Welcome to pyPESTO’s documentation!¶
Install and upgrade¶
Requirements¶
This package requires Python 3.6 or later. It is tested on Linux using Travis continuous integration.
I cannot use my system’s Python distribution, what now?¶
Several Python distributions can co-exist on a single system. If you don’t have access to a recent Python version via your system’s package manager (this might be the case for old operating systems), it is recommended to install the latest version of the Anaconda Python 3 distribution.
Also, there is the possibility to use multiple virtual environments via:
python3 -m virtualenv ENV_NAME
source ENV_NAME/bin/activate
where ENV_NAME denotes an individual environment name, if you do not want to mess up the system environment.
Install from PIP¶
The package can be installed from the Python Package Index PyPI via pip:
pip3 install pypesto
Install from GIT¶
If you want the bleeding edge version, install directly from github:
pip3 install git+https://github.com/icb-dcm/pypesto.git
If you need to have access to the source code, you can download it via:
git clone https://github.com/icb-dcm/pypesto.git
and then install from the local repository via:
cd pypesto
pip3 install .
Upgrade¶
If you want to upgrade from an existing previous version, replace
install by ìnstall --upgrade in the above commands.
Install optional packages¶
This package includes multiple comfort methods simplyfing its use for parameter estimation for models generated using the toolbox amici. To use AMICI, install it via pip:
pip3 install amici
This package inherently supports optimization using the dlib toolbox. To use it, install dlib via:
pip3 install dlib
Examples¶
The following examples cover typical use cases and should help get a better idea of how to use this package:
Rosenbrock banana¶
Here, we perform optimization for the Rosenbrock banana function, which does not require an AMICI model. In particular, we try several ways of specifying derivative information.
[1]:
import pypesto
import numpy as np
import scipy as sp
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
%matplotlib inline
Define the objective and problem¶
[2]:
# first type of objective
objective1 = pypesto.Objective(fun=sp.optimize.rosen,
grad=sp.optimize.rosen_der,
hess=sp.optimize.rosen_hess)
# second type of objective
def rosen2(x):
return sp.optimize.rosen(x), sp.optimize.rosen_der(x), sp.optimize.rosen_hess(x)
objective2 = pypesto.Objective(fun=rosen2, grad=True, hess=True)
dim_full = 10
lb = -5 * np.ones((dim_full, 1))
ub = 5 * np.ones((dim_full, 1))
problem1 = pypesto.Problem(objective=objective1, lb=lb, ub=ub)
problem2 = pypesto.Problem(objective=objective2, lb=lb, ub=ub)
Illustration¶
[3]:
x = np.arange(-2, 2, 0.1)
y = np.arange(-2, 2, 0.1)
x, y = np.meshgrid(x, y)
z = np.zeros_like(x)
for j in range(0, x.shape[0]):
for k in range(0, x.shape[1]):
z[j,k] = objective1([x[j,k], y[j,k]], (0,))
[4]:
fig = plt.figure()
fig.set_size_inches(*(14,10))
ax = plt.axes(projection='3d')
ax.plot_surface(X=x, Y=y, Z=z)
plt.xlabel('x axis')
plt.ylabel('y axis')
ax.set_title('cost function values')
[4]:
Text(0.5, 0.92, 'cost function values')
Run optimization¶
[5]:
# create different optimizers
optimizer_bfgs = pypesto.ScipyOptimizer(method='l-bfgs-b')
optimizer_tnc = pypesto.ScipyOptimizer(method='TNC')
optimizer_dogleg = pypesto.ScipyOptimizer(method='dogleg')
# set number of starts
n_starts = 20
# save optimizer trace
history_options = pypesto.HistoryOptions(trace_record=True)
# Run optimizaitons for different optimzers
result1_bfgs = pypesto.minimize(
problem=problem1, optimizer=optimizer_bfgs,
n_starts=n_starts, history_options=history_options)
result1_tnc = pypesto.minimize(
problem=problem1, optimizer=optimizer_tnc,
n_starts=n_starts, history_options=history_options)
result1_dogleg = pypesto.minimize(
problem=problem1, optimizer=optimizer_dogleg,
n_starts=n_starts, history_options=history_options)
# Optimize second type of objective
result2 = pypesto.minimize(problem=problem2, optimizer=optimizer_tnc, n_starts=n_starts)
Visualize and compare optimization results¶
[6]:
import pypesto.visualize
# plot separated waterfalls
pypesto.visualize.waterfall(result1_bfgs, size=(15,6))
pypesto.visualize.waterfall(result1_tnc, size=(15,6))
pypesto.visualize.waterfall(result1_dogleg, size=(15,6))
[6]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd9397beb90>
We can now have a closer look, which method perfomred better: Let’s first compare bfgs and TNC, since both methods gave good results. How does the fine convergence look like?
[7]:
# plot one list of waterfalls
pypesto.visualize.waterfall([result1_bfgs, result1_tnc],
legends=['L-BFGS-B', 'TNC'],
start_indices=10,
scale_y='lin')
[7]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd93936a410>
[8]:
# retrieve second optimum
all_x = result1_bfgs.optimize_result.get_for_key('x')
all_fval = result1_bfgs.optimize_result.get_for_key('fval')
x = all_x[19]
fval = all_fval[19]
print('Second optimum at: ' + str(fval))
# create a reference point from it
ref = {'x': x, 'fval': fval, 'color': [
0.2, 0.4, 1., 1.], 'legend': 'second optimum'}
ref = pypesto.visualize.create_references(ref)
# new waterfall plot with reference point for second optimum
pypesto.visualize.waterfall(result1_dogleg, size=(15,6),
scale_y='lin', y_limits=[-1, 101],
reference=ref, colors=[0., 0., 0., 1.])
Second optimum at: 3.9865791124344874
[8]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd9393120d0>
Visualize parameters¶
There seems to be a second local optimum. We want to see whether it was also found by the dogleg method
[9]:
pypesto.visualize.parameters([result1_bfgs, result1_tnc],
legends=['L-BFGS-B', 'TNC'],
balance_alpha=False)
pypesto.visualize.parameters(result1_dogleg,
legends='dogleg',
reference=ref,
size=(15,10),
start_indices=[0, 1, 2, 3, 4, 5],
balance_alpha=False)
[9]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd9394ccf10>
Optimizer history¶
Let’s compare optimzer progress over time.
[10]:
# plot one list of waterfalls
pypesto.visualize.optimizer_history([result1_bfgs, result1_tnc],
legends=['L-BFGS-B', 'TNC'],
reference=ref)
# plot one list of waterfalls
pypesto.visualize.optimizer_history(result1_dogleg,
reference=ref)
[10]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd93983b750>
We can also visualize this usign other scalings or offsets…
[11]:
# plot one list of waterfalls
pypesto.visualize.optimizer_history([result1_bfgs, result1_tnc],
legends=['L-BFGS-B', 'TNC'],
reference=ref,
offset_y=0.)
# plot one list of waterfalls
pypesto.visualize.optimizer_history([result1_bfgs, result1_tnc],
legends=['L-BFGS-B', 'TNC'],
reference=ref,
scale_y='lin',
y_limits=[-1., 11.])
[11]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fd9390d9690>
Compute profiles¶
The profiling routine needs a problem, a results object and an optimizer.
Moreover it accepts an index of integer (profile_index), whether or not a profile should be computed.
Finally, an integer (result_index) can be passed, in order to specify the local optimum, from which profiling should be started.
[12]:
# compute profiles
profile_options = pypesto.ProfileOptions(min_step_size=0.0005,
delta_ratio_max=0.05,
default_step_size=0.005,
ratio_min=0.03)
result1_tnc = pypesto.parameter_profile(
problem=problem1,
result=result1_tnc,
optimizer=optimizer_tnc,
profile_index=np.array([1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0]),
result_index=0,
profile_options=profile_options)
# compute profiles from second optimum
result1_tnc = pypesto.parameter_profile(
problem=problem1,
result=result1_tnc,
optimizer=optimizer_tnc,
profile_index=np.array([1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0]),
result_index=19,
profile_options=profile_options)
Visualize and analyze results¶
pypesto offers easy-to-use visualization routines:
[13]:
# specify the parameters, for which profiles should be computed
ax = pypesto.visualize.profiles(result1_tnc, profile_indices = [0,1,2,5],
reference=ref, profile_list_id=0)
# plot profiles again, now from second optimum
ax = pypesto.visualize.profiles(result1_tnc, profile_indices = [0,1,2,5],
reference=ref, profile_list_id=1)
If the result needs to be examined in more detail, it can easily be exported as a pandas.DataFrame:
[14]:
result1_tnc.optimize_result.as_dataframe(['fval', 'n_fval', 'n_grad',
'n_hess', 'n_res', 'n_sres', 'time'])
[14]:
| fval | n_fval | n_grad | n_hess | n_res | n_sres | time | |
|---|---|---|---|---|---|---|---|
| 0 | 1.968227e-13 | 193 | 193 | 0 | 0 | 0 | 0.018201 |
| 1 | 2.202262e-13 | 165 | 165 | 0 | 0 | 0 | 0.039069 |
| 2 | 1.550811e-12 | 188 | 188 | 0 | 0 | 0 | 0.017980 |
| 3 | 1.553846e-12 | 188 | 188 | 0 | 0 | 0 | 0.017905 |
| 4 | 3.138476e-12 | 162 | 162 | 0 | 0 | 0 | 0.015469 |
| 5 | 8.042668e-12 | 229 | 229 | 0 | 0 | 0 | 0.021637 |
| 6 | 8.268731e-12 | 209 | 209 | 0 | 0 | 0 | 0.049976 |
| 7 | 8.310174e-12 | 171 | 171 | 0 | 0 | 0 | 0.016296 |
| 8 | 1.364149e-11 | 219 | 219 | 0 | 0 | 0 | 0.021103 |
| 9 | 1.382298e-11 | 134 | 134 | 0 | 0 | 0 | 0.013395 |
| 10 | 3.487863e-11 | 186 | 186 | 0 | 0 | 0 | 0.017843 |
| 11 | 1.211274e-10 | 160 | 160 | 0 | 0 | 0 | 0.042303 |
| 12 | 1.568336e-10 | 167 | 167 | 0 | 0 | 0 | 0.016380 |
| 13 | 1.557791e-09 | 170 | 170 | 0 | 0 | 0 | 0.016406 |
| 14 | 2.989273e-09 | 174 | 174 | 0 | 0 | 0 | 0.017172 |
| 15 | 3.045916e-08 | 199 | 199 | 0 | 0 | 0 | 0.026230 |
| 16 | 3.194012e-08 | 176 | 176 | 0 | 0 | 0 | 0.033760 |
| 17 | 3.986579e+00 | 134 | 134 | 0 | 0 | 0 | 0.013941 |
| 18 | 3.986579e+00 | 152 | 152 | 0 | 0 | 0 | 0.017177 |
| 19 | 3.986579e+00 | 103 | 103 | 0 | 0 | 0 | 0.009830 |
Conversion reaction¶
[1]:
import importlib
import os
import sys
import numpy as np
import amici
import amici.plotting
import pypesto
# sbml file we want to import
sbml_file = 'conversion_reaction/model_conversion_reaction.xml'
# name of the model that will also be the name of the python module
model_name = 'model_conversion_reaction'
# directory to which the generated model code is written
model_output_dir = 'tmp/' + model_name
Compile AMICI model¶
[2]:
# import sbml model, compile and generate amici module
sbml_importer = amici.SbmlImporter(sbml_file)
sbml_importer.sbml2amici(model_name,
model_output_dir,
verbose=False)
Load AMICI model¶
[3]:
# load amici module (the usual starting point later for the analysis)
sys.path.insert(0, os.path.abspath(model_output_dir))
model_module = importlib.import_module(model_name)
model = model_module.getModel()
model.requireSensitivitiesForAllParameters()
model.setTimepoints(amici.DoubleVector(np.linspace(0, 10, 11)))
model.setParameterScale(amici.ParameterScaling_log10)
model.setParameters(amici.DoubleVector([-0.3,-0.7]))
solver = model.getSolver()
solver.setSensitivityMethod(amici.SensitivityMethod_forward)
solver.setSensitivityOrder(amici.SensitivityOrder_first)
# how to run amici now:
rdata = amici.runAmiciSimulation(model, solver, None)
amici.plotting.plotStateTrajectories(rdata)
edata = amici.ExpData(rdata, 0.2, 0.0)
Optimize¶
[4]:
# create objective function from amici model
# pesto.AmiciObjective is derived from pesto.Objective,
# the general pesto objective function class
objective = pypesto.AmiciObjective(model, solver, [edata], 1)
# create optimizer object which contains all information for doing the optimization
optimizer = pypesto.ScipyOptimizer(method='ls_trf')
#optimizer.solver = 'bfgs|meigo'
# if select meigo -> also set default values in solver_options
#optimizer.options = {'maxiter': 1000, 'disp': True} # = pesto.default_options_meigo()
#optimizer.startpoints = []
#optimizer.startpoint_method = 'lhs|uniform|something|function'
#optimizer.n_starts = 100
# see PestoOptions.m for more required options here
# returns OptimizationResult, see parameters.MS for what to return
# list of final optim results foreach multistart, times, hess, grad,
# flags, meta information (which optimizer -> optimizer.get_repr())
# create problem object containing all information on the problem to be solved
problem = pypesto.Problem(objective=objective,
lb=[-2,-2], ub=[2,2])
# maybe lb, ub = inf
# other constraints: kwargs, class pesto.Constraints
# constraints on pams, states, esp. pesto.AmiciConstraints (e.g. pam1 + pam2<= const)
# if optimizer cannot handle -> error
# maybe also scaling / transformation of parameters encoded here
# do the optimization
result = pypesto.minimize(problem=problem,
optimizer=optimizer,
n_starts=10)
# optimize is a function since it does not need an internal memory,
# just takes input and returns output in the form of a Result object
# 'result' parameter: e.g. some results from somewhere -> pick best start points
Visualize¶
[5]:
# waterfall, parameter space, scatter plots, fits to data
# different functions for different plotting types
import pypesto.visualize
pypesto.visualize.waterfall(result)
pypesto.visualize.parameters(result)
[5]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f48849e1150>
Data storage¶
[6]:
# result = pypesto.storage.load('db_file.db')
Profiles¶
[7]:
# there are three main parts: optimize, profile, sample. the overall structure of profiles and sampling
# will be similar to optimizer like above.
# we intend to only have just one result object which can be reused everywhere, but the problem of how to
# not have one huge class but
# maybe simplified views on it for optimization, profiles and sampling is still to be solved
# profiler = pypesto.Profiler()
# result = pypesto.profile(problem, profiler, result=None)
# possibly pass result object from optimization to get good parameter guesses
Sampling¶
[8]:
# sampler = pypesto.Sampler()
# result = pypesto.sample(problem, sampler, result=None)
[9]:
# open: how to parallelize. the idea is to use methods similar to those in pyabc for working on clusters.
# one way would be to specify an additional 'engine' object passed to optimize(), profile(), sample(),
# which in the default setting just does a for loop, but can also be customized.
Fixed parameters¶
In this notebook we will show how to use fixed parameters. Therefore, we employ our Rosenbrock example. We define two problems, where for the first problem all parameters are optimized, and for the second we fix some of them to specified values.
Define problem¶
[1]:
import pypesto
import pypesto.visualize
import numpy as np
import scipy as sp
import matplotlib.pyplot as plt
%matplotlib inline
[2]:
objective = pypesto.Objective(fun=sp.optimize.rosen,
grad=sp.optimize.rosen_der,
hess=sp.optimize.rosen_hess)
dim_full = 5
lb = -2 * np.ones((dim_full,1))
ub = 2 * np.ones((dim_full,1))
problem1 = pypesto.Problem(objective=objective, lb=lb, ub=ub)
x_fixed_indices = [1, 3]
x_fixed_vals = [1, 1]
problem2 = pypesto.Problem(objective=objective, lb=lb, ub=ub,
x_fixed_indices=x_fixed_indices,
x_fixed_vals=x_fixed_vals)
Optimize¶
[3]:
optimizer = pypesto.ScipyOptimizer()
n_starts = 10
result1 = pypesto.minimize(problem=problem1, optimizer=optimizer,
n_starts=n_starts)
result2 = pypesto.minimize(problem=problem2, optimizer=optimizer,
n_starts=n_starts)
Visualize¶
[4]:
fig, ax = plt.subplots()
pypesto.visualize.waterfall(result1, ax)
pypesto.visualize.waterfall(result2, ax)
pypesto.visualize.parameters(result1)
pypesto.visualize.parameters(result2)
pypesto.visualize.parameters(result2, free_indices_only=False)
[4]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fe98fe12650>
[5]:
result1.optimize_result.as_dataframe(['fval', 'x', 'grad'])
[5]:
| fval | x | grad | |
|---|---|---|---|
| 0 | 4.689854e-14 | [0.9999999815635604, 0.9999999601303594, 0.999... | [1.1618317897690609e-06, 1.2817350405303142e-0... |
| 1 | 5.590260e-14 | [0.9999999788673749, 0.9999999710531022, 0.999... | [-5.3696059283362676e-06, -3.17862832097996e-0... |
| 2 | 3.799999e-13 | [0.9999999894531796, 0.9999999867856568, 1.000... | [-3.1728126266944613e-06, -1.6951548081005115e... |
| 3 | 4.110300e-13 | [1.0000000452200102, 1.0000000894442573, 1.000... | [4.887460487692825e-07, 3.7999227735179252e-06... |
| 4 | 6.110801e-13 | [0.9999999640563018, 0.9999999933489752, 0.999... | [-2.6166434580847274e-05, 8.102944813282283e-0... |
| 5 | 1.231774e-12 | [1.0000000581023145, 1.000000113974549, 1.0000... | [1.0082380613547457e-06, 1.8071346123806702e-0... |
| 6 | 6.918491e-12 | [1.0000000005661727, 1.0000001076896814, 1.000... | [-4.262180210516511e-05, 7.700439519900304e-05... |
| 7 | 2.552430e-11 | [0.9999999192005172, 1.0000000369812814, 1.000... | [-7.959368869249685e-05, -8.749066771695052e-0... |
| 8 | 2.855055e-11 | [0.9999995268339968, 0.9999992696810054, 0.999... | [-8.735140622961871e-05, 4.373181014008344e-05... |
| 9 | 3.930839e+00 | [-0.9620508371994185, 0.9357391790840979, 0.88... | [5.2019939965397555e-05, 2.44790688288532e-05,... |
[6]:
result2.optimize_result.as_dataframe(['fval', 'x', 'grad'])
[6]:
| fval | x | grad | |
|---|---|---|---|
| 0 | 3.757636e-21 | [0.9999999999987625, 1.0, 1.0000000000008613, ... | [-9.92438575763029e-10, nan, 8.630336445497983... |
| 1 | 3.403187e-16 | [1.000000000296693, 1.0, 0.9999999993226122, 1... | [2.3794769983646036e-07, nan, -6.7874258776674... |
| 2 | 1.078998e-15 | [0.9999999994275696, 1.0, 0.9999999986532816, ... | [-4.5908920971880386e-07, nan, -1.349411838817... |
| 3 | 1.986582e-15 | [0.9999999981059375, 1.0, 0.9999999990468542, ... | [-1.5190380919959983e-06, nan, -9.550520753635... |
| 4 | 1.777511e-14 | [1.0000000013724608, 1.0, 0.9999999967022452, ... | [1.100713572463121e-06, nan, -3.30435034796972... |
| 5 | 2.344376e-14 | [0.9999999959141576, 1.0, 0.9999999996553852, ... | [-3.2768456252559055e-06, nan, -3.453040106231... |
| 6 | 2.096868e-13 | [0.9999999984695723, 1.0, 1.0000000066163943, ... | [-1.2274029922568326e-06, nan, 6.6296271449766... |
| 7 | 3.989975e+00 | [-0.9949747468749881, 1.0, 0.9999999997811084,... | [-3.507845658390352e-08, nan, -2.1932935463983... |
| 8 | 3.989975e+00 | [-0.9949747444243423, 1.0, 1.000000008470282, ... | [1.900857428349667e-06, nan, 8.487222652784575... |
| 9 | 3.989975e+00 | [-0.9949747299382655, 1.0, 1.0000000336260297,... | [1.334441878420023e-05, nan, 3.369328314657676... |
AMICI Python example “Boehm”¶
This is an example using the model [boehm_ProteomeRes2014.xml] model to demonstrate and test SBML import and AMICI Python interface.
[1]:
import libsbml
import importlib
import amici
import pypesto
import os
import sys
import numpy as np
import matplotlib.pyplot as plt
# temporarily add the simulate file
sys.path.insert(0, 'boehm_JProteomeRes2014')
from benchmark_import import DataProvider
# sbml file
sbml_file = 'boehm_JProteomeRes2014/boehm_JProteomeRes2014.xml'
# name of the model that will also be the name of the python module
model_name = 'boehm_JProteomeRes2014'
# output directory
model_output_dir = 'tmp/' + model_name
The example model¶
Here we use libsbml to show the reactions and species described by the model (this is independent of AMICI).
[2]:
sbml_reader = libsbml.SBMLReader()
sbml_doc = sbml_reader.readSBML(os.path.abspath(sbml_file))
sbml_model = sbml_doc.getModel()
dir(sbml_doc)
print(os.path.abspath(sbml_file))
print('Species: ', [s.getId() for s in sbml_model.getListOfSpecies()])
print('\nReactions:')
for reaction in sbml_model.getListOfReactions():
reactants = ' + '.join(['%s %s'%(int(r.getStoichiometry()) if r.getStoichiometry() > 1 else '', r.getSpecies()) for r in reaction.getListOfReactants()])
products = ' + '.join(['%s %s'%(int(r.getStoichiometry()) if r.getStoichiometry() > 1 else '', r.getSpecies()) for r in reaction.getListOfProducts()])
reversible = '<' if reaction.getReversible() else ''
print('%3s: %10s %1s->%10s\t\t[%s]' % (reaction.getId(),
reactants,
reversible,
products,
libsbml.formulaToL3String(reaction.getKineticLaw().getMath())))
/home/yannik/pypesto/doc/example/boehm_JProteomeRes2014/boehm_JProteomeRes2014.xml
Species: ['STAT5A', 'STAT5B', 'pApB', 'pApA', 'pBpB', 'nucpApA', 'nucpApB', 'nucpBpB']
Reactions:
v1_v_0: 2 STAT5A -> pApA [cyt * BaF3_Epo * STAT5A^2 * k_phos]
v2_v_1: STAT5A + STAT5B -> pApB [cyt * BaF3_Epo * STAT5A * STAT5B * k_phos]
v3_v_2: 2 STAT5B -> pBpB [cyt * BaF3_Epo * STAT5B^2 * k_phos]
v4_v_3: pApA -> nucpApA [cyt * k_imp_homo * pApA]
v5_v_4: pApB -> nucpApB [cyt * k_imp_hetero * pApB]
v6_v_5: pBpB -> nucpBpB [cyt * k_imp_homo * pBpB]
v7_v_6: nucpApA -> 2 STAT5A [nuc * k_exp_homo * nucpApA]
v8_v_7: nucpApB -> STAT5A + STAT5B [nuc * k_exp_hetero * nucpApB]
v9_v_8: nucpBpB -> 2 STAT5B [nuc * k_exp_homo * nucpBpB]
Importing an SBML model, compiling and generating an AMICI module¶
Before we can use AMICI to simulate our model, the SBML model needs to be translated to C++ code. This is done by amici.SbmlImporter.
[3]:
# Create an SbmlImporter instance for our SBML model
sbml_importer = amici.SbmlImporter(sbml_file)
In this example, we want to specify fixed parameters, observables and a \(\sigma\) parameter. Unfortunately, the latter two are not part of the SBML standard. However, they can be provided to amici.SbmlImporter.sbml2amici as demonstrated in the following.
Constant parameters¶
Constant parameters, i.e. parameters with respect to which no sensitivities are to be computed (these are often parameters specifying a certain experimental condition) are provided as a list of parameter names.
[4]:
constantParameters = {'ratio', 'specC17'}
Observables¶
We used SBML’s `AssignmentRule <http://sbml.org/Software/libSBML/5.13.0/docs//python-api/classlibsbml_1_1_rule.html>`__ as a non-standard way to specify Model outputs within the SBML file. These rules need to be removed prior to the model import (AMICI does at this time not support these Rules). This can be easily done using amici.assignmentRules2observables().
In this example, we introduced parameters named observable_* as targets of the observable AssignmentRules. Where applicable we have observable_*_sigma parameters for \(\sigma\) parameters (see below).
[5]:
# Retrieve model output names and formulae from AssignmentRules and remove the respective rules
observables = amici.assignmentRules2observables(
sbml_importer.sbml, # the libsbml model object
filter_function=lambda variable: variable.getId().startswith('observable_') and not variable.getId().endswith('_sigma')
)
print('Observables:', observables)
Observables: {'observable_pSTAT5A_rel': {'name': '', 'formula': '(100 * pApB + 200 * pApA * specC17) / (pApB + STAT5A * specC17 + 2 * pApA * specC17)'}, 'observable_pSTAT5B_rel': {'name': '', 'formula': '-(100 * pApB - 200 * pBpB * (specC17 - 1)) / (STAT5B * (specC17 - 1) - pApB + 2 * pBpB * (specC17 - 1))'}, 'observable_rSTAT5A_rel': {'name': '', 'formula': '(100 * pApB + 100 * STAT5A * specC17 + 200 * pApA * specC17) / (2 * pApB + STAT5A * specC17 + 2 * pApA * specC17 - STAT5B * (specC17 - 1) - 2 * pBpB * (specC17 - 1))'}}
\(\sigma\) parameters¶
To specify measurement noise as a parameter, we simply provide a dictionary with (preexisting) parameter names as keys and a list of observable names as values to indicate which sigma parameter is to be used for which observable.
[6]:
sigma_vals = ['sd_pSTAT5A_rel', 'sd_pSTAT5B_rel', 'sd_rSTAT5A_rel']
observable_names = observables.keys()
sigmas = dict(zip(list(observable_names), sigma_vals))
print(sigmas)
{'observable_pSTAT5A_rel': 'sd_pSTAT5A_rel', 'observable_pSTAT5B_rel': 'sd_pSTAT5B_rel', 'observable_rSTAT5A_rel': 'sd_rSTAT5A_rel'}
Generating the module¶
Now we can generate the python module for our model. amici.SbmlImporter.sbml2amici will symbolically derive the sensitivity equations, generate C++ code for model simulation, and assemble the python module.
[7]:
sbml_importer.sbml2amici(model_name,
model_output_dir,
verbose=False,
observables=observables,
constantParameters=constantParameters,
sigmas=sigmas
)
Importing the module and loading the model¶
If everything went well, we need to add the previously selected model output directory to our PYTHON_PATH and are then ready to load newly generated model:
[8]:
sys.path.insert(0, os.path.abspath(model_output_dir))
model_module = importlib.import_module(model_name)
And get an instance of our model from which we can retrieve information such as parameter names:
[9]:
model = model_module.getModel()
print("Model parameters:", list(model.getParameterIds()))
print("Model outputs: ", list(model.getObservableIds()))
print("Model states: ", list(model.getStateIds()))
Model parameters: ['Epo_degradation_BaF3', 'k_exp_hetero', 'k_exp_homo', 'k_imp_hetero', 'k_imp_homo', 'k_phos', 'sd_pSTAT5A_rel', 'sd_pSTAT5B_rel', 'sd_rSTAT5A_rel']
Model outputs: ['observable_pSTAT5A_rel', 'observable_pSTAT5B_rel', 'observable_rSTAT5A_rel']
Model states: ['STAT5A', 'STAT5B', 'pApB', 'pApA', 'pBpB', 'nucpApA', 'nucpApB', 'nucpBpB']
Running simulations and analyzing results¶
After importing the model, we can run simulations using amici.runAmiciSimulation. This requires a Model instance and a Solver instance. Optionally you can provide measurements inside an ExpData instance, as shown later in this notebook.
[10]:
h5_file = 'boehm_JProteomeRes2014/data_boehm_JProteomeRes2014.h5'
dp = DataProvider(h5_file)
[11]:
# set timepoints for which we want to simulate the model
timepoints = amici.DoubleVector(dp.get_timepoints())
model.setTimepoints(timepoints)
# set fixed parameters for which we want to simulate the model
model.setFixedParameters(amici.DoubleVector(np.array([0.693, 0.107])))
# set parameters to optimal values found in the benchmark collection
model.setParameterScale(2)
model.setParameters(amici.DoubleVector(np.array([-1.568917588,
-4.999704894,
-2.209698782,
-1.786006548,
4.990114009,
4.197735488,
0.585755271,
0.818982819,
0.498684404
])))
# Create solver instance
solver = model.getSolver()
# Run simulation using model parameters from the benchmark collection and default solver options
rdata = amici.runAmiciSimulation(model, solver)
[12]:
# Create edata
edata = amici.ExpData(rdata, 1.0, 0)
# set observed data
edata.setObservedData(amici.DoubleVector(dp.get_measurements()[0][:, 0]), 0)
edata.setObservedData(amici.DoubleVector(dp.get_measurements()[0][:, 1]), 1)
edata.setObservedData(amici.DoubleVector(dp.get_measurements()[0][:, 2]), 2)
# set standard deviations to optimal values found in the benchmark collection
edata.setObservedDataStdDev(amici.DoubleVector(np.array(16*[10**0.585755271])), 0)
edata.setObservedDataStdDev(amici.DoubleVector(np.array(16*[10**0.818982819])), 1)
edata.setObservedDataStdDev(amici.DoubleVector(np.array(16*[10**0.498684404])), 2)
[13]:
rdata = amici.runAmiciSimulation(model, solver, edata)
print('Chi2 value reported in benchmark collection: 47.9765479')
print('chi2 value using AMICI:')
print(rdata['chi2'])
Chi2 value reported in benchmark collection: 47.9765479
chi2 value using AMICI:
47.97654266893465
Run optimization using pyPESTO¶
[14]:
# create objective function from amici model
# pesto.AmiciObjective is derived from pesto.Objective,
# the general pesto objective function class
model.requireSensitivitiesForAllParameters()
solver.setSensitivityMethod(amici.SensitivityMethod_forward)
solver.setSensitivityOrder(amici.SensitivityOrder_first)
objective = pypesto.AmiciObjective(model, solver, [edata], 1)
[15]:
# create optimizer object which contains all information for doing the optimization
optimizer = pypesto.ScipyOptimizer()
optimizer.solver = 'bfgs'
[16]:
# create problem object containing all information on the problem to be solved
x_names = ['x' + str(j) for j in range(0, 9)]
problem = pypesto.Problem(objective=objective,
lb=-5*np.ones((9)), ub=5*np.ones((9)),
x_names=x_names)
[17]:
# do the optimization
result = pypesto.minimize(problem=problem,
optimizer=optimizer,
n_starts=10) # 200
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 221.821 and h = 3.00478e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 221.821149:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 221.821 and h = 3.00478e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 221.821149:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 221.821 and h = 3.00478e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 221.821149:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 147.199 and h = 2.90261e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 147.198629:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 147.199 and h = 2.90261e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 147.198629:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 147.199 and h = 2.90261e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 147.198629:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 198 and h = 2.97875e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 197.999609:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 197.697 and h = 2.98464e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 197.696730:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 197.697 and h = 2.98464e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 197.696730:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 197.697 and h = 2.98464e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 197.696730:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 66.4603 and h = 6.88533e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 66.460272:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 66.3735 and h = 8.78908e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 66.373478:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 85.8974 and h = 2.05376e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 85.897359:
AMICI failed to integrate the forward problem
Visualization¶
Create waterfall and parameter plot
[18]:
# waterfall, parameter space,
import pypesto.visualize
pypesto.visualize.waterfall(result)
pypesto.visualize.parameters(result)
[18]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f3bc5881c10>
Model import using the Petab format¶
In this notebook, we illustrate how to use pyPESTO together with PEtab and AMICI. We employ models from the benchmark collection, which we first download:
[1]:
import pypesto
import amici
import petab
import os
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
!git clone --depth 1 https://github.com/Benchmarking-Initiative/Benchmark-Models-PEtab.git tmp/benchmark-models || (cd tmp/benchmark-models && git pull)
folder_base = "tmp/benchmark-models/Benchmark-Models/"
fatal: destination path 'tmp/benchmark-models' already exists and is not an empty directory.
Already up to date.
Import¶
Manage PEtab model¶
A PEtab problem comprises all the information on the model, the data and the parameters to perform parameter estimation. We import a model as a petab.Problem.
[2]:
# a collection of models that can be simulated
#model_name = "Zheng_PNAS2012"
model_name = "Boehm_JProteomeRes2014"
#model_name = "Fujita_SciSignal2010"
#model_name = "Sneyd_PNAS2002"
#model_name = "Borghans_BiophysChem1997"
#model_name = "Elowitz_Nature2000"
#model_name = "Crauste_CellSystems2017"
#model_name = "Lucarelli_CellSystems2018"
#model_name = "Schwen_PONE2014"
#model_name = "Blasi_CellSystems2016"
# the yaml configuration file links to all needed files
yaml_config = os.path.join(folder_base, model_name, model_name + '.yaml')
# create a petab problem
petab_problem = petab.Problem.from_yaml(yaml_config)
Import model to AMICI¶
The model must be imported to pyPESTO and AMICI. Therefore, we create a pypesto.PetabImporter from the problem, and create an AMICI model.
[3]:
importer = pypesto.PetabImporter(petab_problem)
model = importer.create_model()
# some model properties
print("Model parameters:", list(model.getParameterIds()), '\n')
print("Model const parameters:", list(model.getFixedParameterIds()), '\n')
print("Model outputs: ", list(model.getObservableIds()), '\n')
print("Model states: ", list(model.getStateIds()), '\n')
Model parameters: ['Epo_degradation_BaF3', 'k_exp_hetero', 'k_exp_homo', 'k_imp_hetero', 'k_imp_homo', 'k_phos', 'ratio', 'specC17', 'noiseParameter1_pSTAT5A_rel', 'noiseParameter1_pSTAT5B_rel', 'noiseParameter1_rSTAT5A_rel']
Model const parameters: []
Model outputs: ['pSTAT5A_rel', 'pSTAT5B_rel', 'rSTAT5A_rel']
Model states: ['STAT5A', 'STAT5B', 'pApB', 'pApA', 'pBpB', 'nucpApA', 'nucpApB', 'nucpBpB']
Create objective function¶
To perform parameter estimation, we need to define an objective function, which integrates the model, data, and noise model defined in the PEtab problem.
[4]:
import libsbml
converter_config = libsbml.SBMLLocalParameterConverter()\
.getDefaultProperties()
petab_problem.sbml_document.convert(converter_config)
obj = importer.create_objective()
# for some models, hyperparamters need to be adjusted
#obj.amici_solver.setMaxSteps(10000)
#obj.amici_solver.setRelativeTolerance(1e-7)
#obj.amici_solver.setAbsoluteTolerance(1e-7)
We can request variable derivatives via sensi_orders, or function values or residuals as specified via mode. Passing return_dict, we obtain the direct result of the AMICI simulation.
[5]:
ret = obj(petab_problem.x_nominal_scaled, mode='mode_fun', sensi_orders=(0,1), return_dict=True)
print(ret)
{'fval': 138.22199677513575, 'grad': array([ 2.20386015e-02, 5.53227506e-02, 5.78886452e-03, 5.40656415e-03,
-4.51595809e-05, 7.91163446e-03, 0.00000000e+00, 1.07840959e-02,
2.40378735e-02, 1.91919657e-02, 0.00000000e+00]), 'hess': array([[ 2.11105595e+03, 5.89390039e-01, 1.07159910e+02,
2.81393973e+03, 8.94333861e-06, -7.86055092e+02,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 5.89390039e-01, 1.91513744e-03, -1.72774945e-01,
7.12558479e-01, -3.69774927e-08, -3.20531692e-01,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 1.07159910e+02, -1.72774945e-01, 6.99839693e+01,
1.61497679e+02, 7.16323554e-06, -8.83572656e+01,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 2.81393973e+03, 7.12558479e-01, 1.61497679e+02,
3.76058352e+03, 8.40044683e-06, -1.04136909e+03,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 8.94333861e-06, -3.69774927e-08, 7.16323554e-06,
8.40044683e-06, 2.86438192e-10, -2.24927732e-04,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[-7.86055092e+02, -3.20531692e-01, -8.83572656e+01,
-1.04136909e+03, -2.24927732e-04, 9.29902113e+02,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00],
[ 0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00, 0.00000000e+00,
0.00000000e+00, 0.00000000e+00]]), 'res': array([], dtype=float64), 'sres': array([], shape=(0, 11), dtype=float64), 'rdatas': [<amici.numpy.ReturnDataView object at 0x7f7802f86610>]}
The problem defined in PEtab also defines the fixing of parameters, and parameter bounds. This information is contained in a pypesto.Problem.
[6]:
problem = importer.create_problem(obj)
In particular, the problem accounts for the fixing of parametes.
[7]:
print(problem.x_fixed_indices, problem.x_free_indices)
[6, 10] [0, 1, 2, 3, 4, 5, 7, 8, 9]
The problem creates a copy of he objective function that takes into account the fixed parameters. The objective function is able to calculate function values and derivatives. A finite difference check whether the computed gradient is accurate:
[8]:
objective = problem.objective
ret = objective(petab_problem.x_nominal_free_scaled, sensi_orders=(0,1))
print(ret)
(138.22199677513575, array([ 2.20386015e-02, 5.53227506e-02, 5.78886452e-03, 5.40656415e-03,
-4.51595809e-05, 7.91163446e-03, 1.07840959e-02, 2.40378735e-02,
1.91919657e-02]))
[9]:
eps = 1e-4
def fd(x):
grad = np.zeros_like(x)
j = 0
for i, xi in enumerate(x):
mask = np.zeros_like(x)
mask[i] += eps
valinc, _ = objective(x+mask, sensi_orders=(0,1))
valdec, _ = objective(x-mask, sensi_orders=(0,1))
grad[j] = (valinc - valdec) / (2*eps)
j += 1
return grad
fdval = fd(petab_problem.x_nominal_free_scaled)
print("fd: ", fdval)
print("l2 difference: ", np.linalg.norm(ret[1] - fdval))
fd: [0.02493368 0.05309659 0.00530587 0.01291083 0.00587754 0.01473653
0.01078279 0.02403657 0.01919066]
l2 difference: 0.012310244824532144
In short¶
All of the previous steps can be shortened by directly creating an importer object and then a problem:
[10]:
importer = pypesto.PetabImporter.from_yaml(yaml_config)
problem = importer.create_problem()
Run optimization¶
Given the problem, we can perform optimization. We can specify an optimizer to use, and a parallelization engine to speed things up.
[11]:
optimizer = pypesto.ScipyOptimizer()
# engine = pypesto.SingleCoreEngine()
engine = pypesto.MultiProcessEngine()
# do the optimization
result = pypesto.minimize(problem=problem, optimizer=optimizer,
n_starts=10, engine=engine)
Engine set up to use up to 4 processes in total. The number was automatically determined and might not be appropriate on some systems.
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 38.1195 and h = 5.55541e-06, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 38.119511:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 88.9211 and h = 2.14177e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 88.921131:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 88.9211 and h = 2.14177e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 88.921131:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 88.9211 and h = 2.14177e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 88.921131:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 145.551 and h = 1.32433e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 145.550813:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 145.551 and h = 1.32433e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 145.550813:
AMICI failed to integrate the forward problem
[Warning] AMICI:CVODES:CVode:ERR_FAILURE: AMICI ERROR: in module CVODES in function CVode : At t = 145.551 and h = 1.32433e-05, the error test failed repeatedly or with |h| = hmin.
[Warning] AMICI:simulation: AMICI forward simulation failed at t = 145.550813:
AMICI failed to integrate the forward problem
Visualize¶
The results are contained in a pypesto.Result object. It contains e.g. the optimal function values.
[12]:
result.optimize_result.get_for_key('fval')
[12]:
[138.2219740350346,
138.22404611978106,
145.7594099868979,
147.54397516143254,
149.58782926326572,
151.16644923400784,
154.73312826411254,
205.61953652493594,
249.27713115708494,
249.7459974433355]
We can use the standard pyPESTO plotting routines to visualize and analyze the results.
[13]:
import pypesto.visualize
ref = pypesto.visualize.create_references(x=petab_problem.x_nominal_scaled, fval=obj(petab_problem.x_nominal_scaled))
pypesto.visualize.waterfall(result, reference=ref, scale_y='lin')
pypesto.visualize.parameters(result, reference=ref)
[13]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f7802e43510>
Save and load results as HDF5 files¶
[1]:
import pypesto
import numpy as np
import scipy as sp
import matplotlib.pyplot as plt
from pypesto.storage import (save_to_hdf5, read_from_hdf5)
import tempfile
%matplotlib inline
Define the objective and problem¶
[2]:
objective = pypesto.Objective(fun=sp.optimize.rosen,
grad=sp.optimize.rosen_der,
hess=sp.optimize.rosen_hess)
dim_full = 10
lb = -5 * np.ones((dim_full, 1))
ub = 5 * np.ones((dim_full, 1))
problem = pypesto.Problem(objective=objective, lb=lb, ub=ub)
Run optimization¶
[3]:
# create optimizers
optimizer = pypesto.ScipyOptimizer(method='l-bfgs-b')
# set number of starts
n_starts = 20
# Run optimizaitons
result = pypesto.minimize(
problem=problem, optimizer=optimizer,
n_starts=n_starts)
[4]:
result.optimize_result.list
[4]:
[{'id': '8',
'x': array([1.00000001, 0.99999999, 0.99999997, 0.99999998, 0.99999998,
0.99999999, 0.99999996, 0.99999987, 0.99999971, 0.9999995 ]),
'fval': 1.4448869867069234e-12,
'grad': array([ 9.91613312e-06, -2.33793663e-07, -1.84487477e-05, -1.24826804e-06,
-7.03416051e-06, 1.12040576e-05, 1.88713028e-05, -4.68014961e-07,
-3.65179645e-05, 1.53152743e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 73,
'n_grad': 73,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 0.99999999, 0.99999997, 0.99999998, 0.99999998,
0.99999999, 0.99999996, 0.99999987, 0.99999971, 0.9999995 ]),
'fval0': 116315.16334351365,
'history': <pypesto.objective.history.History at 0x7fefee65a750>,
'exitflag': 0,
'time': 0.010613441467285156,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '16',
'x': array([0.99999994, 0.99999998, 1. , 1.00000002, 1.00000004,
1.00000001, 1. , 1.00000012, 1.0000003 , 1.00000051]),
'fval': 4.371307796809753e-12,
'grad': array([-4.26293148e-05, 9.13631144e-06, -1.31339486e-06, 2.51280250e-06,
2.59501842e-05, 4.21294205e-07, -5.58158396e-05, 7.08567852e-07,
4.41611237e-05, -1.57413407e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 79,
'n_grad': 79,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999994, 0.99999998, 1. , 1.00000002, 1.00000004,
1.00000001, 1. , 1.00000012, 1.0000003 , 1.00000051]),
'fval0': 127542.57197202934,
'history': <pypesto.objective.history.History at 0x7fefee65a910>,
'exitflag': 0,
'time': 0.011748075485229492,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '4',
'x': array([1.00000007, 0.99999999, 0.99999998, 1. , 1.00000008,
1.00000021, 1.00000015, 1.00000006, 1.00000004, 1.00000013]),
'fval': 1.7134261938953258e-11,
'grad': array([ 5.79464879e-05, -3.23661397e-05, -1.13616716e-05, -2.69343079e-05,
-1.67474293e-06, 1.20454131e-04, 3.83436764e-05, -1.71072644e-05,
-3.54339727e-05, 1.03840629e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 88,
'n_grad': 88,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000007, 0.99999999, 0.99999998, 1. , 1.00000008,
1.00000021, 1.00000015, 1.00000006, 1.00000004, 1.00000013]),
'fval0': 169143.33089007522,
'history': <pypesto.objective.history.History at 0x7fefee9008d0>,
'exitflag': 0,
'time': 0.016152620315551758,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '11',
'x': array([0.99999983, 0.99999997, 0.99999989, 0.9999998 , 1.00000006,
0.99999986, 0.99999963, 0.99999927, 0.99999861, 0.99999709]),
'fval': 4.473573948366185e-11,
'grad': array([-1.25925183e-04, 8.34342658e-05, -1.58946249e-05, -1.85224905e-04,
2.00742516e-04, -1.80384056e-05, -1.83734314e-05, -2.93938826e-05,
5.98755497e-05, -2.43744695e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 78,
'n_grad': 78,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999983, 0.99999997, 0.99999989, 0.9999998 , 1.00000006,
0.99999986, 0.99999963, 0.99999927, 0.99999861, 0.99999709]),
'fval0': 111440.55513257613,
'history': <pypesto.objective.history.History at 0x7fefee65a7d0>,
'exitflag': 0,
'time': 0.009877920150756836,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '0',
'x': array([1.00000016, 1.00000029, 1.00000013, 1.00000007, 0.99999984,
0.99999968, 0.99999955, 0.99999932, 0.99999871, 0.99999745]),
'fval': 4.512690733773355e-11,
'grad': array([ 6.88017381e-06, 1.82437618e-04, -1.71219792e-05, 7.83029016e-05,
-5.64629619e-05, -7.75613657e-05, -5.33124129e-05, 2.00358870e-05,
-3.38990540e-06, 6.36234430e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 80,
'n_grad': 80,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000016, 1.00000029, 1.00000013, 1.00000007, 0.99999984,
0.99999968, 0.99999955, 0.99999932, 0.99999871, 0.99999745]),
'fval0': 179787.03971937217,
'history': <pypesto.objective.history.History at 0x7fefee6d1e50>,
'exitflag': 0,
'time': 0.05712604522705078,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '18',
'x': array([0.9999999 , 0.99999988, 1.00000002, 1.00000017, 1.00000034,
1.00000045, 1.00000052, 1.00000076, 1.00000146, 1.00000315]),
'fval': 4.530338040953872e-11,
'grad': array([-3.46616032e-05, -8.55052094e-05, 4.32179353e-06, 2.27795791e-05,
8.77325561e-05, 1.11150847e-04, 4.14626291e-05, -3.72317820e-05,
-9.65698380e-05, 4.47893709e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 85,
'n_grad': 85,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.9999999 , 0.99999988, 1.00000002, 1.00000017, 1.00000034,
1.00000045, 1.00000052, 1.00000076, 1.00000146, 1.00000315]),
'fval0': 84246.35907849146,
'history': <pypesto.objective.history.History at 0x7fefee65a990>,
'exitflag': 0,
'time': 0.01163339614868164,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '17',
'x': array([1.00000014, 1.00000006, 1.00000012, 1.00000002, 0.99999972,
0.99999976, 0.99999924, 0.9999986 , 0.99999728, 0.99999457]),
'fval': 5.187501773111393e-11,
'grad': array([ 8.98076519e-05, -4.45109249e-05, 8.61160519e-05, 8.83761172e-05,
-1.98032428e-04, 1.80982671e-04, -1.05227326e-04, -1.64856814e-05,
1.06897803e-05, 1.09849767e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 99,
'n_grad': 99,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000014, 1.00000006, 1.00000012, 1.00000002, 0.99999972,
0.99999976, 0.99999924, 0.9999986 , 0.99999728, 0.99999457]),
'fval0': 164257.74387447865,
'history': <pypesto.objective.history.History at 0x7fefee65a950>,
'exitflag': 0,
'time': 0.013841629028320312,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '13',
'x': array([1.00000009, 1.00000004, 1.00000001, 1.0000001 , 1.00000019,
1.00000029, 1.00000006, 1.00000059, 1.00000094, 1.00000194]),
'fval': 6.113882079371062e-11,
'grad': array([ 5.38201966e-05, 1.12643005e-06, -4.79180197e-05, 1.70876435e-05,
3.76896795e-05, 1.90462489e-04, -2.94335966e-04, 1.92128181e-04,
-7.02675278e-05, 1.21112443e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 84,
'n_grad': 84,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000009, 1.00000004, 1.00000001, 1.0000001 , 1.00000019,
1.00000029, 1.00000006, 1.00000059, 1.00000094, 1.00000194]),
'fval0': 128057.90608516608,
'history': <pypesto.objective.history.History at 0x7fefee65a850>,
'exitflag': 0,
'time': 0.011581659317016602,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '9',
'x': array([1.0000003 , 1.00000024, 0.99999999, 0.99999981, 0.99999975,
0.99999984, 1.00000003, 1.00000007, 1.00000014, 1.00000029]),
'fval': 6.603657918190325e-11,
'grad': array([ 1.45585786e-04, 1.26131400e-04, -3.15965052e-05, -8.20696700e-05,
-1.17487544e-04, -6.42577094e-05, 5.98749705e-05, 4.78675947e-06,
-7.72296480e-06, 3.75853482e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 83,
'n_grad': 83,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.0000003 , 1.00000024, 0.99999999, 0.99999981, 0.99999975,
0.99999984, 1.00000003, 1.00000007, 1.00000014, 1.00000029]),
'fval0': 66001.21516931924,
'history': <pypesto.objective.history.History at 0x7ff00035a3d0>,
'exitflag': 0,
'time': 0.012688159942626953,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '19',
'x': array([1.00000001, 1. , 0.99999996, 0.99999988, 1.00000026,
1.00000032, 1.00000033, 1.00000085, 1.00000187, 1.00000333]),
'fval': 6.709274183733498e-11,
'grad': array([ 8.70631729e-06, 9.70587008e-06, 1.05693559e-05, -2.12546166e-04,
1.84978271e-04, 8.28659608e-05, -1.39703429e-04, -2.30820323e-05,
1.96299302e-04, -8.02670983e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 67,
'n_grad': 67,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 1. , 0.99999996, 0.99999988, 1.00000026,
1.00000032, 1.00000033, 1.00000085, 1.00000187, 1.00000333]),
'fval0': 80442.06208067665,
'history': <pypesto.objective.history.History at 0x7fefee65a9d0>,
'exitflag': 0,
'time': 0.008251428604125977,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '10',
'x': array([0.9999998 , 1.00000012, 0.99999997, 1.00000017, 1.00000028,
1.00000037, 1.0000004 , 1.0000005 , 1.00000069, 1.00000195]),
'fval': 1.065068443513822e-10,
'grad': array([-2.07797214e-04, 2.08384202e-04, -1.40258436e-04, 6.72135025e-05,
6.29120089e-05, 9.92424271e-05, 5.28220148e-05, 6.30692991e-05,
-2.90590939e-04, 1.15498187e-04]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 135,
'n_grad': 135,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.9999998 , 1.00000012, 0.99999997, 1.00000017, 1.00000028,
1.00000037, 1.0000004 , 1.0000005 , 1.00000069, 1.00000195]),
'fval0': 218642.53588542074,
'history': <pypesto.objective.history.History at 0x7fefee65a510>,
'exitflag': 0,
'time': 0.017060041427612305,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '6',
'x': array([1.00000009, 0.9999998 , 1.00000026, 1.00000009, 1.00000031,
1.00000016, 1.00000029, 1.00000052, 1.00000131, 1.00000213]),
'fval': 1.3507352118480165e-10,
'grad': array([ 1.50715573e-04, -3.42587714e-04, 3.08461493e-04, -1.41826830e-04,
2.10332130e-04, -7.64162802e-05, 1.68271418e-05, -1.20534088e-04,
2.55005501e-04, -9.87312659e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 86,
'n_grad': 86,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000009, 0.9999998 , 1.00000026, 1.00000009, 1.00000031,
1.00000016, 1.00000029, 1.00000052, 1.00000131, 1.00000213]),
'fval0': 49550.65276671963,
'history': <pypesto.objective.history.History at 0x7fefef785510>,
'exitflag': 0,
'time': 0.012197017669677734,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '14',
'x': array([1.00000021, 0.99999999, 0.99999968, 0.99999939, 0.99999938,
0.99999947, 0.9999996 , 0.99999955, 0.99999928, 0.99999839]),
'fval': 1.76508815611245e-10,
'grad': array([ 1.69974542e-04, 3.68276755e-05, -7.46786006e-05, -2.36880611e-04,
-1.60596472e-04, -1.26090555e-04, -5.76633224e-06, -1.01269226e-06,
9.59426070e-05, -3.14643190e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 94,
'n_grad': 94,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000021, 0.99999999, 0.99999968, 0.99999939, 0.99999938,
0.99999947, 0.9999996 , 0.99999955, 0.99999928, 0.99999839]),
'fval0': 113189.5063880412,
'history': <pypesto.objective.history.History at 0x7fefee65a890>,
'exitflag': 0,
'time': 0.010685205459594727,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '7',
'x': array([1.00000001, 1.00000008, 1.00000032, 0.9999997 , 0.99999988,
0.9999994 , 0.99999945, 0.99999874, 0.9999976 , 0.99999541]),
'fval': 1.8629501787028135e-10,
'grad': array([-1.89690853e-05, -5.75176850e-05, 4.08684636e-04, -3.79119426e-04,
2.41840701e-04, -3.35502015e-04, 1.94400780e-04, -8.41625352e-05,
-6.36788959e-05, 4.17276302e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 86,
'n_grad': 86,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 1.00000008, 1.00000032, 0.9999997 , 0.99999988,
0.9999994 , 0.99999945, 0.99999874, 0.9999976 , 0.99999541]),
'fval0': 275340.0482345366,
'history': <pypesto.objective.history.History at 0x7fefee65a550>,
'exitflag': 0,
'time': 0.011858463287353516,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '5',
'x': array([1.00000063, 1.00000079, 1.00000065, 1.00000046, 1.00000018,
0.99999966, 0.99999884, 0.99999747, 0.99999501, 0.99999004]),
'fval': 3.4686626972583124e-10,
'grad': array([ 1.88621968e-04, 2.82853573e-04, 1.47613974e-04, 1.32427091e-04,
1.34368505e-04, 4.58042858e-05, -1.05802446e-05, -7.23486331e-05,
-8.94165854e-06, 5.57700990e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 98,
'n_grad': 98,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000063, 1.00000079, 1.00000065, 1.00000046, 1.00000018,
0.99999966, 0.99999884, 0.99999747, 0.99999501, 0.99999004]),
'fval0': 95086.57486034792,
'history': <pypesto.objective.history.History at 0x7fefee65a650>,
'exitflag': 0,
'time': 0.013525247573852539,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '12',
'x': array([0.99999993, 1.00000061, 1.00000043, 0.9999998 , 1.00000026,
1.00000048, 1.00000068, 1.00000084, 1.00000215, 1.00000338]),
'fval': 4.2760857236608074e-10,
'grad': array([-0.0003034 , 0.00046938, 0.00026867, -0.00047909, 0.00014639,
0.00010591, 0.00015162, -0.00028974, 0.00047086, -0.00018617]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 77,
'n_grad': 77,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999993, 1.00000061, 1.00000043, 0.9999998 , 1.00000026,
1.00000048, 1.00000068, 1.00000084, 1.00000215, 1.00000338]),
'fval0': 278737.0766282746,
'history': <pypesto.objective.history.History at 0x7fefee65a710>,
'exitflag': 0,
'time': 0.010965824127197266,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '1',
'x': array([0.99999964, 1.00000006, 0.99999975, 1.00000012, 0.99999977,
0.99999967, 0.99999847, 0.99999623, 0.9999936 , 0.99998718]),
'fval': 4.4824491354324123e-10,
'grad': array([-3.15440907e-04, 3.04897788e-04, -3.24323195e-04, 3.10014361e-04,
-1.40439929e-04, 3.69237832e-04, 1.05035392e-04, -6.01846648e-04,
2.19479487e-04, -2.71116767e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 81,
'n_grad': 81,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999964, 1.00000006, 0.99999975, 1.00000012, 0.99999977,
0.99999967, 0.99999847, 0.99999623, 0.9999936 , 0.99998718]),
'fval0': 231983.4016462493,
'history': <pypesto.objective.history.History at 0x7ff030770110>,
'exitflag': 0,
'time': 0.022760868072509766,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '15',
'x': array([-0.99326331, 0.99660594, 0.99824067, 0.9989884 , 0.99922624,
0.99907383, 0.99845405, 0.99705586, 0.9941786 , 0.98839115]),
'fval': 3.9865791123861647,
'grad': array([ 1.37836191e-05, -9.56428278e-05, 1.15714471e-04, -9.46304780e-05,
2.76772792e-05, 1.99653191e-04, -4.63338544e-05, -2.78221136e-05,
-2.26556385e-05, 1.07817815e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 72,
'n_grad': 72,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326331, 0.99660594, 0.99824067, 0.9989884 , 0.99922624,
0.99907383, 0.99845405, 0.99705586, 0.9941786 , 0.98839115]),
'fval0': 117406.38350731946,
'history': <pypesto.objective.history.History at 0x7fefee65a8d0>,
'exitflag': 0,
'time': 0.01103520393371582,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '2',
'x': array([-0.99326315, 0.99660609, 0.99824021, 0.99898815, 0.99922624,
0.99907353, 0.99845395, 0.99705588, 0.99417909, 0.98839179]),
'fval': 3.986579112503477,
'grad': array([ 1.91428162e-04, 2.94649756e-04, -3.04516493e-04, -1.62074006e-04,
2.52224941e-04, -6.28043726e-05, -3.29243223e-05, -1.67555012e-04,
1.96809844e-04, -5.29321104e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 78,
'n_grad': 78,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326315, 0.99660609, 0.99824021, 0.99898815, 0.99922624,
0.99907353, 0.99845395, 0.99705588, 0.99417909, 0.98839179]),
'fval0': 90060.0282425554,
'history': <pypesto.objective.history.History at 0x7fefee6d19d0>,
'exitflag': 0,
'time': 0.016544103622436523,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '3',
'x': array([-0.99326274, 0.99660589, 0.99824031, 0.99898835, 0.99922654,
0.99907324, 0.99845446, 0.99705616, 0.99417916, 0.98839222]),
'fval': 3.9865791128374686,
'grad': array([ 4.43724847e-04, 2.30363468e-04, -2.12480723e-04, -1.17788491e-04,
5.86174712e-04, -6.78812087e-04, 4.85626746e-04, -1.17917644e-04,
-1.48260191e-05, 3.53218137e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 68,
'n_grad': 68,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326274, 0.99660589, 0.99824031, 0.99898835, 0.99922654,
0.99907324, 0.99845446, 0.99705616, 0.99417916, 0.98839222]),
'fval0': 58537.15752301021,
'history': <pypesto.objective.history.History at 0x7ff0003b3c90>,
'exitflag': 0,
'time': 0.007930994033813477,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'}]
Plot results¶
[5]:
import pypesto.visualize
# plot waterfalls
pypesto.visualize.waterfall(result, size=(15,6))
[5]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fefef14a310>
Save optimization result as HDF5 file¶
[6]:
fn = tempfile.mktemp(".hdf5")
# Write result
hdf5_writer = save_to_hdf5.OptimizationResultHDF5Writer(fn)
hdf5_writer.write(result)
# Write problem
hdf5_writer = save_to_hdf5.ProblemHDF5Writer(fn)
hdf5_writer.write(problem)
[7]:
# Read result and problem
hdf5_reader = read_from_hdf5.OptimizationResultHDF5Reader(fn)
result = hdf5_reader.read()
[8]:
result.optimize_result.list
[8]:
[{'id': '8',
'x': array([1.00000001, 0.99999999, 0.99999997, 0.99999998, 0.99999998,
0.99999999, 0.99999996, 0.99999987, 0.99999971, 0.9999995 ]),
'fval': 1.4448869867069234e-12,
'grad': array([ 9.91613312e-06, -2.33793663e-07, -1.84487477e-05, -1.24826804e-06,
-7.03416051e-06, 1.12040576e-05, 1.88713028e-05, -4.68014961e-07,
-3.65179645e-05, 1.53152743e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 73,
'n_grad': 73,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 0.99999999, 0.99999997, 0.99999998, 0.99999998,
0.99999999, 0.99999996, 0.99999987, 0.99999971, 0.9999995 ]),
'fval0': 116315.16334351365,
'history': None,
'exitflag': 0,
'time': 0.010613441467285156,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '16',
'x': array([0.99999994, 0.99999998, 1. , 1.00000002, 1.00000004,
1.00000001, 1. , 1.00000012, 1.0000003 , 1.00000051]),
'fval': 4.371307796809753e-12,
'grad': array([-4.26293148e-05, 9.13631144e-06, -1.31339486e-06, 2.51280250e-06,
2.59501842e-05, 4.21294205e-07, -5.58158396e-05, 7.08567852e-07,
4.41611237e-05, -1.57413407e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 79,
'n_grad': 79,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999994, 0.99999998, 1. , 1.00000002, 1.00000004,
1.00000001, 1. , 1.00000012, 1.0000003 , 1.00000051]),
'fval0': 127542.57197202934,
'history': None,
'exitflag': 0,
'time': 0.011748075485229492,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '4',
'x': array([1.00000007, 0.99999999, 0.99999998, 1. , 1.00000008,
1.00000021, 1.00000015, 1.00000006, 1.00000004, 1.00000013]),
'fval': 1.7134261938953258e-11,
'grad': array([ 5.79464879e-05, -3.23661397e-05, -1.13616716e-05, -2.69343079e-05,
-1.67474293e-06, 1.20454131e-04, 3.83436764e-05, -1.71072644e-05,
-3.54339727e-05, 1.03840629e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 88,
'n_grad': 88,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000007, 0.99999999, 0.99999998, 1. , 1.00000008,
1.00000021, 1.00000015, 1.00000006, 1.00000004, 1.00000013]),
'fval0': 169143.33089007522,
'history': None,
'exitflag': 0,
'time': 0.016152620315551758,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '11',
'x': array([0.99999983, 0.99999997, 0.99999989, 0.9999998 , 1.00000006,
0.99999986, 0.99999963, 0.99999927, 0.99999861, 0.99999709]),
'fval': 4.473573948366185e-11,
'grad': array([-1.25925183e-04, 8.34342658e-05, -1.58946249e-05, -1.85224905e-04,
2.00742516e-04, -1.80384056e-05, -1.83734314e-05, -2.93938826e-05,
5.98755497e-05, -2.43744695e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 78,
'n_grad': 78,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999983, 0.99999997, 0.99999989, 0.9999998 , 1.00000006,
0.99999986, 0.99999963, 0.99999927, 0.99999861, 0.99999709]),
'fval0': 111440.55513257613,
'history': None,
'exitflag': 0,
'time': 0.009877920150756836,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '0',
'x': array([1.00000016, 1.00000029, 1.00000013, 1.00000007, 0.99999984,
0.99999968, 0.99999955, 0.99999932, 0.99999871, 0.99999745]),
'fval': 4.512690733773355e-11,
'grad': array([ 6.88017381e-06, 1.82437618e-04, -1.71219792e-05, 7.83029016e-05,
-5.64629619e-05, -7.75613657e-05, -5.33124129e-05, 2.00358870e-05,
-3.38990540e-06, 6.36234430e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 80,
'n_grad': 80,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000016, 1.00000029, 1.00000013, 1.00000007, 0.99999984,
0.99999968, 0.99999955, 0.99999932, 0.99999871, 0.99999745]),
'fval0': 179787.03971937217,
'history': None,
'exitflag': 0,
'time': 0.05712604522705078,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '18',
'x': array([0.9999999 , 0.99999988, 1.00000002, 1.00000017, 1.00000034,
1.00000045, 1.00000052, 1.00000076, 1.00000146, 1.00000315]),
'fval': 4.530338040953872e-11,
'grad': array([-3.46616032e-05, -8.55052094e-05, 4.32179353e-06, 2.27795791e-05,
8.77325561e-05, 1.11150847e-04, 4.14626291e-05, -3.72317820e-05,
-9.65698380e-05, 4.47893709e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 85,
'n_grad': 85,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.9999999 , 0.99999988, 1.00000002, 1.00000017, 1.00000034,
1.00000045, 1.00000052, 1.00000076, 1.00000146, 1.00000315]),
'fval0': 84246.35907849146,
'history': None,
'exitflag': 0,
'time': 0.01163339614868164,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '17',
'x': array([1.00000014, 1.00000006, 1.00000012, 1.00000002, 0.99999972,
0.99999976, 0.99999924, 0.9999986 , 0.99999728, 0.99999457]),
'fval': 5.187501773111393e-11,
'grad': array([ 8.98076519e-05, -4.45109249e-05, 8.61160519e-05, 8.83761172e-05,
-1.98032428e-04, 1.80982671e-04, -1.05227326e-04, -1.64856814e-05,
1.06897803e-05, 1.09849767e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 99,
'n_grad': 99,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000014, 1.00000006, 1.00000012, 1.00000002, 0.99999972,
0.99999976, 0.99999924, 0.9999986 , 0.99999728, 0.99999457]),
'fval0': 164257.74387447865,
'history': None,
'exitflag': 0,
'time': 0.013841629028320312,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '13',
'x': array([1.00000009, 1.00000004, 1.00000001, 1.0000001 , 1.00000019,
1.00000029, 1.00000006, 1.00000059, 1.00000094, 1.00000194]),
'fval': 6.113882079371062e-11,
'grad': array([ 5.38201966e-05, 1.12643005e-06, -4.79180197e-05, 1.70876435e-05,
3.76896795e-05, 1.90462489e-04, -2.94335966e-04, 1.92128181e-04,
-7.02675278e-05, 1.21112443e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 84,
'n_grad': 84,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000009, 1.00000004, 1.00000001, 1.0000001 , 1.00000019,
1.00000029, 1.00000006, 1.00000059, 1.00000094, 1.00000194]),
'fval0': 128057.90608516608,
'history': None,
'exitflag': 0,
'time': 0.011581659317016602,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '9',
'x': array([1.0000003 , 1.00000024, 0.99999999, 0.99999981, 0.99999975,
0.99999984, 1.00000003, 1.00000007, 1.00000014, 1.00000029]),
'fval': 6.603657918190325e-11,
'grad': array([ 1.45585786e-04, 1.26131400e-04, -3.15965052e-05, -8.20696700e-05,
-1.17487544e-04, -6.42577094e-05, 5.98749705e-05, 4.78675947e-06,
-7.72296480e-06, 3.75853482e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 83,
'n_grad': 83,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.0000003 , 1.00000024, 0.99999999, 0.99999981, 0.99999975,
0.99999984, 1.00000003, 1.00000007, 1.00000014, 1.00000029]),
'fval0': 66001.21516931924,
'history': None,
'exitflag': 0,
'time': 0.012688159942626953,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '19',
'x': array([1.00000001, 1. , 0.99999996, 0.99999988, 1.00000026,
1.00000032, 1.00000033, 1.00000085, 1.00000187, 1.00000333]),
'fval': 6.709274183733498e-11,
'grad': array([ 8.70631729e-06, 9.70587008e-06, 1.05693559e-05, -2.12546166e-04,
1.84978271e-04, 8.28659608e-05, -1.39703429e-04, -2.30820323e-05,
1.96299302e-04, -8.02670983e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 67,
'n_grad': 67,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 1. , 0.99999996, 0.99999988, 1.00000026,
1.00000032, 1.00000033, 1.00000085, 1.00000187, 1.00000333]),
'fval0': 80442.06208067665,
'history': None,
'exitflag': 0,
'time': 0.008251428604125977,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '10',
'x': array([0.9999998 , 1.00000012, 0.99999997, 1.00000017, 1.00000028,
1.00000037, 1.0000004 , 1.0000005 , 1.00000069, 1.00000195]),
'fval': 1.065068443513822e-10,
'grad': array([-2.07797214e-04, 2.08384202e-04, -1.40258436e-04, 6.72135025e-05,
6.29120089e-05, 9.92424271e-05, 5.28220148e-05, 6.30692991e-05,
-2.90590939e-04, 1.15498187e-04]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 135,
'n_grad': 135,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.9999998 , 1.00000012, 0.99999997, 1.00000017, 1.00000028,
1.00000037, 1.0000004 , 1.0000005 , 1.00000069, 1.00000195]),
'fval0': 218642.53588542074,
'history': None,
'exitflag': 0,
'time': 0.017060041427612305,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '6',
'x': array([1.00000009, 0.9999998 , 1.00000026, 1.00000009, 1.00000031,
1.00000016, 1.00000029, 1.00000052, 1.00000131, 1.00000213]),
'fval': 1.3507352118480165e-10,
'grad': array([ 1.50715573e-04, -3.42587714e-04, 3.08461493e-04, -1.41826830e-04,
2.10332130e-04, -7.64162802e-05, 1.68271418e-05, -1.20534088e-04,
2.55005501e-04, -9.87312659e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 86,
'n_grad': 86,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000009, 0.9999998 , 1.00000026, 1.00000009, 1.00000031,
1.00000016, 1.00000029, 1.00000052, 1.00000131, 1.00000213]),
'fval0': 49550.65276671963,
'history': None,
'exitflag': 0,
'time': 0.012197017669677734,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '14',
'x': array([1.00000021, 0.99999999, 0.99999968, 0.99999939, 0.99999938,
0.99999947, 0.9999996 , 0.99999955, 0.99999928, 0.99999839]),
'fval': 1.76508815611245e-10,
'grad': array([ 1.69974542e-04, 3.68276755e-05, -7.46786006e-05, -2.36880611e-04,
-1.60596472e-04, -1.26090555e-04, -5.76633224e-06, -1.01269226e-06,
9.59426070e-05, -3.14643190e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 94,
'n_grad': 94,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000021, 0.99999999, 0.99999968, 0.99999939, 0.99999938,
0.99999947, 0.9999996 , 0.99999955, 0.99999928, 0.99999839]),
'fval0': 113189.5063880412,
'history': None,
'exitflag': 0,
'time': 0.010685205459594727,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '7',
'x': array([1.00000001, 1.00000008, 1.00000032, 0.9999997 , 0.99999988,
0.9999994 , 0.99999945, 0.99999874, 0.9999976 , 0.99999541]),
'fval': 1.8629501787028135e-10,
'grad': array([-1.89690853e-05, -5.75176850e-05, 4.08684636e-04, -3.79119426e-04,
2.41840701e-04, -3.35502015e-04, 1.94400780e-04, -8.41625352e-05,
-6.36788959e-05, 4.17276302e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 86,
'n_grad': 86,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000001, 1.00000008, 1.00000032, 0.9999997 , 0.99999988,
0.9999994 , 0.99999945, 0.99999874, 0.9999976 , 0.99999541]),
'fval0': 275340.0482345366,
'history': None,
'exitflag': 0,
'time': 0.011858463287353516,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '5',
'x': array([1.00000063, 1.00000079, 1.00000065, 1.00000046, 1.00000018,
0.99999966, 0.99999884, 0.99999747, 0.99999501, 0.99999004]),
'fval': 3.4686626972583124e-10,
'grad': array([ 1.88621968e-04, 2.82853573e-04, 1.47613974e-04, 1.32427091e-04,
1.34368505e-04, 4.58042858e-05, -1.05802446e-05, -7.23486331e-05,
-8.94165854e-06, 5.57700990e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 98,
'n_grad': 98,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([1.00000063, 1.00000079, 1.00000065, 1.00000046, 1.00000018,
0.99999966, 0.99999884, 0.99999747, 0.99999501, 0.99999004]),
'fval0': 95086.57486034792,
'history': None,
'exitflag': 0,
'time': 0.013525247573852539,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '12',
'x': array([0.99999993, 1.00000061, 1.00000043, 0.9999998 , 1.00000026,
1.00000048, 1.00000068, 1.00000084, 1.00000215, 1.00000338]),
'fval': 4.2760857236608074e-10,
'grad': array([-0.0003034 , 0.00046938, 0.00026867, -0.00047909, 0.00014639,
0.00010591, 0.00015162, -0.00028974, 0.00047086, -0.00018617]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 77,
'n_grad': 77,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999993, 1.00000061, 1.00000043, 0.9999998 , 1.00000026,
1.00000048, 1.00000068, 1.00000084, 1.00000215, 1.00000338]),
'fval0': 278737.0766282746,
'history': None,
'exitflag': 0,
'time': 0.010965824127197266,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '1',
'x': array([0.99999964, 1.00000006, 0.99999975, 1.00000012, 0.99999977,
0.99999967, 0.99999847, 0.99999623, 0.9999936 , 0.99998718]),
'fval': 4.4824491354324123e-10,
'grad': array([-3.15440907e-04, 3.04897788e-04, -3.24323195e-04, 3.10014361e-04,
-1.40439929e-04, 3.69237832e-04, 1.05035392e-04, -6.01846648e-04,
2.19479487e-04, -2.71116767e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 81,
'n_grad': 81,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([0.99999964, 1.00000006, 0.99999975, 1.00000012, 0.99999977,
0.99999967, 0.99999847, 0.99999623, 0.9999936 , 0.99998718]),
'fval0': 231983.4016462493,
'history': None,
'exitflag': 0,
'time': 0.022760868072509766,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '15',
'x': array([-0.99326331, 0.99660594, 0.99824067, 0.9989884 , 0.99922624,
0.99907383, 0.99845405, 0.99705586, 0.9941786 , 0.98839115]),
'fval': 3.9865791123861647,
'grad': array([ 1.37836191e-05, -9.56428278e-05, 1.15714471e-04, -9.46304780e-05,
2.76772792e-05, 1.99653191e-04, -4.63338544e-05, -2.78221136e-05,
-2.26556385e-05, 1.07817815e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 72,
'n_grad': 72,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326331, 0.99660594, 0.99824067, 0.9989884 , 0.99922624,
0.99907383, 0.99845405, 0.99705586, 0.9941786 , 0.98839115]),
'fval0': 117406.38350731946,
'history': None,
'exitflag': 0,
'time': 0.01103520393371582,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '2',
'x': array([-0.99326315, 0.99660609, 0.99824021, 0.99898815, 0.99922624,
0.99907353, 0.99845395, 0.99705588, 0.99417909, 0.98839179]),
'fval': 3.986579112503477,
'grad': array([ 1.91428162e-04, 2.94649756e-04, -3.04516493e-04, -1.62074006e-04,
2.52224941e-04, -6.28043726e-05, -3.29243223e-05, -1.67555012e-04,
1.96809844e-04, -5.29321104e-05]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 78,
'n_grad': 78,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326315, 0.99660609, 0.99824021, 0.99898815, 0.99922624,
0.99907353, 0.99845395, 0.99705588, 0.99417909, 0.98839179]),
'fval0': 90060.0282425554,
'history': None,
'exitflag': 0,
'time': 0.016544103622436523,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'},
{'id': '3',
'x': array([-0.99326274, 0.99660589, 0.99824031, 0.99898835, 0.99922654,
0.99907324, 0.99845446, 0.99705616, 0.99417916, 0.98839222]),
'fval': 3.9865791128374686,
'grad': array([ 4.43724847e-04, 2.30363468e-04, -2.12480723e-04, -1.17788491e-04,
5.86174712e-04, -6.78812087e-04, 4.85626746e-04, -1.17917644e-04,
-1.48260191e-05, 3.53218137e-06]),
'hess': None,
'res': None,
'sres': None,
'n_fval': 68,
'n_grad': 68,
'n_hess': 0,
'n_res': 0,
'n_sres': 0,
'x0': array([-0.99326274, 0.99660589, 0.99824031, 0.99898835, 0.99922654,
0.99907324, 0.99845446, 0.99705616, 0.99417916, 0.98839222]),
'fval0': 58537.15752301021,
'history': None,
'exitflag': 0,
'time': 0.007930994033813477,
'message': b'CONVERGENCE: REL_REDUCTION_OF_F_<=_FACTR*EPSMCH'}]
Plot results¶
[9]:
# plot waterfalls
pypesto.visualize.waterfall(result, size=(15,6))
[9]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fefec49cbd0>
[10]:
pypesto.visualize.parameters(result,
balance_alpha=False)
[10]:
<matplotlib.axes._subplots.AxesSubplot at 0x7fefebca04d0>
[ ]:
A sampler study¶
In this notebook, we perform a short study of how various samplers implemented in pyPESTO perform.
The pipeline¶
First, we show a typical workflow, fully integrating the samplers with a PEtab problem, using a toy example of a conversion reaction.
[1]:
import pypesto
import petab
# import to petab
petab_problem = petab.Problem.from_yaml(
"conversion_reaction/conversion_reaction.yaml")
# import to pypesto
importer = pypesto.PetabImporter(petab_problem)
# create problem
problem = importer.create_problem()
Commonly, as a first step, optimization is performed, in order to find good parameter point estimates.
[2]:
result = pypesto.minimize(problem, n_starts=10)
[3]:
pypesto.visualize.waterfall(result, size=(4,4))
[3]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f1ccd9d8090>
Next, we perform sampling. Here, we employ a pypesto.sample.AdaptiveParallelTemperingSampler sampler, which runs Markov Chain Monte Carlo (MCMC) chains on different temperatures. For each chain, we employ a pypesto.sample.AdaptiveMetropolisSampler. For more on the samplers see below or the API documentation.
[4]:
sampler = pypesto.AdaptiveParallelTemperingSampler(
internal_sampler=pypesto.AdaptiveMetropolisSampler(),
n_chains=3)
For the actual sampling, we call the pypesto.sample function. By passing the result object to the function, the previously found global optimum is used as starting point for the MCMC sampling.
[5]:
result = pypesto.sample(problem, n_samples=10000, sampler=sampler, result=result)
When the sampling is finished, we can analyse our results. pyPESTO provides functions to analyse both the sampling process as well as the obtained sampling result. Visualizing the traces e.g. allows to detect burn-in phases, or fine-tune hyperparameters. First, the parameter trajectories can be visualized:
[6]:
pypesto.visualize.sampling_parameters_trace(result, use_problem_bounds=False)
[6]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f1cccf78dd0>
Next, also the log posterior trace can be visualized:
[7]:
pypesto.visualize.sampling_fval_trace(result)
[7]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f1cccf16b90>
To visualize the result, there are various options. The scatter plot shows histograms of 1-dim parameter marginals and scatter plots of 2-dimensional parameter combinations:
[8]:
pypesto.visualize.sampling_scatter(result, size=[13,6])
[8]:
<seaborn.axisgrid.PairGrid at 0x7f1d12243cd0>
sampling_1d_marginals allows to plot e.g. kernel density estimates or histograms (internally using seaborn):
[9]:
for i_chain in range(len(result.sample_result.betas)):
pypesto.visualize.sampling_1d_marginals(
result, i_chain=i_chain, suptitle=f"Chain: {i_chain}")
That’s it for the moment on using the sampling pipeline.
1-dim test problem¶
To compare and test the various implemented samplers, we first study a 1-dimensional test problem of a gaussian mixture density, together with a flat prior.
[10]:
import numpy as np
from scipy.stats import multivariate_normal
import seaborn as sns
import pypesto
def density(x):
return 0.3*multivariate_normal.pdf(x, mean=-1.5, cov=0.1) + \
0.7*multivariate_normal.pdf(x, mean=2.5, cov=0.2)
def p(x):
return - np.log(density(x))
objective = pypesto.Objective(fun=p)
problem = pypesto.Problem(
objective=objective, lb=np.array(-10), ub=np.array(10), x_names=['x'])
The likelihood has two separate modes:
[11]:
xs = np.linspace(-10, 10, 100)
ys = [density(x) for x in xs]
sns.lineplot(xs, ys, color='C1')
[11]:
<matplotlib.axes._subplots.AxesSubplot at 0x7f1cbb116550>
Metropolis sampler¶
For this problem, let us try out the simplest sampler, the pypesto.sample.MetropolisSampler.
[12]:
sampler = pypesto.MetropolisSampler({'std': 0.5})
result = pypesto.sample(problem, 1e4, sampler, x0=np.array([0.5]))
ax = pypesto.visualize.sampling_1d_marginals(result)
ax[0][0].plot(xs, ys)
[12]:
[<matplotlib.lines.Line2D at 0x7f1cccffaa50>]
The obtained posterior does not accurately represent the distribution, often only capturing one mode. This is because it is hard for the Markov chain to jump between the distribution’s two modes. This can be fixed by choosing a higher proposal variation std:
[13]:
sampler = pypesto.MetropolisSampler({'std': 1})
result = pypesto.sample(problem, 1e4, sampler, x0=np.array([0.5]))
ax = pypesto.visualize.sampling_1d_marginals(result)
ax[0][0].plot(xs, ys)
[13]:
[<matplotlib.lines.Line2D at 0x7f1cba518750>]
In general, MCMC have difficulties exploring multimodel landscapes. One way to overcome this is to used parallel tempering. There, various chains are run, lifting the densities to different temperatures. At high temperatures, proposed steps are more likely to get accepted and thus jumps between modes more likely.
Parallel tempering sampler¶
In pyPESTO, the most basic parallel tempering algorithm is the pypesto.sample.ParallelTemperingSampler. It takes an internal_sampler parameter, to specify what sampler to use for performing sampling the different chains. Further, we can directly specify what inverse temperatures betas to use. When not specifying the betas explicitly but just the number of chains n_chains, an established near-exponential decay scheme is used.
[14]:
sampler = pypesto.ParallelTemperingSampler(
internal_sampler=pypesto.MetropolisSampler(),
betas=[1, 1e-1, 1e-2])
result = pypesto.sample(problem, 1e4, sampler, x0=np.array([0.5]))
[15]:
for i_chain in range(len(result.sample_result.betas)):
pypesto.visualize.sampling_1d_marginals(
result, i_chain=i_chain, suptitle=f"Chain: {i_chain}")
Of interest is here finally the first chain at index i_chain=0, which approximates the posterior well.
Adaptive Metropolis sampler¶
The problem of having to specify the proposal step variation manually can be overcome by using the pypesto.sample.AdaptiveMetropolisSampler, which iteratively adjusts the proposal steps to the function landscape.
[16]:
sampler = pypesto.AdaptiveMetropolisSampler()
result = pypesto.sample(problem, 1e4, sampler, x0=np.array([0.5]))
[17]:
pypesto.visualize.sampling_1d_marginals(result)
[17]:
array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7f1cbc3c6290>]],
dtype=object)
Adaptive parallel tempering sampler¶
The pypesto.sample.AdaptiveParallelTemperingSampler iteratively adjusts the temperatures to obtain good swapping rates between chains.
[18]:
sampler = pypesto.AdaptiveParallelTemperingSampler(
internal_sampler=pypesto.AdaptiveMetropolisSampler(), n_chains=3)
result = pypesto.sample(problem, 1e4, sampler, x0=np.array([0.5]))
[19]:
for i_chain in range(len(result.sample_result.betas)):
pypesto.visualize.sampling_1d_marginals(
result, i_chain=i_chain, suptitle=f"Chain: {i_chain}")
[20]:
result.sample_result.betas
[20]:
array([1.00000000e+00, 8.02757714e-02, 2.00000000e-05])
2-dim test problem: Rosenbrock banana¶
The adaptive parallel tempering sampler with chains running adaptive Metropolis samplers is also able to sample from more challenging posterior distributions. To illustrates this shortly, we use the Rosenbrock function.
[21]:
import scipy.optimize as so
import pypesto
# first type of objective
objective = pypesto.Objective(fun=so.rosen)
dim_full = 4
lb = -5 * np.ones((dim_full, 1))
ub = 5 * np.ones((dim_full, 1))
problem = pypesto.Problem(objective=objective, lb=lb, ub=ub)
[23]:
sampler = pypesto.AdaptiveParallelTemperingSampler(
internal_sampler=pypesto.AdaptiveMetropolisSampler(), n_chains=10)
result = pypesto.sample(problem, 1e4, sampler, x0=np.zeros(dim_full))
[24]:
pypesto.visualize.sampling_scatter(result)
pypesto.visualize.sampling_1d_marginals(result)
[24]:
array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7f1cbb524390>,
<matplotlib.axes._subplots.AxesSubplot object at 0x7f1cbc726250>],
[<matplotlib.axes._subplots.AxesSubplot object at 0x7f1cbcb1d310>,
<matplotlib.axes._subplots.AxesSubplot object at 0x7f1cbcac95d0>]],
dtype=object)
[ ]:
Download the examples as notebooks¶
Note
Some of the notebooks have extra dependencies.
Contribute¶
Contribute documentation¶
To make pypesto easily usable, we are committed to documenting extensively. This involves in particular documenting the functionality of methods and classes, the purpose of single lines of code, and giving usage examples. The documentation is hosted on pypesto.readthedocs.io and updated automatically every time the master branch on github.com/icb-dcm/pypesto is updated. To compile the documentation locally, use:
cd doc
make html
Contribute tests¶
Tests are located in the test folder. All files starting with test_
contain tests and are automatically run on Travis CI. To run them manually,
type:
python3 -m pytest test
or alternatively:
python3 -m unittest test
You can also run specific tests.
Tests can be written with pytest or the unittest module.
PEP8¶
We try to respect the PEP8 coding standards. We run flake8 as part of the tests. If flake8 complains, the tests won’t pass. You can run it via:
./run_flake8.sh
in Linux from the base directory, or directly from python. More, you can use the tool autopep8 to automatically fix various coding issues.
Contribute code¶
If you start working on a new feature or a fix, if not already done, please create an issue on github shortly describing your plans and assign it to yourself.
To get your code merged, please:
- create a pull request to develop
- if not already done in a commit message already, use the pull request description to reference and automatically close the respective issue (see https://help.github.com/articles/closing-issues-using-keywords/)
- check that all tests on travis pass
- check that the documentation is up-to-date
- request a code review
General notes:
- Internally, we use
numpyfor arrays. In particular, vectors are represented as arrays of shape (n,). - Use informative commmit messages.
Deploy¶
New features and bug fixes are continuously added to the develop branch. On
every merge to master, the version number in pypesto/version.py should
be incremented as described below.
Versioning scheme¶
For version numbers, we use A.B.C, where
Cis increased for bug fixes,Bis increased for new features and minor API breaking changes,Ais increased for major API breaking changes.
Creating a new release¶
After new commits have been added to the develop branch, changes can be merged to master and a new version of pyPESTO can be released. Every merge to master should coincide with an incremented version number and a git tag on the respective merge commit.
Merge into master¶
- create a pull request from develop to master
- check that all tests on travis pass
- check that the documentation is up-to-date
- adapt the version number in the file
pesto/version.py(see above) - update the release notes in
doc/releasenotes.rst - request a code review
- merge into the origin master branch
To be able to actually perform the merge, sufficient rights may be required. Also, at least one review is required.
Creating a release on github¶
After merging into master, create a new release on Github. In the release form:
- specify a tag with the new version as specified in
pesto/version.py, prefixed withv(e.g.v0.0.1) - include the latest additions to
doc/releasenotes.rstin the release description
Tagging the release commit will automatically trigger deployment of the new version to pypi.
Objective¶
-
class
pypesto.objective.AggregatedObjective(objectives: List[pypesto.objective.objective.Objective], x_names: List[str] = None)¶ Bases:
pypesto.objective.objective.ObjectiveThis class aggregates multiple objectives into one objective.
-
__call__()¶ Method to obtain arbitrary sensitivities. This is the central method which is always called, also by the get_* methods.
There are different ways in which an optimizer calls the objective function, and in how the objective function provides information (e.g. derivatives via separate functions or along with the function values). The different calling modes increase efficiency in space and time and make the objective flexible.
Parameters: - x – The parameters for which to evaluate the objective function.
- sensi_orders – Specifies which sensitivities to compute, e.g. (0,1) -> fval, grad.
- mode – Whether to compute function values or residuals.
- return_dict – If False (default), the result is a Tuple of the requested values in the requested order. Tuples of length one are flattened. If True, instead a dict is returned which can carry further information.
Returns: By default, this is a tuple of the requested function values and derivatives in the requested order (if only 1 value, the tuple is flattened). If return_dict, then instead a dict is returned with function values and derivatives indicated by ids.
Return type: result
-
__class__¶ alias of
builtins.type
-
__deepcopy__(memodict=None)¶
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.aggregated', '__doc__': '\n This class aggregates multiple objectives into one objective.\n ', '__init__': <function AggregatedObjective.__init__>, '__deepcopy__': <function AggregatedObjective.__deepcopy__>, 'aggregate_fun_sensi_orders': <function AggregatedObjective.aggregate_fun_sensi_orders>, 'aggregate_res_sensi_orders': <function AggregatedObjective.aggregate_res_sensi_orders>, 'aggregate_res': <function AggregatedObjective.aggregate_res>, 'aggregate_sres': <function AggregatedObjective.aggregate_sres>, 'aggregate_fun': <function AggregatedObjective.aggregate_fun>, 'aggregate_grad': <function AggregatedObjective.aggregate_grad>, 'aggregate_hess': <function AggregatedObjective.aggregate_hess>, 'aggregate_hessp': <function AggregatedObjective.aggregate_hessp>, 'reset_steadystate_guesses': <function AggregatedObjective.reset_steadystate_guesses>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(objectives: List[pypesto.objective.objective.Objective], x_names: List[str] = None)¶ Constructor.
Parameters: objectives (list) – List of pypesto.objetive instances
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.aggregated'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
aggregate_fun(x)¶
-
aggregate_fun_sensi_orders(x, sensi_orders)¶
-
aggregate_grad(x)¶
-
aggregate_hess(x)¶
-
aggregate_hessp(x)¶
-
aggregate_res(x)¶
-
aggregate_res_sensi_orders(x, sensi_orders)¶
-
aggregate_sres(x)¶
-
check_grad(x: numpy.ndarray, x_indices: List[int] = None, eps: float = 1e-05, verbosity: int = 1, mode: str = 'mode_fun') → pandas.core.frame.DataFrame¶ Compare gradient evaluation: Firstly approximate via finite differences, and secondly use the objective gradient.
Parameters: - x – The parameters for which to evaluate the gradient.
- x_indices – List of index values for which to compute gradients. Default: all.
- eps – Finite differences step size. Default: 1e-5.
- verbosity – Level of verbosity for function output. * 0: no output, * 1: summary for all parameters, * 2: summary for individual parameters. Default: 1.
- mode – Residual (MODE_RES) or objective function value (MODE_FUN, default) computation mode.
Returns: gradient, finite difference approximations and error estimates.
Return type: result
-
check_sensi_orders(sensi_orders, mode) → None¶ Check if the objective is able to compute the requested sensitivities. If not, throw an exception.
Raises: - ValueError if the objective function cannot be called as
requested.
-
get_fval(x: numpy.ndarray) → float¶ Get the function value at x.
-
get_grad(x: numpy.ndarray) → numpy.ndarray¶ Get the gradient at x.
-
get_hess(x: numpy.ndarray) → numpy.ndarray¶ Get the Hessian at x.
-
get_res(x: numpy.ndarray) → numpy.ndarray¶ Get the residuals at x.
-
get_sres(x: numpy.ndarray) → numpy.ndarray¶ Get the residual sensitivities at x.
-
has_fun¶
-
has_grad¶
-
has_hess¶
-
has_hessp¶
-
has_res¶
-
has_sres¶
-
initialize()¶ Initialize the objective function. This function is used at the beginning of an analysis, e.g. optimization, and can e.g. reset the objective memory. By default does nothing.
-
static
output_to_dict()¶ Convert output tuple to dict.
-
static
output_to_tuple()¶ Return values as requested by the caller, since usually only a subset is demanded. One output is returned as-is, more than one output are returned as a tuple in order (fval, grad, hess).
-
reset_steadystate_guesses()¶ Propagates reset_steadystate_guesses() to child objectives if available (currently only applies for amici_objective)
-
update_from_problem(dim_full: int, x_free_indices: List[int], x_fixed_indices: List[int], x_fixed_vals: List[int])¶ Handle fixed parameters. Later, the objective will be given parameter vectors x of dimension dim, which have to be filled up with fixed parameter values to form a vector of dimension dim_full >= dim. This vector is then used to compute function value and derivatives. The derivatives must later be reduced again to dimension dim.
This is so as to make the fixing of parameters transparent to the caller.
The methods preprocess, postprocess are overwritten for the above functionality, respectively.
Parameters: - dim_full – Dimension of the full vector including fixed parameters.
- x_free_indices – Vector containing the indices (zero-based) of free parameters (complimentary to x_fixed_indices).
- x_fixed_indices – Vector containing the indices (zero-based) of parameter components that are not to be optimized.
- x_fixed_vals – Vector of the same length as x_fixed_indices, containing the values of the fixed parameters.
-
-
class
pypesto.objective.AmiciCalculator¶ Bases:
objectClass to perform the actual call to AMICI and obtain requested objective function values.
-
__call__(x_dct: Dict, sensi_order: int, mode: str, amici_model: Union[amici.Model, amici.ModelPtr], amici_solver: Union[amici.Solver, amici.SolverPtr], edatas: List[amici.ExpData], n_threads: int, x_ids: Sequence[str], parameter_mapping: ParameterMapping)¶ Perform the actual AMICI call.
Called within the
AmiciObjective.__call__()method.Parameters: - x_dct – Parameters for which to compute function value and derivatives.
- sensi_order – Maximum sensitivity order.
- mode – Call mode (function value or residual based).
- amici_model – The AMICI model.
- amici_solver – The AMICI solver.
- edatas – The experimental data.
- n_threads – Number of threads for AMICI call.
- x_ids – Ids of optimization parameters.
- parameter_mapping – Mapping of optimization to simulation parameters.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.amici_calculator', '__doc__': '\n Class to perform the actual call to AMICI and obtain requested objective\n function values.\n ', 'initialize': <function AmiciCalculator.initialize>, '__call__': <function AmiciCalculator.__call__>, '__dict__': <attribute '__dict__' of 'AmiciCalculator' objects>, '__weakref__': <attribute '__weakref__' of 'AmiciCalculator' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.amici_calculator'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
initialize()¶ Initialize the calculator. Default: Do nothing.
-
-
class
pypesto.objective.AmiciObjectBuilder¶ Bases:
abc.ABCAllows to build AMICI model, solver, and edatas.
This class is useful for pickling an
pypesto.AmiciObjective, which is required in some parallelization schemes. Therefore, this class itself must be picklable.-
__abstractmethods__= frozenset({'create_edatas', 'create_model', 'create_solver'})¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.amici_objective', '__doc__': 'Allows to build AMICI model, solver, and edatas.\n\n This class is useful for pickling an :class:`pypesto.AmiciObjective`,\n which is required in some parallelization schemes. Therefore, this\n class itself must be picklable.\n ', 'create_model': <function AmiciObjectBuilder.create_model>, 'create_solver': <function AmiciObjectBuilder.create_solver>, 'create_edatas': <function AmiciObjectBuilder.create_edatas>, '__abstractmethods__': frozenset({'create_edatas', 'create_model', 'create_solver'}), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.amici_objective'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
create_edatas(model: Union[amici.Model, amici.ModelPtr]) → Sequence[amici.ExpData]¶ Create AMICI experimental data.
-
create_model() → Union[amici.Model, amici.ModelPtr]¶ Create an AMICI model.
-
create_solver(model: Union[amici.Model, amici.ModelPtr]) → Union[amici.Solver, amici.SolverPtr]¶ Create an AMICI solver.
-
-
class
pypesto.objective.AmiciObjective(amici_model: Union[amici.Model, amici.ModelPtr], amici_solver: Union[amici.Solver, amici.SolverPtr], edatas: Union[Sequence[amici.ExpData], amici.ExpData], max_sensi_order: int = None, x_ids: Sequence[str] = None, x_names: Sequence[str] = None, parameter_mapping: ParameterMapping = None, guess_steadystate: bool = True, n_threads: int = 1, amici_object_builder: pypesto.objective.amici_objective.AmiciObjectBuilder = None, calculator: pypesto.objective.amici_calculator.AmiciCalculator = None)¶ Bases:
pypesto.objective.objective.ObjectiveThis class allows to create an objective directly from an amici model.
-
__call__()¶ Method to obtain arbitrary sensitivities. This is the central method which is always called, also by the get_* methods.
There are different ways in which an optimizer calls the objective function, and in how the objective function provides information (e.g. derivatives via separate functions or along with the function values). The different calling modes increase efficiency in space and time and make the objective flexible.
Parameters: - x – The parameters for which to evaluate the objective function.
- sensi_orders – Specifies which sensitivities to compute, e.g. (0,1) -> fval, grad.
- mode – Whether to compute function values or residuals.
- return_dict – If False (default), the result is a Tuple of the requested values in the requested order. Tuples of length one are flattened. If True, instead a dict is returned which can carry further information.
Returns: By default, this is a tuple of the requested function values and derivatives in the requested order (if only 1 value, the tuple is flattened). If return_dict, then instead a dict is returned with function values and derivatives indicated by ids.
Return type: result
-
__class__¶ alias of
builtins.type
-
__deepcopy__(memodict: Dict = None) → pypesto.objective.amici_objective.AmiciObjective¶
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.amici_objective', '__doc__': '\n This class allows to create an objective directly from an amici model.\n ', '__init__': <function AmiciObjective.__init__>, 'initialize': <function AmiciObjective.initialize>, 'get_bound_fun': <function AmiciObjective.get_bound_fun>, 'get_bound_res': <function AmiciObjective.get_bound_res>, 'rebind_fun': <function AmiciObjective.rebind_fun>, 'rebind_res': <function AmiciObjective.rebind_res>, '__deepcopy__': <function AmiciObjective.__deepcopy__>, '__getstate__': <function AmiciObjective.__getstate__>, '__setstate__': <function AmiciObjective.__setstate__>, '_call_amici': <function AmiciObjective._call_amici>, 'par_arr_to_dct': <function AmiciObjective.par_arr_to_dct>, 'apply_steadystate_guess': <function AmiciObjective.apply_steadystate_guess>, 'store_steadystate_guess': <function AmiciObjective.store_steadystate_guess>, 'reset_steadystate_guesses': <function AmiciObjective.reset_steadystate_guesses>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__getstate__() → Dict¶
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(amici_model: Union[amici.Model, amici.ModelPtr], amici_solver: Union[amici.Solver, amici.SolverPtr], edatas: Union[Sequence[amici.ExpData], amici.ExpData], max_sensi_order: int = None, x_ids: Sequence[str] = None, x_names: Sequence[str] = None, parameter_mapping: ParameterMapping = None, guess_steadystate: bool = True, n_threads: int = 1, amici_object_builder: pypesto.objective.amici_objective.AmiciObjectBuilder = None, calculator: pypesto.objective.amici_calculator.AmiciCalculator = None)¶ Constructor.
Parameters: - amici_model – The amici model.
- amici_solver – The solver to use for the numeric integration of the model.
- edatas – The experimental data. If a list is passed, its entries correspond to multiple experimental conditions.
- max_sensi_order – Maximum sensitivity order supported by the model. Defaults to 2 if the model was compiled with o2mode, otherwise 1.
- x_ids – Ids of optimization parameters. In the simplest case, this will be the AMICI model parameters (default).
- x_names – Names of optimization parameters.
- parameter_mapping – Mapping of optimization parameters to model parameters. Format as created by amici.petab_objective.create_parameter_mapping. The default is just to assume that optimization and simulation parameters coincide.
- guess_steadystate – Whether to guess steadystates based on previous steadystates and respective derivatives. This option may lead to unexpected results for models with conservation laws and should accordingly be deactivated for those models.
- n_threads – Number of threads that are used for parallelization over experimental conditions. If amici was not installed with openMP support this option will have no effect.
- amici_object_builder – AMICI object builder. Allows recreating the objective for pickling, required in some parallelization schemes.
- calculator – Performs the actual calculation of the function values and derivatives.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.amici_objective'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__setstate__(state: Dict)¶
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
apply_steadystate_guess(condition_ix: int, x_dct: Dict)¶ Use the stored steadystate as well as the respective sensitivity ( if available) and parameter value to approximate the steadystate at the current parameters using a zeroth or first order taylor approximation: x_ss(x’) = x_ss(x) [+ dx_ss/dx(x)*(x’-x)]
-
check_grad(x: numpy.ndarray, x_indices: List[int] = None, eps: float = 1e-05, verbosity: int = 1, mode: str = 'mode_fun') → pandas.core.frame.DataFrame¶ Compare gradient evaluation: Firstly approximate via finite differences, and secondly use the objective gradient.
Parameters: - x – The parameters for which to evaluate the gradient.
- x_indices – List of index values for which to compute gradients. Default: all.
- eps – Finite differences step size. Default: 1e-5.
- verbosity – Level of verbosity for function output. * 0: no output, * 1: summary for all parameters, * 2: summary for individual parameters. Default: 1.
- mode – Residual (MODE_RES) or objective function value (MODE_FUN, default) computation mode.
Returns: gradient, finite difference approximations and error estimates.
Return type: result
-
check_sensi_orders(sensi_orders, mode) → None¶ Check if the objective is able to compute the requested sensitivities. If not, throw an exception.
Raises: - ValueError if the objective function cannot be called as
requested.
-
get_bound_fun()¶ Generate a fun function that calls _call_amici with MODE_FUN. Defining a non-class function that references self as a local variable will bind the function to a copy of the current self object and will accordingly not take future changes to self into account.
-
get_bound_res()¶ Generate a res function that calls _call_amici with MODE_RES. Defining a non-class function that references self as a local variable will bind the function to a copy of the current self object and will accordingly not take future changes to self into account.
-
get_fval(x: numpy.ndarray) → float¶ Get the function value at x.
-
get_grad(x: numpy.ndarray) → numpy.ndarray¶ Get the gradient at x.
-
get_hess(x: numpy.ndarray) → numpy.ndarray¶ Get the Hessian at x.
-
get_res(x: numpy.ndarray) → numpy.ndarray¶ Get the residuals at x.
-
get_sres(x: numpy.ndarray) → numpy.ndarray¶ Get the residual sensitivities at x.
-
has_fun¶
-
has_grad¶
-
has_hess¶
-
has_hessp¶
-
has_res¶
-
has_sres¶
-
initialize()¶ Initialize the objective function. This function is used at the beginning of an analysis, e.g. optimization, and can e.g. reset the objective memory. By default does nothing.
-
static
output_to_dict()¶ Convert output tuple to dict.
-
static
output_to_tuple()¶ Return values as requested by the caller, since usually only a subset is demanded. One output is returned as-is, more than one output are returned as a tuple in order (fval, grad, hess).
-
par_arr_to_dct(x: Sequence[float]) → Dict[str, float]¶ Create dict from parameter vector.
-
rebind_fun()¶ Replace the current fun function with one that is bound to the current instance
-
rebind_res()¶ Replace the current res function with one that is bound to the current instance
-
reset_steadystate_guesses()¶ Resets all steadystate guess data
-
store_steadystate_guess(condition_ix: int, x_dct: Dict, rdata: amici.ReturnData)¶ Store condition parameter, steadystate and steadystate sensitivity in steadystate_guesses if steadystate guesses are enabled for this condition
-
update_from_problem(dim_full: int, x_free_indices: List[int], x_fixed_indices: List[int], x_fixed_vals: List[int])¶ Handle fixed parameters. Later, the objective will be given parameter vectors x of dimension dim, which have to be filled up with fixed parameter values to form a vector of dimension dim_full >= dim. This vector is then used to compute function value and derivatives. The derivatives must later be reduced again to dimension dim.
This is so as to make the fixing of parameters transparent to the caller.
The methods preprocess, postprocess are overwritten for the above functionality, respectively.
Parameters: - dim_full – Dimension of the full vector including fixed parameters.
- x_free_indices – Vector containing the indices (zero-based) of free parameters (complimentary to x_fixed_indices).
- x_fixed_indices – Vector containing the indices (zero-based) of parameter components that are not to be optimized.
- x_fixed_vals – Vector of the same length as x_fixed_indices, containing the values of the fixed parameters.
-
-
class
pypesto.objective.CsvHistory(file: str, x_names: Iterable[str] = None, options: Dict = None)¶ Bases:
pypesto.objective.history.HistoryStores a representation of the history in a CSV file.
Parameters: - file – CSV file name.
- x_names – Parameter names.
- options – History options.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': 'Stores a representation of the history in a CSV file.\n\n Parameters\n ----------\n file:\n CSV file name.\n x_names:\n Parameter names.\n options:\n History options.\n ', '__init__': <function CsvHistory.__init__>, 'update': <function CsvHistory.update>, 'finalize': <function CsvHistory.finalize>, '_update_trace': <function CsvHistory._update_trace>, '_init_trace': <function CsvHistory._init_trace>, '_save_trace': <function CsvHistory._save_trace>, 'get_fval_trace': <function CsvHistory.get_fval_trace>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(file: str, x_names: Iterable[str] = None, options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → pandas.core.series.Series¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int, ...], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history after a function evaluation.
Parameters: - x – The parameter vector.
- sensi_orders – The sensitivity orders computed.
- mode – The objective function mode computed (function value or residuals).
- result – The objective function values for parameters x, sensitivities sensi_orders and mode mode.
-
class
pypesto.objective.Hdf5History(id: str, file: str, options: Dict = None)¶ Bases:
pypesto.objective.history.HistoryStores a representation of the history in an HDF5 file.
Parameters: - id – Id of the history
- file – HDF5 file name.
- options – History options.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': 'Stores a representation of the history in an HDF5 file.\n\n Parameters\n ----------\n id:\n Id of the history\n file:\n HDF5 file name.\n options:\n History options.\n ', '__init__': <function Hdf5History.__init__>, 'update': <function Hdf5History.update>, 'finalize': <function Hdf5History.finalize>, 'get_x_trace': <function Hdf5History.get_x_trace>, 'get_fval_trace': <function Hdf5History.get_fval_trace>, 'get_grad_trace': <function Hdf5History.get_grad_trace>, 'get_hess_trace': <function Hdf5History.get_hess_trace>, 'get_res_trace': <function Hdf5History.get_res_trace>, 'get_sres_trace': <function Hdf5History.get_sres_trace>, 'get_chi2_trace': <function Hdf5History.get_chi2_trace>, 'get_schi2_trace': <function Hdf5History.get_schi2_trace>, 'get_time_trace': <function Hdf5History.get_time_trace>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(id: str, file: str, options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → Sequence[float]¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int, ...], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history after a function evaluation.
Parameters: - x – The parameter vector.
- sensi_orders – The sensitivity orders computed.
- mode – The objective function mode computed (function value or residuals).
- result – The objective function values for parameters x, sensitivities sensi_orders and mode mode.
-
class
pypesto.objective.History(options: Dict = None)¶ Bases:
pypesto.objective.history.HistoryBaseTracks numbers of function evaluations only, no trace.
Parameters: options – History options. -
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': 'Tracks numbers of function evaluations only, no trace.\n\n Parameters\n ----------\n options:\n History options.\n ', '__init__': <function History.__init__>, 'update': <function History.update>, 'finalize': <function History.finalize>, '_update_counts': <function History._update_counts>, 'n_fval': <property object>, 'n_grad': <property object>, 'n_hess': <property object>, 'n_res': <property object>, 'n_sres': <property object>, 'start_time': <property object>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → Sequence[float]¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int, ...], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history after a function evaluation.
Parameters: - x – The parameter vector.
- sensi_orders – The sensitivity orders computed.
- mode – The objective function mode computed (function value or residuals).
- result – The objective function values for parameters x, sensitivities sensi_orders and mode mode.
-
-
class
pypesto.objective.HistoryBase¶ Bases:
abc.ABCAbstract base class for history objects.
Can be used as a dummy history, but does not implement any history functionality.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': 'Abstract base class for history objects.\n\n Can be used as a dummy history, but does not implement any history\n functionality.\n ', 'update': <function HistoryBase.update>, 'finalize': <function HistoryBase.finalize>, 'n_fval': <property object>, 'n_grad': <property object>, 'n_hess': <property object>, 'n_res': <property object>, 'n_sres': <property object>, 'start_time': <property object>, 'get_x_trace': <function HistoryBase.get_x_trace>, 'get_fval_trace': <function HistoryBase.get_fval_trace>, 'get_grad_trace': <function HistoryBase.get_grad_trace>, 'get_hess_trace': <function HistoryBase.get_hess_trace>, 'get_res_trace': <function HistoryBase.get_res_trace>, 'get_sres_trace': <function HistoryBase.get_sres_trace>, 'get_chi2_trace': <function HistoryBase.get_chi2_trace>, 'get_schi2_trace': <function HistoryBase.get_schi2_trace>, 'get_time_trace': <function HistoryBase.get_time_trace>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → Sequence[float]¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int, ...], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history after a function evaluation.
Parameters: - x – The parameter vector.
- sensi_orders – The sensitivity orders computed.
- mode – The objective function mode computed (function value or residuals).
- result – The objective function values for parameters x, sensitivities sensi_orders and mode mode.
-
-
class
pypesto.objective.HistoryOptions(trace_record: bool = False, trace_record_grad: bool = True, trace_record_hess: bool = True, trace_record_res: bool = True, trace_record_sres: bool = True, trace_record_chi2: bool = True, trace_record_schi2: bool = True, trace_save_iter: int = 10, storage_file: str = None)¶ Bases:
dictOptions for the objective that are used in optimization, profiles and sampling.
In addition implements a factory pattern to generate history objects.
Parameters: - trace_record – Flag indicating whether to record the trace of function calls. The trace_record_* flags only become effective if trace_record is True. Default: False.
- trace_record_grad – Flag indicating whether to record the gradient in the trace. Default: True.
- trace_record_hess – Flag indicating whether to record the Hessian in the trace. Default: False.
- trace_record_res – Flag indicating whether to record the residual in the trace. Default: False.
- trace_record_sres – Flag indicating whether to record the residual sensitivities in the trace. Default: False.
- trace_record_chi2 – Flag indicating whether to record the chi2 in the trace. Default: True.
- trace_record_schi2 – Flag indicating whether to record the chi2 sensitivities in the trace. Default: True.
- trace_save_iter – After how many iterations to store the trace.
- storage_file – File to save the history to. Can be any of None, a “{filename}.csv”, or a “{filename}.hdf5” file. Depending on the values, the create_history method creates the appropriate object. Occurrences of “{id}” in the file name are replaced by the id upon creation of a history, if applicable.
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': '\n Options for the objective that are used in optimization, profiles\n and sampling.\n\n In addition implements a factory pattern to generate history objects.\n\n Parameters\n ----------\n trace_record:\n Flag indicating whether to record the trace of function calls.\n The trace_record_* flags only become effective if\n trace_record is True.\n Default: False.\n trace_record_grad:\n Flag indicating whether to record the gradient in the trace.\n Default: True.\n trace_record_hess:\n Flag indicating whether to record the Hessian in the trace.\n Default: False.\n trace_record_res:\n Flag indicating whether to record the residual in\n the trace.\n Default: False.\n trace_record_sres:\n Flag indicating whether to record the residual sensitivities in\n the trace.\n Default: False.\n trace_record_chi2:\n Flag indicating whether to record the chi2 in the trace.\n Default: True.\n trace_record_schi2:\n Flag indicating whether to record the chi2 sensitivities in the\n trace.\n Default: True.\n trace_save_iter:\n After how many iterations to store the trace.\n storage_file:\n File to save the history to. Can be any of None, a\n "{filename}.csv", or a "{filename}.hdf5" file. Depending on the values,\n the `create_history` method creates the appropriate object.\n Occurrences of "{id}" in the file name are replaced by the `id`\n upon creation of a history, if applicable.\n ', '__init__': <function HistoryOptions.__init__>, '__getattr__': <function HistoryOptions.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, 'assert_instance': <staticmethod object>, 'create_history': <function HistoryOptions.create_history>, '__dict__': <attribute '__dict__' of 'HistoryOptions' objects>, '__weakref__': <attribute '__weakref__' of 'HistoryOptions' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(trace_record: bool = False, trace_record_grad: bool = True, trace_record_hess: bool = True, trace_record_res: bool = True, trace_record_sres: bool = True, trace_record_chi2: bool = True, trace_record_schi2: bool = True, trace_save_iter: int = 10, storage_file: str = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
assert_instance(maybe_options: Union[HistoryOptions, Dict]) → pypesto.objective.history.HistoryOptions¶ Returns a valid options object.
Parameters: maybe_options (HistoryOptions or dict) –
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
create_history(id: str, x_names: Iterable[str]) → pypesto.objective.history.History¶ Factory method creating a
Historyobject.Parameters: - id – Identifier for the history.
- x_names – Parameter names.
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
class
pypesto.objective.MemoryHistory(options: Dict = None)¶ Bases:
pypesto.objective.history.HistoryTracks numbers of function evaluations and keeps an in-memory trace of function evaluations.
Parameters: options – History options. -
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': 'Tracks numbers of function evaluations and keeps an in-memory\n trace of function evaluations.\n\n Parameters\n ----------\n options:\n History options.\n ', '__init__': <function MemoryHistory.__init__>, 'update': <function MemoryHistory.update>, '_update_trace': <function MemoryHistory._update_trace>, 'get_x_trace': <function MemoryHistory.get_x_trace>, 'get_fval_trace': <function MemoryHistory.get_fval_trace>, 'get_grad_trace': <function MemoryHistory.get_grad_trace>, 'get_hess_trace': <function MemoryHistory.get_hess_trace>, 'get_res_trace': <function MemoryHistory.get_res_trace>, 'get_sres_trace': <function MemoryHistory.get_sres_trace>, 'get_chi2_trace': <function MemoryHistory.get_chi2_trace>, 'get_schi2_trace': <function MemoryHistory.get_schi2_trace>, 'get_time_trace': <function MemoryHistory.get_time_trace>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → Sequence[float]¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int, ...], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history after a function evaluation.
Parameters: - x – The parameter vector.
- sensi_orders – The sensitivity orders computed.
- mode – The objective function mode computed (function value or residuals).
- result – The objective function values for parameters x, sensitivities sensi_orders and mode mode.
-
-
class
pypesto.objective.Objective(fun: Callable = None, grad: Union[Callable, bool] = None, hess: Callable = None, hessp: Callable = None, res: Callable = None, sres: Union[Callable, bool] = None, fun_accept_sensi_orders: bool = False, res_accept_sensi_orders: bool = False, x_names: List[str] = None)¶ Bases:
objectThe objective class is a simple wrapper around the objective function, giving a standardized way of calling. Apart from that, it manages several things including fixing of parameters and history.
The objective function is assumed to be in the format of a cost function, log-likelihood function, or log-posterior function. These functions are subject to minimization. For profiling and sampling, the sign is internally flipped, all returned and stored values are however given as returned by this objective function. If maximization is to be performed, the sign should be flipped before creating the objective function.
Parameters: - fun –
The objective function to be minimized. If it only computes the objective function value, it should be of the form
fun(x) -> floatwhere x is an 1-D array with shape (n,), and n is the parameter space dimension.
- grad –
Method for computing the gradient vector. If it is a callable, it should be of the form
grad(x) -> array_like, shape (n,).If its value is True, then fun should return the gradient as a second output.
- hess –
Method for computing the Hessian matrix. If it is a callable, it should be of the form
hess(x) -> array, shape (n,n).If its value is True, then fun should return the gradient as a second, and the Hessian as a third output, and grad should be True as well.
- hessp –
Method for computing the Hessian vector product, i.e.
hessp(x, v) -> array_like, shape (n,)computes the product H*v of the Hessian of fun at x with v.
- res –
Method for computing residuals, i.e.
res(x) -> array_like, shape(m,). - sres –
Method for computing residual sensitivities. If its is a callable, it should be of the form
sres(x) -> array, shape (m,n).If its value is True, then res should return the residual sensitivities as a second output.
- fun_accept_sensi_orders – Flag indicating whether fun takes sensi_orders as an argument. Default: False.
- res_accept_sensi_orders – Flag indicating whether res takes sensi_orders as an argument. Default: False
- x_names – Parameter names. None if no names provided, otherwise a list of str, length dim_full (as in the Problem class). Can be read by the problem.
-
history¶ For storing the call history. Initialized by the methods, e.g. the optimizer, in initialize_history().
-
pre_post_processor¶ Preprocess input values to and postprocess output values from __call__. Configured in update_from_problem().
Notes
If fun_accept_sensi_orders resp. res_accept_sensi_orders is True, fun resp. res can also return dictionaries instead of tuples. In that case, they are expected to follow the naming conventions in
constants.py. This is of interest, because when __call__ is called with return_dict = True, the full dictionary is returned, which can contain e.g. also simulation data or debugging information.-
__call__()¶ Method to obtain arbitrary sensitivities. This is the central method which is always called, also by the get_* methods.
There are different ways in which an optimizer calls the objective function, and in how the objective function provides information (e.g. derivatives via separate functions or along with the function values). The different calling modes increase efficiency in space and time and make the objective flexible.
Parameters: - x – The parameters for which to evaluate the objective function.
- sensi_orders – Specifies which sensitivities to compute, e.g. (0,1) -> fval, grad.
- mode – Whether to compute function values or residuals.
- return_dict – If False (default), the result is a Tuple of the requested values in the requested order. Tuples of length one are flattened. If True, instead a dict is returned which can carry further information.
Returns: By default, this is a tuple of the requested function values and derivatives in the requested order (if only 1 value, the tuple is flattened). If return_dict, then instead a dict is returned with function values and derivatives indicated by ids.
Return type: result
-
__class__¶ alias of
builtins.type
-
__deepcopy__(memodict=None) → pypesto.objective.objective.Objective¶
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.objective', '__doc__': '\n The objective class is a simple wrapper around the objective function,\n giving a standardized way of calling. Apart from that, it manages several\n things including fixing of parameters and history.\n\n The objective function is assumed to be in the format of a cost function,\n log-likelihood function, or log-posterior function. These functions are\n subject to minimization. For profiling and sampling, the sign is internally\n flipped, all returned and stored values are however given as returned\n by this objective function. If maximization is to be performed, the sign\n should be flipped before creating the objective function.\n\n Parameters\n ----------\n\n fun:\n The objective function to be minimized. If it only computes the\n objective function value, it should be of the form\n\n ``fun(x) -> float``\n\n where x is an 1-D array with shape (n,), and n is the parameter space\n dimension.\n\n grad:\n Method for computing the gradient vector. If it is a callable,\n it should be of the form\n\n ``grad(x) -> array_like, shape (n,).``\n\n If its value is True, then fun should return the gradient as a second\n output.\n\n hess:\n Method for computing the Hessian matrix. If it is a callable,\n it should be of the form\n\n ``hess(x) -> array, shape (n,n).``\n\n If its value is True, then fun should return the gradient as a\n second, and the Hessian as a third output, and grad should be True as\n well.\n\n hessp:\n Method for computing the Hessian vector product, i.e.\n\n ``hessp(x, v) -> array_like, shape (n,)``\n\n computes the product H*v of the Hessian of fun at x with v.\n\n res:\n Method for computing residuals, i.e.\n\n ``res(x) -> array_like, shape(m,).``\n\n sres:\n Method for computing residual sensitivities. If its is a callable,\n it should be of the form\n\n ``sres(x) -> array, shape (m,n).``\n\n If its value is True, then res should return the residual\n sensitivities as a second output.\n\n fun_accept_sensi_orders:\n Flag indicating whether fun takes sensi_orders as an argument.\n Default: False.\n\n res_accept_sensi_orders:\n Flag indicating whether res takes sensi_orders as an argument.\n Default: False\n\n x_names:\n Parameter names. None if no names provided, otherwise a list of str,\n length dim_full (as in the Problem class). Can be read by the\n problem.\n\n Attributes\n ----------\n\n history:\n For storing the call history. Initialized by the methods, e.g. the\n optimizer, in `initialize_history()`.\n\n pre_post_processor:\n Preprocess input values to and postprocess output values from\n __call__. Configured in `update_from_problem()`.\n\n Notes\n -----\n\n If fun_accept_sensi_orders resp. res_accept_sensi_orders is True,\n fun resp. res can also return dictionaries instead of tuples.\n In that case, they are expected to follow the naming conventions\n in ``constants.py``. This is of interest, because when __call__ is\n called with return_dict = True, the full dictionary is returned, which\n can contain e.g. also simulation data or debugging information.\n ', '__init__': <function Objective.__init__>, '__deepcopy__': <function Objective.__deepcopy__>, 'initialize': <function Objective.initialize>, 'has_fun': <property object>, 'has_grad': <property object>, 'has_hess': <property object>, 'has_hessp': <property object>, 'has_res': <property object>, 'has_sres': <property object>, 'check_sensi_orders': <function Objective.check_sensi_orders>, '__call__': <function Objective.__call__>, '_call_unprocessed': <function Objective._call_unprocessed>, '_call_mode_fun': <function Objective._call_mode_fun>, '_call_mode_res': <function Objective._call_mode_res>, 'output_to_dict': <staticmethod object>, 'output_to_tuple': <staticmethod object>, 'get_fval': <function Objective.get_fval>, 'get_grad': <function Objective.get_grad>, 'get_hess': <function Objective.get_hess>, 'get_res': <function Objective.get_res>, 'get_sres': <function Objective.get_sres>, 'update_from_problem': <function Objective.update_from_problem>, 'check_grad': <function Objective.check_grad>, '__dict__': <attribute '__dict__' of 'Objective' objects>, '__weakref__': <attribute '__weakref__' of 'Objective' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(fun: Callable = None, grad: Union[Callable, bool] = None, hess: Callable = None, hessp: Callable = None, res: Callable = None, sres: Union[Callable, bool] = None, fun_accept_sensi_orders: bool = False, res_accept_sensi_orders: bool = False, x_names: List[str] = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.objective'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
check_grad(x: numpy.ndarray, x_indices: List[int] = None, eps: float = 1e-05, verbosity: int = 1, mode: str = 'mode_fun') → pandas.core.frame.DataFrame¶ Compare gradient evaluation: Firstly approximate via finite differences, and secondly use the objective gradient.
Parameters: - x – The parameters for which to evaluate the gradient.
- x_indices – List of index values for which to compute gradients. Default: all.
- eps – Finite differences step size. Default: 1e-5.
- verbosity – Level of verbosity for function output. * 0: no output, * 1: summary for all parameters, * 2: summary for individual parameters. Default: 1.
- mode – Residual (MODE_RES) or objective function value (MODE_FUN, default) computation mode.
Returns: gradient, finite difference approximations and error estimates.
Return type: result
-
check_sensi_orders(sensi_orders, mode) → None¶ Check if the objective is able to compute the requested sensitivities. If not, throw an exception.
Raises: - ValueError if the objective function cannot be called as
requested.
-
get_fval(x: numpy.ndarray) → float¶ Get the function value at x.
-
get_grad(x: numpy.ndarray) → numpy.ndarray¶ Get the gradient at x.
-
get_hess(x: numpy.ndarray) → numpy.ndarray¶ Get the Hessian at x.
-
get_res(x: numpy.ndarray) → numpy.ndarray¶ Get the residuals at x.
-
get_sres(x: numpy.ndarray) → numpy.ndarray¶ Get the residual sensitivities at x.
-
has_fun¶
-
has_grad¶
-
has_hess¶
-
has_hessp¶
-
has_res¶
-
has_sres¶
-
initialize()¶ Initialize the objective function. This function is used at the beginning of an analysis, e.g. optimization, and can e.g. reset the objective memory. By default does nothing.
-
static
output_to_dict()¶ Convert output tuple to dict.
-
static
output_to_tuple()¶ Return values as requested by the caller, since usually only a subset is demanded. One output is returned as-is, more than one output are returned as a tuple in order (fval, grad, hess).
-
update_from_problem(dim_full: int, x_free_indices: List[int], x_fixed_indices: List[int], x_fixed_vals: List[int])¶ Handle fixed parameters. Later, the objective will be given parameter vectors x of dimension dim, which have to be filled up with fixed parameter values to form a vector of dimension dim_full >= dim. This vector is then used to compute function value and derivatives. The derivatives must later be reduced again to dimension dim.
This is so as to make the fixing of parameters transparent to the caller.
The methods preprocess, postprocess are overwritten for the above functionality, respectively.
Parameters: - dim_full – Dimension of the full vector including fixed parameters.
- x_free_indices – Vector containing the indices (zero-based) of free parameters (complimentary to x_fixed_indices).
- x_fixed_indices – Vector containing the indices (zero-based) of parameter components that are not to be optimized.
- x_fixed_vals – Vector of the same length as x_fixed_indices, containing the values of the fixed parameters.
- fun –
-
class
pypesto.objective.OptimizerHistory(history: pypesto.objective.history.History, x0: numpy.ndarray)¶ Bases:
pypesto.objective.history.HistoryBaseObjective call history. Also handles saving of intermediate results.
-
fval0, fval_min Initial and best function value found.
-
x0, x_min Initial and best parameters found.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.history', '__doc__': '\n Objective call history. Also handles saving of intermediate results.\n\n Attributes\n ----------\n fval0, fval_min:\n Initial and best function value found.\n x0, x_min:\n Initial and best parameters found.\n ', '__init__': <function OptimizerHistory.__init__>, 'update': <function OptimizerHistory.update>, 'finalize': <function OptimizerHistory.finalize>, '_update_vals': <function OptimizerHistory._update_vals>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(history: pypesto.objective.history.History, x0: numpy.ndarray) → None¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.history'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
finalize()¶ Finalize history. Called after a run.
-
get_chi2_trace() → Sequence[numpy.ndarray]¶ Chi2 value trace.
-
get_fval_trace() → Sequence[float]¶ Function value trace.
-
get_grad_trace() → Sequence[numpy.ndarray]¶ Gradient trace.
-
get_hess_trace() → Sequence[numpy.ndarray]¶ Hessian trace.
-
get_res_trace() → Sequence[numpy.ndarray]¶ Residual trace.
-
get_schi2_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Chi2 value sensitivity trace.
-
get_sres_trace() → Sequence[numpy.ndarray]¶ Residual sensitivity trace.
-
get_time_trace(t: Optional[int] = None) → Sequence[numpy.ndarray]¶ Execution time trace.
-
get_x_trace() → Sequence[numpy.ndarray]¶ Parameter trace.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residual evaluations.
-
n_sres¶ Number or residual sensitivity evaluations.
-
start_time¶ Start time.
-
update(x: numpy.ndarray, sensi_orders: Tuple[int], mode: str, result: Dict[str, Union[float, numpy.ndarray]]) → None¶ Update history and best found value.
-
-
pypesto.objective.res_to_chi2(res: numpy.ndarray)¶ We assume that the residuals res are related to an objective function value fval = chi2 via:
fval = 0.5 * sum(res**2)
which is the ‘Linear’ formulation in scipy.
-
pypesto.objective.sres_to_schi2(res: numpy.ndarray, sres: numpy.ndarray)¶ In line with the assumptions in res_to_chi2.
Problem¶
A problem contains the objective as well as all information like prior describing the problem to be solved.
-
class
pypesto.problem.Iterable¶ Bases:
collections.abc.Iterable,typing.Generic-
__abstractmethods__= frozenset({'__iter__'})¶
-
__args__= None¶
-
__class__¶ alias of
GenericMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__extra__¶ alias of
collections.abc.Iterable
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__()¶
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'typing'¶
-
__ne__¶ Return self!=value.
-
static
__new__(cls, *args, **kwds)¶ Create and return a new object. See help(type) for accurate signature.
-
__next_in_mro__¶ alias of
builtins.object
-
__orig_bases__= (typing.Generic[+T_co],)¶
-
__origin__= None¶
-
__parameters__= (+T_co,)¶
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__slots__= ()¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶
-
__tree_hash__= -9223366114679874455¶
-
-
class
pypesto.problem.List¶ Bases:
list,typing.MutableSequence-
__abstractmethods__= frozenset()¶
-
__add__¶ Return self+value.
-
__args__= None¶
-
__class__¶ alias of
GenericMeta
-
__contains__¶ Return key in self.
-
__delattr__¶ Implement delattr(self, name).
-
__delitem__¶ Delete self[key].
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__extra__¶ alias of
builtins.list
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__iadd__¶ Implement self+=value.
-
__imul__¶ Implement self*=value.
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'typing'¶
-
__mul__¶ Return self*value.
-
__ne__¶ Return self!=value.
-
static
__new__(cls, *args, **kwds)¶ Create and return a new object. See help(type) for accurate signature.
-
__next_in_mro__¶ alias of
builtins.object
-
__orig_bases__= (<class 'list'>, typing.MutableSequence[~T])¶
-
__origin__= None¶
-
__parameters__= (~T,)¶
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__reversed__()¶ L.__reversed__() – return a reverse iterator over the list
-
__rmul__¶ Return value*self.
-
__setattr__¶ Implement setattr(self, name, value).
-
__setitem__¶ Set self[key] to value.
-
__sizeof__()¶ L.__sizeof__() – size of L in memory, in bytes
-
__slots__= ()¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶
-
__tree_hash__= -9223366114679871464¶
-
append(object) → None -- append object to end¶
-
clear() → None -- remove all items from L¶
-
copy() → list -- a shallow copy of L¶
-
count(value) → integer -- return number of occurrences of value¶
-
extend(iterable) → None -- extend list by appending elements from the iterable¶
-
index(value[, start[, stop]]) → integer -- return first index of value.¶ Raises ValueError if the value is not present.
-
insert()¶ L.insert(index, object) – insert object before index
-
pop([index]) → item -- remove and return item at index (default last).¶ Raises IndexError if list is empty or index is out of range.
-
remove(value) → None -- remove first occurrence of value.¶ Raises ValueError if the value is not present.
-
reverse()¶ L.reverse() – reverse IN PLACE
-
sort(key=None, reverse=False) → None -- stable sort *IN PLACE*¶
-
-
class
pypesto.problem.Objective(fun: Callable = None, grad: Union[Callable, bool] = None, hess: Callable = None, hessp: Callable = None, res: Callable = None, sres: Union[Callable, bool] = None, fun_accept_sensi_orders: bool = False, res_accept_sensi_orders: bool = False, x_names: List[str] = None)¶ Bases:
objectThe objective class is a simple wrapper around the objective function, giving a standardized way of calling. Apart from that, it manages several things including fixing of parameters and history.
The objective function is assumed to be in the format of a cost function, log-likelihood function, or log-posterior function. These functions are subject to minimization. For profiling and sampling, the sign is internally flipped, all returned and stored values are however given as returned by this objective function. If maximization is to be performed, the sign should be flipped before creating the objective function.
Parameters: - fun –
The objective function to be minimized. If it only computes the objective function value, it should be of the form
fun(x) -> floatwhere x is an 1-D array with shape (n,), and n is the parameter space dimension.
- grad –
Method for computing the gradient vector. If it is a callable, it should be of the form
grad(x) -> array_like, shape (n,).If its value is True, then fun should return the gradient as a second output.
- hess –
Method for computing the Hessian matrix. If it is a callable, it should be of the form
hess(x) -> array, shape (n,n).If its value is True, then fun should return the gradient as a second, and the Hessian as a third output, and grad should be True as well.
- hessp –
Method for computing the Hessian vector product, i.e.
hessp(x, v) -> array_like, shape (n,)computes the product H*v of the Hessian of fun at x with v.
- res –
Method for computing residuals, i.e.
res(x) -> array_like, shape(m,). - sres –
Method for computing residual sensitivities. If its is a callable, it should be of the form
sres(x) -> array, shape (m,n).If its value is True, then res should return the residual sensitivities as a second output.
- fun_accept_sensi_orders – Flag indicating whether fun takes sensi_orders as an argument. Default: False.
- res_accept_sensi_orders – Flag indicating whether res takes sensi_orders as an argument. Default: False
- x_names – Parameter names. None if no names provided, otherwise a list of str, length dim_full (as in the Problem class). Can be read by the problem.
-
history¶ For storing the call history. Initialized by the methods, e.g. the optimizer, in initialize_history().
-
pre_post_processor¶ Preprocess input values to and postprocess output values from __call__. Configured in update_from_problem().
Notes
If fun_accept_sensi_orders resp. res_accept_sensi_orders is True, fun resp. res can also return dictionaries instead of tuples. In that case, they are expected to follow the naming conventions in
constants.py. This is of interest, because when __call__ is called with return_dict = True, the full dictionary is returned, which can contain e.g. also simulation data or debugging information.-
__call__()¶ Method to obtain arbitrary sensitivities. This is the central method which is always called, also by the get_* methods.
There are different ways in which an optimizer calls the objective function, and in how the objective function provides information (e.g. derivatives via separate functions or along with the function values). The different calling modes increase efficiency in space and time and make the objective flexible.
Parameters: - x – The parameters for which to evaluate the objective function.
- sensi_orders – Specifies which sensitivities to compute, e.g. (0,1) -> fval, grad.
- mode – Whether to compute function values or residuals.
- return_dict – If False (default), the result is a Tuple of the requested values in the requested order. Tuples of length one are flattened. If True, instead a dict is returned which can carry further information.
Returns: By default, this is a tuple of the requested function values and derivatives in the requested order (if only 1 value, the tuple is flattened). If return_dict, then instead a dict is returned with function values and derivatives indicated by ids.
Return type: result
-
__class__¶ alias of
builtins.type
-
__deepcopy__(memodict=None) → pypesto.objective.objective.Objective¶
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.objective.objective', '__doc__': '\n The objective class is a simple wrapper around the objective function,\n giving a standardized way of calling. Apart from that, it manages several\n things including fixing of parameters and history.\n\n The objective function is assumed to be in the format of a cost function,\n log-likelihood function, or log-posterior function. These functions are\n subject to minimization. For profiling and sampling, the sign is internally\n flipped, all returned and stored values are however given as returned\n by this objective function. If maximization is to be performed, the sign\n should be flipped before creating the objective function.\n\n Parameters\n ----------\n\n fun:\n The objective function to be minimized. If it only computes the\n objective function value, it should be of the form\n\n ``fun(x) -> float``\n\n where x is an 1-D array with shape (n,), and n is the parameter space\n dimension.\n\n grad:\n Method for computing the gradient vector. If it is a callable,\n it should be of the form\n\n ``grad(x) -> array_like, shape (n,).``\n\n If its value is True, then fun should return the gradient as a second\n output.\n\n hess:\n Method for computing the Hessian matrix. If it is a callable,\n it should be of the form\n\n ``hess(x) -> array, shape (n,n).``\n\n If its value is True, then fun should return the gradient as a\n second, and the Hessian as a third output, and grad should be True as\n well.\n\n hessp:\n Method for computing the Hessian vector product, i.e.\n\n ``hessp(x, v) -> array_like, shape (n,)``\n\n computes the product H*v of the Hessian of fun at x with v.\n\n res:\n Method for computing residuals, i.e.\n\n ``res(x) -> array_like, shape(m,).``\n\n sres:\n Method for computing residual sensitivities. If its is a callable,\n it should be of the form\n\n ``sres(x) -> array, shape (m,n).``\n\n If its value is True, then res should return the residual\n sensitivities as a second output.\n\n fun_accept_sensi_orders:\n Flag indicating whether fun takes sensi_orders as an argument.\n Default: False.\n\n res_accept_sensi_orders:\n Flag indicating whether res takes sensi_orders as an argument.\n Default: False\n\n x_names:\n Parameter names. None if no names provided, otherwise a list of str,\n length dim_full (as in the Problem class). Can be read by the\n problem.\n\n Attributes\n ----------\n\n history:\n For storing the call history. Initialized by the methods, e.g. the\n optimizer, in `initialize_history()`.\n\n pre_post_processor:\n Preprocess input values to and postprocess output values from\n __call__. Configured in `update_from_problem()`.\n\n Notes\n -----\n\n If fun_accept_sensi_orders resp. res_accept_sensi_orders is True,\n fun resp. res can also return dictionaries instead of tuples.\n In that case, they are expected to follow the naming conventions\n in ``constants.py``. This is of interest, because when __call__ is\n called with return_dict = True, the full dictionary is returned, which\n can contain e.g. also simulation data or debugging information.\n ', '__init__': <function Objective.__init__>, '__deepcopy__': <function Objective.__deepcopy__>, 'initialize': <function Objective.initialize>, 'has_fun': <property object>, 'has_grad': <property object>, 'has_hess': <property object>, 'has_hessp': <property object>, 'has_res': <property object>, 'has_sres': <property object>, 'check_sensi_orders': <function Objective.check_sensi_orders>, '__call__': <function Objective.__call__>, '_call_unprocessed': <function Objective._call_unprocessed>, '_call_mode_fun': <function Objective._call_mode_fun>, '_call_mode_res': <function Objective._call_mode_res>, 'output_to_dict': <staticmethod object>, 'output_to_tuple': <staticmethod object>, 'get_fval': <function Objective.get_fval>, 'get_grad': <function Objective.get_grad>, 'get_hess': <function Objective.get_hess>, 'get_res': <function Objective.get_res>, 'get_sres': <function Objective.get_sres>, 'update_from_problem': <function Objective.update_from_problem>, 'check_grad': <function Objective.check_grad>, '__dict__': <attribute '__dict__' of 'Objective' objects>, '__weakref__': <attribute '__weakref__' of 'Objective' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(fun: Callable = None, grad: Union[Callable, bool] = None, hess: Callable = None, hessp: Callable = None, res: Callable = None, sres: Union[Callable, bool] = None, fun_accept_sensi_orders: bool = False, res_accept_sensi_orders: bool = False, x_names: List[str] = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.objective.objective'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
check_grad(x: numpy.ndarray, x_indices: List[int] = None, eps: float = 1e-05, verbosity: int = 1, mode: str = 'mode_fun') → pandas.core.frame.DataFrame¶ Compare gradient evaluation: Firstly approximate via finite differences, and secondly use the objective gradient.
Parameters: - x – The parameters for which to evaluate the gradient.
- x_indices – List of index values for which to compute gradients. Default: all.
- eps – Finite differences step size. Default: 1e-5.
- verbosity – Level of verbosity for function output. * 0: no output, * 1: summary for all parameters, * 2: summary for individual parameters. Default: 1.
- mode – Residual (MODE_RES) or objective function value (MODE_FUN, default) computation mode.
Returns: gradient, finite difference approximations and error estimates.
Return type: result
-
check_sensi_orders(sensi_orders, mode) → None¶ Check if the objective is able to compute the requested sensitivities. If not, throw an exception.
Raises: - ValueError if the objective function cannot be called as
requested.
-
get_fval(x: numpy.ndarray) → float¶ Get the function value at x.
-
get_grad(x: numpy.ndarray) → numpy.ndarray¶ Get the gradient at x.
-
get_hess(x: numpy.ndarray) → numpy.ndarray¶ Get the Hessian at x.
-
get_res(x: numpy.ndarray) → numpy.ndarray¶ Get the residuals at x.
-
get_sres(x: numpy.ndarray) → numpy.ndarray¶ Get the residual sensitivities at x.
-
has_fun¶
-
has_grad¶
-
has_hess¶
-
has_hessp¶
-
has_res¶
-
has_sres¶
-
initialize()¶ Initialize the objective function. This function is used at the beginning of an analysis, e.g. optimization, and can e.g. reset the objective memory. By default does nothing.
-
static
output_to_dict()¶ Convert output tuple to dict.
-
static
output_to_tuple()¶ Return values as requested by the caller, since usually only a subset is demanded. One output is returned as-is, more than one output are returned as a tuple in order (fval, grad, hess).
-
update_from_problem(dim_full: int, x_free_indices: List[int], x_fixed_indices: List[int], x_fixed_vals: List[int])¶ Handle fixed parameters. Later, the objective will be given parameter vectors x of dimension dim, which have to be filled up with fixed parameter values to form a vector of dimension dim_full >= dim. This vector is then used to compute function value and derivatives. The derivatives must later be reduced again to dimension dim.
This is so as to make the fixing of parameters transparent to the caller.
The methods preprocess, postprocess are overwritten for the above functionality, respectively.
Parameters: - dim_full – Dimension of the full vector including fixed parameters.
- x_free_indices – Vector containing the indices (zero-based) of free parameters (complimentary to x_fixed_indices).
- x_fixed_indices – Vector containing the indices (zero-based) of parameter components that are not to be optimized.
- x_fixed_vals – Vector of the same length as x_fixed_indices, containing the values of the fixed parameters.
- fun –
-
class
pypesto.problem.Problem(objective: pypesto.objective.objective.Objective, lb: Union[numpy.ndarray, List[float]], ub: Union[numpy.ndarray, List[float]], dim_full: Optional[int] = None, x_fixed_indices: Optional[Iterable[int]] = None, x_fixed_vals: Optional[Iterable[float]] = None, x_guesses: Optional[Iterable[float]] = None, x_names: Optional[Iterable[str]] = None)¶ Bases:
objectThe problem formulation. A problem specifies the objective function, boundaries and constraints, parameter guesses as well as the parameters which are to be optimized.
Parameters: - objective – The objective function for minimization. Note that a shallow copy is created.
- ub (lb,) – The lower and upper bounds. For unbounded directions set to inf.
- dim_full – The full dimension of the problem, including fixed parameters.
- x_fixed_indices – Vector containing the indices (zero-based) of parameter components that are not to be optimized.
- x_fixed_vals – Vector of the same length as x_fixed_indices, containing the values of the fixed parameters.
- x_guesses – Guesses for the parameter values, shape (g, dim), where g denotes the number of guesses. These are used as start points in the optimization.
- x_names – Parameter names that can be optionally used e.g. in visualizations. If objective.get_x_names() is not None, those values are used, else the values specified here are used if not None, otherwise the variable names are set to [‘x0’, … ‘x{dim_full}’]. The list must always be of length dim_full.
-
dim¶ The number of non-fixed parameters. Computed from the other values.
-
x_free_indices¶ Vector containing the indices (zero-based) of free parameters (complimentary to x_fixed_indices).
Type: array_like of int
Notes
On the fixing of parameter values:
The number of parameters dim_full the objective takes as input must be known, so it must be either lb a vector of that size, or dim_full specified as a parameter.
All vectors are mapped to the reduced space of dimension dim in __init__, regardless of whether they were in dimension dim or dim_full before. If the full representation is needed, the methods get_full_vector() and get_full_matrix() can be used.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.problem', '__doc__': "\n The problem formulation. A problem specifies the objective function,\n boundaries and constraints, parameter guesses as well as the parameters\n which are to be optimized.\n\n Parameters\n ----------\n objective:\n The objective function for minimization. Note that a shallow copy\n is created.\n lb, ub:\n The lower and upper bounds. For unbounded directions set to inf.\n dim_full:\n The full dimension of the problem, including fixed parameters.\n x_fixed_indices:\n Vector containing the indices (zero-based) of parameter components\n that are not to be optimized.\n x_fixed_vals:\n Vector of the same length as x_fixed_indices, containing the values\n of the fixed parameters.\n x_guesses:\n Guesses for the parameter values, shape (g, dim), where g denotes the\n number of guesses. These are used as start points in the optimization.\n x_names:\n Parameter names that can be optionally used e.g. in visualizations.\n If objective.get_x_names() is not None, those values are used,\n else the values specified here are used if not None, otherwise\n the variable names are set to ['x0', ... 'x{dim_full}']. The list\n must always be of length dim_full.\n\n Attributes\n ----------\n\n dim:\n The number of non-fixed parameters.\n Computed from the other values.\n x_free_indices: array_like of int\n Vector containing the indices (zero-based) of free parameters\n (complimentary to x_fixed_indices).\n\n Notes\n -----\n\n On the fixing of parameter values:\n\n The number of parameters dim_full the objective takes as input must\n be known, so it must be either lb a vector of that size, or dim_full\n specified as a parameter.\n\n All vectors are mapped to the reduced space of dimension dim in __init__,\n regardless of whether they were in dimension dim or dim_full before. If\n the full representation is needed, the methods get_full_vector() and\n get_full_matrix() can be used.\n ", '__init__': <function Problem.__init__>, 'normalize_input': <function Problem.normalize_input>, 'fix_parameters': <function Problem.fix_parameters>, 'unfix_parameters': <function Problem.unfix_parameters>, 'get_full_vector': <function Problem.get_full_vector>, 'get_full_matrix': <function Problem.get_full_matrix>, 'get_reduced_vector': <function Problem.get_reduced_vector>, 'get_reduced_matrix': <function Problem.get_reduced_matrix>, 'print_parameter_summary': <function Problem.print_parameter_summary>, '__dict__': <attribute '__dict__' of 'Problem' objects>, '__weakref__': <attribute '__weakref__' of 'Problem' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(objective: pypesto.objective.objective.Objective, lb: Union[numpy.ndarray, List[float]], ub: Union[numpy.ndarray, List[float]], dim_full: Optional[int] = None, x_fixed_indices: Optional[Iterable[int]] = None, x_fixed_vals: Optional[Iterable[float]] = None, x_guesses: Optional[Iterable[float]] = None, x_names: Optional[Iterable[str]] = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.problem'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
fix_parameters(parameter_indices: Union[Iterable[int], int], parameter_vals: Union[Iterable[float], float]) → None¶ Fix specified parameters to specified values
-
get_full_matrix(x: Optional[numpy.ndarray]) → Optional[numpy.ndarray]¶ Map matrix from dim to dim_full. Usually used for hessian.
Parameters: x (array_like, shape=(dim, dim)) – The matrix in dimension dim.
-
get_full_vector(x: Optional[numpy.ndarray], x_fixed_vals: Iterable[float] = None) → Optional[numpy.ndarray]¶ Map vector from dim to dim_full. Usually used for x, grad.
Parameters: - x (array_like, shape=(dim,)) – The vector in dimension dim.
- x_fixed_vals (array_like, ndim=1, optional) – The values to be used for the fixed indices. If None, then nans are inserted. Usually, None will be used for grad and problem.x_fixed_vals for x.
-
get_reduced_matrix(x_full: Optional[numpy.ndarray]) → Optional[numpy.ndarray]¶ Map matrix from dim_full to dim, i.e. delete fixed indices.
Parameters: x_full (array_like, ndim=2) – The matrix in dimension dim_full.
-
get_reduced_vector(x_full: Optional[numpy.ndarray]) → Optional[numpy.ndarray]¶ Map vector from dim_full to dim, i.e. delete fixed indices.
Parameters: x_full (array_like, ndim=1) – The vector in dimension dim_full.
-
normalize_input(check_x_guesses: bool = True) → None¶ Reduce all vectors to dimension dim and have the objective accept vectors of dimension dim.
-
print_parameter_summary() → None¶ Prints a summary of what parameters are being optimized and what parameter boundaries are
-
unfix_parameters(parameter_indices: Union[Iterable[int], int]) → None¶ Free specified parameters
PEtab¶
pyPESTO support for the PEtab data format.
-
class
pypesto.petab.PetabImporter(petab_problem: petab.Problem, output_folder: str = None, model_name: str = None)¶ Bases:
pypesto.objective.amici_objective.AmiciObjectBuilder-
MODEL_BASE_DIR= 'amici_models'¶
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.petab.importer', 'MODEL_BASE_DIR': 'amici_models', '__init__': <function PetabImporter.__init__>, 'from_yaml': <staticmethod object>, 'create_model': <function PetabImporter.create_model>, '_create_model': <function PetabImporter._create_model>, '_must_compile': <function PetabImporter._must_compile>, 'compile_model': <function PetabImporter.compile_model>, 'create_solver': <function PetabImporter.create_solver>, 'create_edatas': <function PetabImporter.create_edatas>, 'create_objective': <function PetabImporter.create_objective>, 'create_problem': <function PetabImporter.create_problem>, 'rdatas_to_measurement_df': <function PetabImporter.rdatas_to_measurement_df>, 'rdatas_to_simulation_df': <function PetabImporter.rdatas_to_simulation_df>, '__doc__': None, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(petab_problem: petab.Problem, output_folder: str = None, model_name: str = None)¶ - petab_problem:
- Managing access to the model and data.
- output_folder:
- Folder to contain the amici model. Defaults to ‘./amici_models/{model_name}’.
- model_name:
- Name of the model, which will in particular be the name of the compiled model python module.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.petab.importer'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
compile_model(**kwargs)¶ Compile the model. If the output folder exists already, it is first deleted.
Parameters: kwargs (Extra arguments passed to amici.SbmlImporter.sbml2amici.) –
-
create_edatas(model: amici.Model = None, simulation_conditions=None) → List[amici.ExpData]¶ Create list of amici.ExpData objects.
-
create_model(force_compile: bool = False, **kwargs) → amici.Model¶ Import amici model. If necessary or force_compile is True, compile first.
Parameters: - force_compile –
If False, the model is compiled only if the output folder does not exist yet. If True, the output folder is deleted and the model (re-)compiled in either case.
Warning
If force_compile, then an existing folder of that name will be deleted.
- kwargs (Extra arguments passed to amici.SbmlImporter.sbml2amici) –
- force_compile –
-
create_objective(model: amici.Model = None, solver: amici.Solver = None, edatas: Sequence[amici.ExpData] = None, force_compile: bool = False, **kwargs) → pypesto.objective.amici_objective.AmiciObjective¶ Create a
pypesto.AmiciObjective.Parameters: - model – The AMICI model.
- solver – The AMICI solver.
- edatas – The experimental data in AMICI format.
- force_compile – Whether to force-compile the model if not passed.
- **kwargs – Additional arguments passed on to the objective.
Returns: A
pypesto.AmiciObjectivefor the model and the data.Return type: objective
-
create_problem(objective: pypesto.objective.amici_objective.AmiciObjective = None, **kwargs) → pypesto.problem.Problem¶ Create a
pypesto.Problem.Parameters: - objective – Objective as created by create_objective.
- **kwargs – Additional key word arguments passed on to the objective, if not provided.
Returns: A
pypesto.Problemfor the objective.Return type: problem
-
create_solver(model: amici.Model = None) → amici.Solver¶ Return model solver.
-
static
from_yaml(yaml_config: Union[dict, str], output_folder: str = None, model_name: str = None) → pypesto.petab.importer.PetabImporter¶ Simplified constructor using a petab yaml file.
-
rdatas_to_measurement_df(rdatas: Sequence[amici.ReturnData], model: amici.Model = None) → pandas.core.frame.DataFrame¶ Create a measurement dataframe in the petab format from the passed rdatas and own information.
Parameters: - rdatas – A list of rdatas as produced by pypesto.AmiciObjective.__call__(x, return_dict=True)[‘rdatas’].
- model – The amici model.
Returns: A dataframe built from the rdatas in the format as in self.petab_problem.measurement_df.
Return type: measurement_df
-
rdatas_to_simulation_df(rdatas: Sequence[amici.ReturnData], model: amici.Model = None) → pandas.core.frame.DataFrame¶ Same as rdatas_to_measurement_df, execpt a petab simulation dataframe is created, i.e. the measurement column label is adjusted.
-
Optimize¶
Multistart optimization with support for various optimizers.
-
class
pypesto.optimize.DlibOptimizer(method: str, options: Dict = None)¶ Bases:
pypesto.optimize.optimizer.OptimizerUse the Dlib toolbox for optimization.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.optimizer', '__doc__': '\n Use the Dlib toolbox for optimization.\n ', '__init__': <function DlibOptimizer.__init__>, 'minimize': <function fix_decorator.<locals>.wrapped_minimize>, 'is_least_squares': <function DlibOptimizer.is_least_squares>, 'get_default_options': <staticmethod object>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(method: str, options: Dict = None)¶ Default constructor.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.optimizer'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
get_default_options(self)¶ Create default options specific for the optimizer.
-
is_least_squares()¶
-
minimize(problem, x0, id, history_options=None)¶
-
-
class
pypesto.optimize.OptimizeOptions(startpoint_resample: bool = False, allow_failed_starts: bool = True)¶ Bases:
dictOptions for the multistart optimization.
Parameters: - startpoint_resample – Flag indicating whether initial points are supposed to be resampled if function evaluation fails at the initial point
- allow_failed_starts (bool, optional) – Flag indicating whether we tolerate that exceptions are thrown during the minimization process.
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.options', '__doc__': '\n Options for the multistart optimization.\n\n Parameters\n ----------\n startpoint_resample:\n Flag indicating whether initial points are supposed to be resampled if\n function evaluation fails at the initial point\n allow_failed_starts: bool, optional\n Flag indicating whether we tolerate that exceptions are thrown during\n the minimization process.\n ', '__init__': <function OptimizeOptions.__init__>, '__getattr__': <function OptimizeOptions.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, 'assert_instance': <staticmethod object>, '__dict__': <attribute '__dict__' of 'OptimizeOptions' objects>, '__weakref__': <attribute '__weakref__' of 'OptimizeOptions' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(startpoint_resample: bool = False, allow_failed_starts: bool = True)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.options'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
assert_instance(maybe_options: Union[OptimizeOptions, Dict]) → pypesto.optimize.options.OptimizeOptions¶ Returns a valid options object.
Parameters: maybe_options (OptimizeOptions or dict) –
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
class
pypesto.optimize.Optimizer¶ Bases:
abc.ABCThis is the optimizer base class, not functional on its own.
An optimizer takes a problem, and possibly a start point, and then performs an optimization. It returns an OptimizerResult.
-
__abstractmethods__= frozenset({'is_least_squares', 'minimize'})¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.optimizer', '__doc__': '\n This is the optimizer base class, not functional on its own.\n\n An optimizer takes a problem, and possibly a start point, and then\n performs an optimization. It returns an OptimizerResult.\n ', '__init__': <function Optimizer.__init__>, 'minimize': <function fix_decorator.<locals>.wrapped_minimize>, 'is_least_squares': <function Optimizer.is_least_squares>, 'get_default_options': <staticmethod object>, '__abstractmethods__': frozenset({'is_least_squares', 'minimize'}), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Default constructor.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.optimizer'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
get_default_options()¶ Create default options specific for the optimizer.
-
is_least_squares()¶
-
minimize(problem, x0, id, history_options=None)¶
-
-
class
pypesto.optimize.OptimizerResult(id: str = None, x: numpy.ndarray = None, fval: float = None, grad: numpy.ndarray = None, hess: numpy.ndarray = None, res: numpy.ndarray = None, sres: numpy.ndarray = None, n_fval: int = None, n_grad: int = None, n_hess: int = None, n_res: int = None, n_sres: int = None, x0: numpy.ndarray = None, fval0: float = None, history: pypesto.objective.history.History = None, exitflag: int = None, time: float = None, message: str = None)¶ Bases:
dictThe result of an optimizer run. Used as a standardized return value to map from the individual result objects returned by the employed optimizers to the format understood by pypesto.
Can be used like a dict.
-
id¶ Id of the optimizer run. Usually the start index.
-
x¶ The best found parameters.
-
fval¶ The best found function value, fun(x).
-
grad¶ The gradient at x.
-
hess¶ The Hessian at x.
-
res¶ The residuals at x.
-
sres¶ The residual sensitivities at x.
-
n_fval¶ Number of function evaluations.
-
n_grad¶ Number of gradient evaluations.
-
n_hess¶ Number of Hessian evaluations.
-
n_res¶ Number of residuals evaluations.
-
n_sres¶ Number of residual sensitivity evaluations.
-
x0¶ The starting parameters.
-
fval0¶ The starting function value, fun(x0).
-
history¶ Objective history.
-
exitflag¶ The exitflag of the optimizer.
-
time¶ Execution time.
-
message¶ Textual comment on the optimization result.
Type: str
Notes
Any field not supported by the optimizer is filled with None.
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.result', '__doc__': '\n The result of an optimizer run. Used as a standardized return value to\n map from the individual result objects returned by the employed\n optimizers to the format understood by pypesto.\n\n Can be used like a dict.\n\n Attributes\n ----------\n id:\n Id of the optimizer run. Usually the start index.\n x:\n The best found parameters.\n fval:\n The best found function value, `fun(x)`.\n grad:\n The gradient at `x`.\n hess:\n The Hessian at `x`.\n res:\n The residuals at `x`.\n sres:\n The residual sensitivities at `x`.\n n_fval\n Number of function evaluations.\n n_grad:\n Number of gradient evaluations.\n n_hess:\n Number of Hessian evaluations.\n n_res:\n Number of residuals evaluations.\n n_sres:\n Number of residual sensitivity evaluations.\n x0:\n The starting parameters.\n fval0:\n The starting function value, `fun(x0)`.\n history:\n Objective history.\n exitflag:\n The exitflag of the optimizer.\n time:\n Execution time.\n message: str\n Textual comment on the optimization result.\n\n Notes\n -----\n\n Any field not supported by the optimizer is filled with None.\n ', '__init__': <function OptimizerResult.__init__>, '__getattr__': <function OptimizerResult.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, '__dict__': <attribute '__dict__' of 'OptimizerResult' objects>, '__weakref__': <attribute '__weakref__' of 'OptimizerResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(id: str = None, x: numpy.ndarray = None, fval: float = None, grad: numpy.ndarray = None, hess: numpy.ndarray = None, res: numpy.ndarray = None, sres: numpy.ndarray = None, n_fval: int = None, n_grad: int = None, n_hess: int = None, n_res: int = None, n_sres: int = None, x0: numpy.ndarray = None, fval0: float = None, history: pypesto.objective.history.History = None, exitflag: int = None, time: float = None, message: str = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
-
class
pypesto.optimize.PyswarmOptimizer(options: Dict = None)¶ Bases:
pypesto.optimize.optimizer.OptimizerGlobal optimization using pyswarm.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.optimizer', '__doc__': '\n Global optimization using pyswarm.\n ', '__init__': <function PyswarmOptimizer.__init__>, 'minimize': <function fix_decorator.<locals>.wrapped_minimize>, 'is_least_squares': <function PyswarmOptimizer.is_least_squares>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Default constructor.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.optimizer'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
get_default_options()¶ Create default options specific for the optimizer.
-
is_least_squares()¶
-
minimize(problem, x0, id, history_options=None)¶
-
-
class
pypesto.optimize.ScipyOptimizer(method: str = 'L-BFGS-B', tol: float = 1e-09, options: Dict = None)¶ Bases:
pypesto.optimize.optimizer.OptimizerUse the SciPy optimizers.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.optimize.optimizer', '__doc__': '\n Use the SciPy optimizers.\n ', '__init__': <function ScipyOptimizer.__init__>, 'minimize': <function fix_decorator.<locals>.wrapped_minimize>, 'is_least_squares': <function ScipyOptimizer.is_least_squares>, 'get_default_options': <staticmethod object>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(method: str = 'L-BFGS-B', tol: float = 1e-09, options: Dict = None)¶ Default constructor.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.optimize.optimizer'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
static
get_default_options(self)¶ Create default options specific for the optimizer.
-
is_least_squares()¶
-
minimize(problem, x0, id, history_options=None)¶
-
-
pypesto.optimize.minimize(problem: pypesto.problem.Problem, optimizer: pypesto.optimize.optimizer.Optimizer = None, n_starts: int = 100, ids: Iterable[str] = None, startpoint_method: Union[Callable, bool] = None, result: pypesto.result.Result = None, engine: pypesto.engine.base.Engine = None, options: pypesto.optimize.options.OptimizeOptions = None, history_options: pypesto.objective.history.HistoryOptions = None) → pypesto.result.Result¶ This is the main function to call to do multistart optimization.
Parameters: - problem – The problem to be solved.
- optimizer – The optimizer to be used n_starts times.
- n_starts – Number of starts of the optimizer.
- ids – Ids assigned to the startpoints.
- startpoint_method – Method for how to choose start points. False means the optimizer does not require start points, e.g. ‘pso’ method in ‘GlobalOptimizer’
- result – A result object to append the optimization results to. For example, one might append more runs to a previous optimization. If None, a new object is created.
- engine – Parallelization engine. Defaults to sequential execution on a SingleCoreEngine.
- options – Various options applied to the multistart optimization.
- history_options – Optimizer history options.
Returns: Result object containing the results of all multistarts in result.optimize_result.
Return type: result
Profile¶
-
class
pypesto.profile.ProfileOptions(default_step_size: float = 0.01, min_step_size: float = 0.001, max_step_size: float = 1.0, step_size_factor: float = 1.25, delta_ratio_max: float = 0.1, ratio_min: float = 0.145, reg_points: int = 10, reg_order: int = 4, magic_factor_obj_value: float = 0.5)¶ Bases:
dictOptions for optimization based profiling.
Parameters: - default_step_size – default step size of the profiling routine along the profile path (adaptive step lengths algorithms will only use this as a first guess and then refine the update)
- min_step_size – lower bound for the step size in adaptive methods
- max_step_size – upper bound for the step size in adaptive methods
- step_size_factor – Adaptive methods recompute the likelihood at the predicted point and try to find a good step length by a sort of line search algorithm. This factor controls step handling in this line search
- delta_ratio_max – maximum allowed drop of the posterior ratio between two profile steps
- ratio_min – lower bound for likelihood ratio of the profile, based on inverse chi2-distribution. The default corresponds to 95% confidence
- reg_points – number of profile points used for regression in regression based adaptive profile points proposal
- reg_order – maximum degree of regression polynomial used in regression based adaptive profile points proposal
- magic_factor_obj_value – There is this magic factor in the old profiling code which slows down profiling at small ratios (must be >= 0 and < 1)
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.profile.profile', '__doc__': '\n Options for optimization based profiling.\n\n Parameters\n ----------\n default_step_size:\n default step size of the profiling routine along the profile path\n (adaptive step lengths algorithms will only use this as a first guess\n and then refine the update)\n min_step_size:\n lower bound for the step size in adaptive methods\n max_step_size:\n upper bound for the step size in adaptive methods\n step_size_factor:\n Adaptive methods recompute the likelihood at the predicted point and\n try to find a good step length by a sort of line search algorithm.\n This factor controls step handling in this line search\n delta_ratio_max:\n maximum allowed drop of the posterior ratio between two profile steps\n ratio_min:\n lower bound for likelihood ratio of the profile, based on inverse\n chi2-distribution.\n The default corresponds to 95% confidence\n reg_points:\n number of profile points used for regression in regression based\n adaptive profile points proposal\n reg_order:\n maximum degree of regression polynomial used in regression based\n adaptive profile points proposal\n magic_factor_obj_value:\n There is this magic factor in the old profiling code which slows down\n profiling at small ratios (must be >= 0 and < 1)\n ', '__init__': <function ProfileOptions.__init__>, '__getattr__': <function ProfileOptions.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, 'create_instance': <staticmethod object>, '__dict__': <attribute '__dict__' of 'ProfileOptions' objects>, '__weakref__': <attribute '__weakref__' of 'ProfileOptions' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(default_step_size: float = 0.01, min_step_size: float = 0.001, max_step_size: float = 1.0, step_size_factor: float = 1.25, delta_ratio_max: float = 0.1, ratio_min: float = 0.145, reg_points: int = 10, reg_order: int = 4, magic_factor_obj_value: float = 0.5)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.profile.profile'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
static
create_instance(maybe_options: Union[ProfileOptions, Dict]) → pypesto.profile.profile.ProfileOptions¶ Returns a valid options object.
Parameters: maybe_options (ProfileOptions or dict) –
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
class
pypesto.profile.ProfilerResult(x_path, fval_path, ratio_path, gradnorm_path=None, exitflag_path=None, time_path=None, time_total=0.0, n_fval=0, n_grad=0, n_hess=0, message=None)¶ Bases:
dictThe result of a profiler run. The standardized return return value from pypesto.profile, which can either be initialized from an OptimizerResult or from an existing ProfilerResult (in order to extend the computation).
Can be used like a dict.
-
x_path¶ The path of the best found parameters along the profile (Dimension: n_par x n_profile_points)
Type: ndarray
-
fval_path¶ The function values, fun(x), along the profile.
Type: ndarray
-
ratio_path¶ The ratio of the posterior function along the profile.
Type: ndarray
-
gradnorm_path¶ The gradient norm along the profile.
Type: ndarray
-
exitflag_path¶ The exitflags of the optimizer along the profile.
Type: ndarray
-
time_path¶ The computation time of the optimizer runs along the profile.
Type: ndarray
-
time_total¶ The total computation time for the profile.
Type: ndarray
-
n_fval¶ Number of function evaluations.
Type: int
-
n_grad¶ Number of gradient evaluations.
Type: int
-
n_hess¶ Number of Hessian evaluations.
Type: int
-
message¶ Textual comment on the profile result.
Type: str
Notes
Any field not supported by the profiler or the profiling optimizer is filled with None. Some fields are filled by pypesto itself.
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.profile.result', '__doc__': '\n The result of a profiler run. The standardized return return value from\n pypesto.profile, which can either be initialized from an OptimizerResult\n or from an existing ProfilerResult (in order to extend the computation).\n\n Can be used like a dict.\n\n Attributes\n ----------\n\n x_path: ndarray\n The path of the best found parameters along the profile\n (Dimension: n_par x n_profile_points)\n\n fval_path: ndarray\n The function values, fun(x), along the profile.\n\n ratio_path: ndarray\n The ratio of the posterior function along the profile.\n\n gradnorm_path: ndarray\n The gradient norm along the profile.\n\n exitflag_path: ndarray\n The exitflags of the optimizer along the profile.\n\n time_path: ndarray\n The computation time of the optimizer runs along the profile.\n\n time_total: ndarray\n The total computation time for the profile.\n\n n_fval: int\n Number of function evaluations.\n\n n_grad: int\n Number of gradient evaluations.\n\n n_hess: int\n Number of Hessian evaluations.\n\n message: str\n Textual comment on the profile result.\n\n Notes\n -----\n\n Any field not supported by the profiler or the profiling optimizer is\n filled with None. Some fields are filled by pypesto itself.\n ', '__init__': <function ProfilerResult.__init__>, '__getattr__': <function ProfilerResult.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, 'append_profile_point': <function ProfilerResult.append_profile_point>, 'flip_profile': <function ProfilerResult.flip_profile>, '__dict__': <attribute '__dict__' of 'ProfilerResult' objects>, '__weakref__': <attribute '__weakref__' of 'ProfilerResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(x_path, fval_path, ratio_path, gradnorm_path=None, exitflag_path=None, time_path=None, time_total=0.0, n_fval=0, n_grad=0, n_hess=0, message=None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.profile.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
append_profile_point(x, fval, ratio, gradnorm=nan, exitflag=nan, time=nan, n_fval=0, n_grad=0, n_hess=0)¶ This function appends a new OptimizerResult to an existing ProfilerResults
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
flip_profile()¶ This function flips the profiling direction (left-right) Profiling direction needs to be changed once (if the profile is new) and twice, if we append to an existing profile
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
-
pypesto.profile.parameter_profile(problem: pypesto.problem.Problem, result: pypesto.result.Result, optimizer: pypesto.optimize.optimizer.Optimizer, profile_index: numpy.ndarray = None, profile_list: int = None, result_index: int = 0, next_guess_method: Callable = None, profile_options: pypesto.profile.profile.ProfileOptions = None) → pypesto.result.Result¶ This is the main function to call to do parameter profiling.
Parameters: - problem – The problem to be solved.
- result – A result object to initialize profiling and to append the profiling results to. For example, one might append more profiling runs to a previous profile, in order to merge these. The existence of an optimization result is obligatory.
- optimizer – The optimizer to be used along each profile.
- profile_index – array with parameter indices, whether a profile should be computed (1) or not (0) Default is all profiles should be computed
- profile_list – integer which specifies whether a call to the profiler should create a new list of profiles (default) or should be added to a specific profile list
- result_index – index from which optimization result profiling should be started (default: global optimum, i.e., index = 0)
- next_guess_method – function handle to a method that creates the next starting point for optimization in profiling.
- profile_options – Various options applied to the profile optimization.
Returns: The profile results are filled into result.profile_result.
Return type: result
Sampling¶
Draw samples from the distribution, with support for various samplers.
-
class
pypesto.sampling.AdaptiveMetropolisSampler(options: Dict = None)¶ Bases:
pypesto.sampling.metropolis.MetropolisSamplerMetropolis-Hastings sampler with adaptive proposal covariance.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.adaptive_metropolis', '__doc__': '\n Metropolis-Hastings sampler with adaptive proposal covariance.\n ', '__init__': <function AdaptiveMetropolisSampler.__init__>, 'default_options': <classmethod object>, 'initialize': <function AdaptiveMetropolisSampler.initialize>, '_propose_parameter': <function AdaptiveMetropolisSampler._propose_parameter>, '_update_proposal': <function AdaptiveMetropolisSampler._update_proposal>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.adaptive_metropolis'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
classmethod
default_options()¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_last_sample() → pypesto.sampling.sampler.InternalSample¶ Get the last sample in the chain.
Returns: The last sample in the chain in the exchange format. Return type: internal_sample
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Get the generated samples.
-
initialize(problem: pypesto.problem.Problem, x0: numpy.ndarray)¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
set_last_sample(sample: pypesto.sampling.sampler.InternalSample)¶ Set the last sample in the chain to the passed value.
Parameters: sample – The sample that will replace the last sample in the chain.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
class
pypesto.sampling.AdaptiveParallelTemperingSampler(internal_sampler: pypesto.sampling.sampler.InternalSampler, betas: Sequence[float] = None, n_chains: int = None, options: Dict = None)¶ Bases:
pypesto.sampling.parallel_tempering.ParallelTemperingSamplerParallel tempering sampler with adaptive temperature adaptation.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.adaptive_parallel_tempering', '__doc__': 'Parallel tempering sampler with adaptive temperature adaptation.', 'default_options': <classmethod object>, 'adjust_betas': <function AdaptiveParallelTemperingSampler.adjust_betas>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(internal_sampler: pypesto.sampling.sampler.InternalSampler, betas: Sequence[float] = None, n_chains: int = None, options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.adaptive_parallel_tempering'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
adjust_betas(i_sample: int, swapped: Sequence[bool])¶ Update temperatures as in Vousden2016.
-
classmethod
default_options() → Dict¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Concatenate all chains.
-
initialize(problem: pypesto.problem.Problem, x0: Union[numpy.ndarray, List[numpy.ndarray]])¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
swap_samples() → Sequence[bool]¶ Swap samples as in Vousden2016.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
class
pypesto.sampling.InternalSampler(options: Dict = None)¶ Bases:
pypesto.sampling.sampler.SamplerSampler to be used inside a parallel tempering sampler.
The last sample can be obtained via get_last_sample and set via set_last_sample.
-
__abstractmethods__= frozenset({'sample', 'initialize', 'get_samples', 'set_last_sample', 'get_last_sample'})¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.sampler', '__doc__': 'Sampler to be used inside a parallel tempering sampler.\n\n The last sample can be obtained via `get_last_sample` and set via\n `set_last_sample`.\n ', 'get_last_sample': <function InternalSampler.get_last_sample>, 'set_last_sample': <function InternalSampler.set_last_sample>, '__abstractmethods__': frozenset({'sample', 'initialize', 'get_samples', 'set_last_sample', 'get_last_sample'}), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.sampler'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
classmethod
default_options() → Dict¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_last_sample() → pypesto.sampling.sampler.InternalSample¶ Get the last sample in the chain.
Returns: The last sample in the chain in the exchange format. Return type: internal_sample
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Get the generated samples.
-
initialize(problem: pypesto.problem.Problem, x0: Union[numpy.ndarray, List[numpy.ndarray]])¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
set_last_sample(sample: pypesto.sampling.sampler.InternalSample)¶ Set the last sample in the chain to the passed value.
Parameters: sample – The sample that will replace the last sample in the chain.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
class
pypesto.sampling.McmcPtResult(trace_x: numpy.ndarray, trace_fval: numpy.ndarray, betas: Iterable[float], message: str = None)¶ Bases:
dictThe result of a sampler run using Markov-chain Monte Carlo, and optionally parallel tempering.
Can be used like a dict.
Parameters: - trace_x ([n_chain, n_iter, n_par]) – Parameters
- trace_fval ([n_chain, n_iter]) – Function values.
- betas ([n_chain]) – The associated inverse temperatures.
- message (str) – Textual comment on the profile result.
- n_chain denotes the number of chains, n_iter the number of (Here,) –
- (i.e., the chain length), and n_par the number of parameters. (iterations) –
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.result', '__doc__': 'The result of a sampler run using Markov-chain Monte Carlo, and\n optionally parallel tempering.\n\n Can be used like a dict.\n\n Parameters\n ----------\n trace_x: [n_chain, n_iter, n_par]\n Parameters\n trace_fval: [n_chain, n_iter]\n Function values.\n betas: [n_chain]\n The associated inverse temperatures.\n message: str\n Textual comment on the profile result.\n\n Here, `n_chain` denotes the number of chains, `n_iter` the number of\n iterations (i.e., the chain length), and `n_par` the number of parameters.\n ', '__init__': <function McmcPtResult.__init__>, '__getattr__': <function McmcPtResult.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, '__dict__': <attribute '__dict__' of 'McmcPtResult' objects>, '__weakref__': <attribute '__weakref__' of 'McmcPtResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(trace_x: numpy.ndarray, trace_fval: numpy.ndarray, betas: Iterable[float], message: str = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
class
pypesto.sampling.MetropolisSampler(options: Dict = None)¶ Bases:
pypesto.sampling.sampler.InternalSamplerSimple Metropolis-Hastings sampler with fixed proposal variance.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.metropolis', '__doc__': '\n Simple Metropolis-Hastings sampler with fixed proposal variance.\n ', '__init__': <function MetropolisSampler.__init__>, 'default_options': <classmethod object>, 'initialize': <function MetropolisSampler.initialize>, 'sample': <function MetropolisSampler.sample>, '_perform_step': <function MetropolisSampler._perform_step>, '_propose_parameter': <function MetropolisSampler._propose_parameter>, '_update_proposal': <function MetropolisSampler._update_proposal>, 'get_last_sample': <function MetropolisSampler.get_last_sample>, 'set_last_sample': <function MetropolisSampler.set_last_sample>, 'get_samples': <function MetropolisSampler.get_samples>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.metropolis'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
classmethod
default_options()¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_last_sample() → pypesto.sampling.sampler.InternalSample¶ Get the last sample in the chain.
Returns: The last sample in the chain in the exchange format. Return type: internal_sample
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Get the generated samples.
-
initialize(problem: pypesto.problem.Problem, x0: numpy.ndarray)¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
set_last_sample(sample: pypesto.sampling.sampler.InternalSample)¶ Set the last sample in the chain to the passed value.
Parameters: sample – The sample that will replace the last sample in the chain.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
class
pypesto.sampling.ParallelTemperingSampler(internal_sampler: pypesto.sampling.sampler.InternalSampler, betas: Sequence[float] = None, n_chains: int = None, options: Dict = None)¶ Bases:
pypesto.sampling.sampler.SamplerSimple parallel tempering sampler.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.parallel_tempering', '__doc__': 'Simple parallel tempering sampler.', '__init__': <function ParallelTemperingSampler.__init__>, 'default_options': <classmethod object>, 'initialize': <function ParallelTemperingSampler.initialize>, 'sample': <function ParallelTemperingSampler.sample>, 'get_samples': <function ParallelTemperingSampler.get_samples>, 'swap_samples': <function ParallelTemperingSampler.swap_samples>, 'adjust_betas': <function ParallelTemperingSampler.adjust_betas>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(internal_sampler: pypesto.sampling.sampler.InternalSampler, betas: Sequence[float] = None, n_chains: int = None, options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.parallel_tempering'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
adjust_betas(i_sample: int, swapped: Sequence[bool])¶ Adjust temperature values. Default: Do nothing.
-
classmethod
default_options() → Dict¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Concatenate all chains.
-
initialize(problem: pypesto.problem.Problem, x0: Union[numpy.ndarray, List[numpy.ndarray]])¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
swap_samples() → Sequence[bool]¶ Swap samples as in Vousden2016.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
class
pypesto.sampling.Sampler(options: Dict = None)¶ Bases:
abc.ABCSampler base class, not functional on its own.
The sampler maintains an internal chain, which is initialized in initialize, and updated in sample.
-
__abstractmethods__= frozenset({'sample', 'initialize', 'get_samples'})¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.sampling.sampler', '__doc__': 'Sampler base class, not functional on its own.\n\n The sampler maintains an internal chain, which is initialized in\n `initialize`, and updated in `sample`.\n ', '__init__': <function Sampler.__init__>, 'initialize': <function Sampler.initialize>, 'sample': <function Sampler.sample>, 'get_samples': <function Sampler.get_samples>, 'default_options': <classmethod object>, 'translate_options': <classmethod object>, '__abstractmethods__': frozenset({'sample', 'initialize', 'get_samples'}), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(options: Dict = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.sampling.sampler'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
classmethod
default_options() → Dict¶ Convenience method to set/get default options.
Returns: Default sampler options. Return type: default_options
-
get_samples() → pypesto.sampling.result.McmcPtResult¶ Get the generated samples.
-
initialize(problem: pypesto.problem.Problem, x0: Union[numpy.ndarray, List[numpy.ndarray]])¶ Initialize the sampler.
Parameters: - problem – The problem for which to sample.
- x0 – Should, but is not required to, be used as initial parameter.
-
sample(n_samples: int, beta: float = 1.0)¶ Perform sampling.
Parameters: - n_samples – Number of samples to generate.
- beta – Inverse of the temperature to which the system is elevated.
-
classmethod
translate_options(options)¶ Convenience method to translate options and fill in defaults.
Parameters: options – Options configuring the sampler.
-
-
pypesto.sampling.sample(problem: pypesto.problem.Problem, n_samples: int, sampler: pypesto.sampling.sampler.Sampler = None, x0: Union[numpy.ndarray, List[numpy.ndarray]] = None, result: pypesto.result.Result = None) → pypesto.result.Result¶ This is the main function to call to do parameter sampling.
Parameters: - problem – The problem to be solved. If None is provided, a
pypesto.AdaptiveMetropolisSampleris used. - n_samples – Number of samples to generate.
- sampler – The sampler to perform the actual sampling.
- x0 – Initial parameter for the Markov chain. If None, the best parameter found in optimization is used. Note that some samplers require an initial parameter, some may ignore it. x0 can also be a list, to have separate starting points for parallel tempering chains.
- result – A result to write to. If None provided, one is created from the problem.
Returns: A result with filled in sample_options part.
Return type: result
- problem – The problem to be solved. If None is provided, a
Visualize¶
pypesto comes with various visualization routines. To use these, import pypesto.visualize.
-
class
pypesto.visualize.ReferencePoint(reference=None, x=None, fval=None, color=None, legend=None)¶ Bases:
dictReference point for plotting. Should contain a parameter value and an objective function value, may alos contain a color and a legend.
Can be used like a dict.
-
x¶ Reference parameters.
Type: ndarray
-
fval¶ Function value, fun(x), for reference parameters.
Type: float
-
color¶ Color which should be used for reference point.
Type: RGBA, optional
-
auto_color¶ flag indicating whether color for this reference point should be assigned automatically or whether it was assigned by user
Type: boolean
-
legend¶ legend text for reference point
Type: str
-
__class__¶ alias of
builtins.type
-
__contains__()¶ True if D has a key k, else False.
-
__delattr__¶ Delete self[key].
-
__delitem__¶ Delete self[key].
-
__dict__= mappingproxy({'__module__': 'pypesto.visualize.reference_points', '__doc__': '\n Reference point for plotting. Should contain a parameter value and an\n objective function value, may alos contain a color and a legend.\n\n Can be used like a dict.\n\n Attributes\n ----------\n\n x: ndarray\n Reference parameters.\n\n fval: float\n Function value, fun(x), for reference parameters.\n\n color: RGBA, optional\n Color which should be used for reference point.\n\n auto_color: boolean\n flag indicating whether color for this reference point should be\n assigned automatically or whether it was assigned by user\n\n legend: str\n legend text for reference point\n ', '__init__': <function ReferencePoint.__init__>, '__getattr__': <function ReferencePoint.__getattr__>, '__setattr__': <slot wrapper '__setitem__' of 'dict' objects>, '__delattr__': <slot wrapper '__delitem__' of 'dict' objects>, '__dict__': <attribute '__dict__' of 'ReferencePoint' objects>, '__weakref__': <attribute '__weakref__' of 'ReferencePoint' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattr__(key)¶
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__()¶ x.__getitem__(y) <==> x[y]
-
__gt__¶ Return self>value.
-
__hash__= None¶
-
__init__(reference=None, x=None, fval=None, color=None, legend=None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__¶ Implement iter(self).
-
__le__¶ Return self<=value.
-
__len__¶ Return len(self).
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.visualize.reference_points'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Set self[key] to value.
-
__setitem__¶ Set self[key] to value.
-
__sizeof__() → size of D in memory, in bytes¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
clear() → None. Remove all items from D.¶
-
copy() → a shallow copy of D¶
-
fromkeys()¶ Returns a new dict with keys from iterable and values equal to value.
-
get(k[, d]) → D[k] if k in D, else d. d defaults to None.¶
-
items() → a set-like object providing a view on D's items¶
-
keys() → a set-like object providing a view on D's keys¶
-
pop(k[, d]) → v, remove specified key and return the corresponding value.¶ If key is not found, d is returned if given, otherwise KeyError is raised
-
popitem() → (k, v), remove and return some (key, value) pair as a¶ 2-tuple; but raise KeyError if D is empty.
-
setdefault(k[, d]) → D.get(k,d), also set D[k]=d if k not in D¶
-
update([E, ]**F) → None. Update D from dict/iterable E and F.¶ If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
-
values() → an object providing a view on D's values¶
-
-
pypesto.visualize.assign_clustered_colors(vals, balance_alpha=True, highlight_global=True)¶ Cluster and assign colors.
Parameters: - vals (numeric list or array) – List to be clustered and assigned colors.
- balance_alpha (bool (optional)) – Flag indicating whether alpha for large clusters should be reduced to avoid overplotting (default: True)
- highlight_global (bool (optional)) – flag indicating whether global optimum should be highlighted
Returns: colors – One for each element in ‘vals’.
Return type: list of RGBA
-
pypesto.visualize.assign_clusters(vals)¶ Find clustering.
Parameters: vals (numeric list or array) – List to be clustered. Returns: - clust (numeric list) – Indicating the corresponding cluster of each element from ‘vals’.
- clustsize (numeric list) – Size of clusters, length equals number of clusters.
-
pypesto.visualize.assign_colors(vals, colors=None, balance_alpha=True, highlight_global=True)¶ Assign colors or format user specified colors.
Parameters: - vals (numeric list or array) – List to be clustered and assigned colors.
- colors (list, or RGBA, optional) – list of colors, or single color
- balance_alpha (bool (optional)) – Flag indicating whether alpha for large clusters should be reduced to avoid overplotting (default: True)
- highlight_global (bool (optional)) – flag indicating whether global optimum should be highlighted
Returns: colors – One for each element in ‘vals’.
Return type: list of RGBA
-
pypesto.visualize.create_references(references=None, x=None, fval=None, color=None, legend=None) → List[pypesto.visualize.reference_points.ReferencePoint]¶ This function creates a list of reference point objects from user inputs
Parameters: - references (ReferencePoint or dict or list, optional) – Will be converted into a list of RefPoints
- x (ndarray, optional) – Parameter vector which should be used for reference point
- fval (float, optional) – Objective function value which should be used for reference point
- color (RGBA, optional) – Color which should be used for reference point.
- legend (str) – legend text for reference point
Returns: colors – One for each element in ‘vals’.
Return type: list of RGBA
-
pypesto.visualize.delete_nan_inf(fvals: numpy.ndarray, x: numpy.ndarray = None) → Tuple[numpy.ndarray, numpy.ndarray]¶ Delete nan and inf values in fvals. If parameters ‘x’ are passend, also the corresponding entries are deleted.
Parameters: - x – array of parameters
- fvals – array of fval
Returns: - x (np.array) – array of parameters without nan or inf
- fvals (np.array) – array of fval without nan or inf
-
pypesto.visualize.optimizer_history(results, ax=None, size=(18.5, 10.5), trace_x='steps', trace_y='fval', scale_y='log10', offset_y=None, colors=None, y_limits=None, start_indices=None, reference=None, legends=None)¶ Plot history of optimizer. Can plot either the history of the cost function or of the gradient norm, over either the optimizer steps or the computation time.
Parameters: - results (pypesto.Result or list) – Optimization result obtained by ‘optimize.py’ or list of those
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- trace_x (str, optional) – What should be plotted on the x-axis? Possibilities: ‘time’, ‘steps’ Default: ‘steps’
- trace_y (str, optional) – What should be plotted on the y-axis? Possibilities: ‘fval’, ‘gradnorm’, ‘stepsize’ Default: ‘fval’
- scale_y (str, optional) – May be logarithmic or linear (‘log10’ or ‘lin’)
- offset_y (float, optional) – Offset for the y-axis-values, as these are plotted on a log10-scale Will be computed automatically if necessary
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- y_limits (float or ndarray, optional) – maximum value to be plotted on the y-axis, or y-limits
- start_indices (list or int) – list of integers specifying the multistart to be plotted or int specifying up to which start index should be plotted
- reference (list, optional) – List of reference points for optimization results, containing et least a function value fval
- legends (list or str) – Labels for line plots, one label per result object
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.optimizer_history_lowlevel(vals, scale_y='log10', colors=None, ax=None, size=(18.5, 10.5), x_label='Optimizer steps', y_label='Objective value', legend_text=None)¶ Plot optimizer history using list of numpy arrays.
Parameters: - vals (list of numpy arrays) – list of 2xn-arrays (x_values and y_values of the trace)
- scale_y (str, optional) – May be logarithmic or linear (‘log10’ or ‘lin’)
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – see waterfall
- x_label (str) – label for x-axis
- y_label (str) – label for y-axis
- legend_text (str) – Label for line plots
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.parameters(results, ax=None, free_indices_only=True, lb=None, ub=None, size=None, reference=None, colors=None, legends=None, balance_alpha=True, start_indices=None)¶ Plot parameter values.
Parameters: - results (pypesto.Result or list) – Optimization result obtained by ‘optimize.py’ or list of those
- ax (matplotlib.Axes, optional) – Axes object to use.
- free_indices_only (bool, optional) – If True, only free parameters are shown. If False, also the fixed parameters are shown.
- ub (lb,) – If not None, override result.problem.lb, problem.problem.ub. Dimension either result.problem.dim or result.problem.dim_full.
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- reference (list, optional) – List of reference points for optimization results, containing et least a function value fval
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- legends (list or str) – Labels for line plots, one label per result object
- balance_alpha (bool (optional)) – Flag indicating whether alpha for large clusters should be reduced to avoid overplotting (default: True)
- start_indices (list or int) – list of integers specifying the multistarts to be plotted or int specifying up to which start index should be plotted
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.parameters_lowlevel(xs, fvals, lb=None, ub=None, x_labels=None, ax=None, size=None, colors=None, linestyle='-', legend_text=None, balance_alpha=True)¶ Plot parameters plot using list of parameters.
Parameters: - xs (nested list or array) – Including optimized parameters for each startpoint. Shape: (n_starts, dim).
- fvals (numeric list or array) – Function values. Needed to assign cluster colors.
- ub (lb,) – The lower and upper bounds.
- x_labels (array_like of str, optional) – Labels to be used for the parameters.
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – see parameters
- colors (list of RGBA) – One for each element in ‘fvals’.
- linestyle (str, optional) – linestyle argument for parameter plot
- legend_text (str) – Label for line plots
- balance_alpha (bool (optional)) – Flag indicating whether alpha for large clusters should be reduced to avoid overplotting (default: True)
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.process_offset_y(offset_y: Optional[float], scale_y: str, min_val: float) → float¶ compute offset for y-axis, depend on user settings
Parameters: - offset_y – value for offsetting the later plotted values, in order to ensure positivity if a semilog-plot is used
- scale_y – Can be ‘lin’ or ‘log10’, specifying whether values should be plotted on linear or on log10-scale
- min_val – Smallest value to be plotted
Returns: offset_y – value for offsetting the later plotted values, in order to ensure positivity if a semilog-plot is used
Return type: float
-
pypesto.visualize.process_result_list(results, colors=None, legends=None)¶ assigns colors and legends to a list of results, chekc user provided lists
Parameters: - results (list or pypesto.Result) – list of pypesto.Result objects or a single pypesto.Result
- colors (list, optional) – list of RGBA colors
- legends (str or list) – labels for line plots
Returns: - results (list of pypesto.Result) – list of pypesto.Result objects
- colors (list of RGBA) – One for each element in ‘results’.
- legends (list of str) – labels for line plots
-
pypesto.visualize.process_y_limits(ax, y_limits)¶ apply user specified limits of y-axis
Parameters: - ax (matplotlib.Axes, optional) – Axes object to use.
- y_limits (ndarray) – y_limits, minimum and maximum, for current axes object
- min_val (float) – Smallest value to be plotted
Returns: ax – Axes object to use.
Return type: matplotlib.Axes, optional
-
pypesto.visualize.profile_lowlevel(fvals, ax=None, size=(18.5, 6.5), color=None, legend_text=None)¶ Lowlevel routine for plotting one profile, working with a numpy array only
Parameters: - fvals (numeric list or array) – Including values need to be plotted.
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- color (RGBA, optional) – color for profiles in plot.
- legend_text (str) – Label for line plots
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.profiles(results, ax=None, profile_indices=None, size=(18.5, 6.5), reference=None, colors=None, legends=None, profile_list_id=0)¶ Plot classical 1D profile plot (using the posterior, e.g. Gaussian like profile)
Parameters: - results (list or pypesto.Result) – list of pypesto.Result or single pypesto.Result
- ax (list of matplotlib.Axes, optional) – List of axes objects to use.
- profile_indices (list of integer values) – list of integer values specifying which profiles should be plotted
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- reference (list, optional) – List of reference points for optimization results, containing et least a function value fval
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- legends (list or str, optional) – Labels for line plots, one label per result object
- profile_list_id (int, optional) – index of the profile list to be used for profiling
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.profiles_lowlevel(fvals, ax=None, size=(18.5, 6.5), color=None, legend_text=None)¶ Lowlevel routine for profile plotting, working with a list of arrays only, opening different axes objects in case
Parameters: - fvals (numeric list or array) – Including values need to be plotted.
- ax (list of matplotlib.Axes, optional) – list of axes object to use.
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- size – Figure size (width, height) in inches. Is only applied when no ax object is specified
- color (RGBA, optional) – color for profiles in plot.
- legend_text (str) – Label for line plots
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.sampling_1d_marginals(result: pypesto.result.Result, i_chain: int = 0, burn_in: int = None, stepsize: int = 1, plot_type: str = 'both', bw: str = 'scott', suptitle: str = None, size: Tuple[float, float] = None)¶ Plot marginals.
Parameters: - result – The pyPESTO result object with filled sample result.
- i_chain – Which chain to plot. Default: First chain.
- burn_in – Index after burn-in phase, thus also the burn-in length.
- stepsize – Only one in stepsize values is plotted.
- plot_type ({'hist'|'kde'|'both'}) – Specify whether to plot a histogram (‘hist’), a kernel density estimate (‘kde’), or both (‘both’).
- bw ({'scott', 'silverman' | scalar | pair of scalars}) – Kernel bandwidth method.
- suptitle – Figure super title.
- size – Figure size in inches.
Returns: ax
Return type: matplotlib-axes
-
pypesto.visualize.sampling_fval_trace(result: pypesto.result.Result, i_chain: int = 0, burn_in: int = None, stepsize: int = 1, title: str = None, size: Tuple[float, float] = None, ax: matplotlib.axes._axes.Axes = None)¶ Plot log-posterior (=function value) over iterations.
Parameters: - result – The pyPESTO result object with filled sample result.
- i_chain – Which chain to plot. Default: First chain.
- burn_in – Index after burn-in phase, thus also the burn-in length.
- stepsize – Only one in stepsize values is plotted.
- title – Axes title.
- size (ndarray) – Figure size in inches.
- ax – Axes object to use.
Returns: The plot axes.
Return type: ax
-
pypesto.visualize.sampling_parameters_trace(result: pypesto.result.Result, i_chain: int = 0, burn_in: int = None, stepsize: int = 1, use_problem_bounds: bool = True, suptitle: str = None, size: Tuple[float, float] = None, ax: matplotlib.axes._axes.Axes = None)¶ Plot parameter values over iterations.
Parameters: - result – The pyPESTO result object with filled sample result.
- i_chain – Which chain to plot. Default: First chain.
- burn_in – Index after burn-in phase, thus also the burn-in length.
- stepsize – Only one in stepsize values is plotted.
- use_problem_bounds – Defines if the y-limits shall be the lower and upper bounds of parameter estimation problem.
- suptitle – Figure suptitle.
- size – Figure size in inches.
- ax – Axes object to use.
Returns: The plot axes.
Return type: ax
-
pypesto.visualize.sampling_scatter(result: pypesto.result.Result, i_chain: int = 0, burn_in: int = None, stepsize: int = 1, suptitle: str = None, size: Tuple[float, float] = None)¶ Parameter scatter plot.
Parameters: - result – The pyPESTO result object with filled sample result.
- i_chain – Which chain to plot. Default: First chain.
- burn_in – Index after burn-in phase, thus also the burn-in length.
- stepsize – Only one in stepsize values is plotted.
- suptitle – Figure super title.
- size – Figure size in inches.
Returns: The plot axes.
Return type: ax
-
pypesto.visualize.waterfall(results, ax=None, size=(18.5, 10.5), y_limits=None, scale_y='log10', offset_y=None, start_indices=None, reference=None, colors=None, legends=None)¶ Plot waterfall plot.
Parameters: - results (pypesto.Result or list) – Optimization result obtained by ‘optimize.py’ or list of those
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – Figure size (width, height) in inches. Is only applied when no ax object is specified
- y_limits (float or ndarray, optional) – maximum value to be plotted on the y-axis, or y-limits
- scale_y (str, optional) – May be logarithmic or linear (‘log10’ or ‘lin’)
- offset_y – offset for the y-axis, if it is supposed to be in log10-scale
- start_indices (list or int) – list of integers specifying the multistart to be plotted or int specifying up to which start index should be plotted
- reference (list, optional) – List of reference points for optimization results, containing et least a function value fval
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- legends (list or str) – Labels for line plots, one label per result object
Returns: ax – The plot axes.
Return type: matplotlib.Axes
-
pypesto.visualize.waterfall_lowlevel(fvals, scale_y='log10', offset_y=0.0, ax=None, size=(18.5, 10.5), colors=None, legend_text=None)¶ Plot waterfall plot using list of function values.
Parameters: - fvals (numeric list or array) – Including values need to be plotted.
- scale_y (str, optional) – May be logarithmic or linear (‘log10’ or ‘lin’)
- offset_y – offset for the y-axis, if it is supposed to be in log10-scale
- ax (matplotlib.Axes, optional) – Axes object to use.
- size (tuple, optional) – see waterfall
- colors (list, or RGBA, optional) – list of colors, or single color color or list of colors for plotting. If not set, clustering is done and colors are assigned automatically
- legend_text (str) – Label for line plots
Returns: ax – The plot axes.
Return type: matplotlib.Axes
Result¶
The pypesto.Result object contains all results generated by the pypesto components. It contains sub-results for optimization, profiles, sampling.
-
class
pypesto.result.OptimizeResult¶ Bases:
objectResult of the minimize() function.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.result', '__doc__': '\n Result of the minimize() function.\n ', '__init__': <function OptimizeResult.__init__>, 'append': <function OptimizeResult.append>, 'sort': <function OptimizeResult.sort>, 'as_dataframe': <function OptimizeResult.as_dataframe>, 'as_list': <function OptimizeResult.as_list>, 'get_for_key': <function OptimizeResult.get_for_key>, '__dict__': <attribute '__dict__' of 'OptimizeResult' objects>, '__weakref__': <attribute '__weakref__' of 'OptimizeResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
append(optimizer_result: pypesto.optimize.result.OptimizerResult)¶ Append an optimizer result to the result object.
Parameters: optimizer_result – The result of one (local) optimizer run.
-
as_dataframe(keys=None) → pandas.core.frame.DataFrame¶ Get as pandas DataFrame. If keys is a list, return only the specified values.
-
as_list(keys=None) → Sequence¶ Get as list. If keys is a list, return only the specified values.
Parameters: keys (list(str), optional) – Labels of the field to extract.
-
get_for_key(key) → list¶ Extract the list of values for the specified key as a list.
-
sort()¶ Sort the optimizer results by function value fval (ascending).
-
-
class
pypesto.result.ProfileResult¶ Bases:
objectResult of the profile() function.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.result', '__doc__': '\n Result of the profile() function.\n ', '__init__': <function ProfileResult.__init__>, 'create_new_profile_list': <function ProfileResult.create_new_profile_list>, 'create_new_profile': <function ProfileResult.create_new_profile>, 'add_profile': <function ProfileResult.add_profile>, 'get_current_profile': <function ProfileResult.get_current_profile>, '__dict__': <attribute '__dict__' of 'ProfileResult' objects>, '__weakref__': <attribute '__weakref__' of 'ProfileResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
add_profile(profiler_result, i_parameter)¶ Writes a profiler result to the result object at i_parameter.
Parameters: - profiler_result – The result of one (local) profiler run.
- i_parameter – integer specifying the parameter index
-
create_new_profile(profiler_result: Optional[pypesto.profile.result.ProfilerResult] = None)¶ Append an profiler result to the result object.
Parameters: - profiler_result – The result of one (local) profiler run or None, if to be left empty
- profile_list (integer) – index specifying the list of profiles, to which we want to append
-
create_new_profile_list()¶ Append an profiler result to the result object.
-
get_current_profile(i_parameter)¶ Append an profiler result to the result object.
Parameters: i_parameter – integer specifying the profile index
-
-
class
pypesto.result.Result(problem=None)¶ Bases:
objectUniversal result object for pypesto. The algorithms like optimize, profile, sample fill different parts of it.
-
problem¶ The problem underlying the results.
Type: pypesto.Problem
-
optimize_result¶ The results of the optimizer runs.
-
profile_result¶ The results of the profiler run.
-
sample_result¶ The results of the sampler run.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.result', '__doc__': '\n Universal result object for pypesto.\n The algorithms like optimize, profile,\n sample fill different parts of it.\n\n Attributes\n ----------\n\n problem: pypesto.Problem\n The problem underlying the results.\n\n optimize_result:\n The results of the optimizer runs.\n\n profile_result:\n The results of the profiler run.\n\n sample_result:\n The results of the sampler run.\n\n ', '__init__': <function Result.__init__>, '__dict__': <attribute '__dict__' of 'Result' objects>, '__weakref__': <attribute '__weakref__' of 'Result' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(problem=None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
-
class
pypesto.result.SampleResult¶ Bases:
objectResult of the sample() function.
-
__class__¶ alias of
builtins.type
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.result', '__doc__': '\n Result of the sample() function.\n ', '__init__': <function SampleResult.__init__>, '__dict__': <attribute '__dict__' of 'SampleResult' objects>, '__weakref__': <attribute '__weakref__' of 'SampleResult' objects>})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.result'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
-
class
pypesto.result.Sequence¶ Bases:
collections.abc.Sequence,typing.Reversible,typing.Collection-
__abstractmethods__= frozenset({'__len__', '__getitem__'})¶
-
__args__= None¶
-
__class__¶ alias of
GenericMeta
-
__contains__(value)¶
-
__delattr__¶ Implement delattr(self, name).
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__extra__¶ alias of
collections.abc.Sequence
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__getitem__(index)¶
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__iter__()¶
-
__le__¶ Return self<=value.
-
__len__()¶
-
__lt__¶ Return self<value.
-
__module__= 'typing'¶
-
__ne__¶ Return self!=value.
-
static
__new__(cls, *args, **kwds)¶ Create and return a new object. See help(type) for accurate signature.
-
__next_in_mro__¶ alias of
builtins.object
-
__orig_bases__= (typing.Reversible[+T_co], typing.Collection[+T_co])¶
-
__origin__= None¶
-
__parameters__= (+T_co,)¶
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__reversed__()¶
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__slots__= ()¶
-
__str__¶ Return str(self).
-
__subclasshook__()¶
-
__tree_hash__= -9223366114679872040¶
-
count(value) → integer -- return number of occurrences of value¶
-
index(value[, start[, stop]]) → integer -- return first index of value.¶ Raises ValueError if the value is not present.
Supporting start and stop arguments is optional, but recommended.
-
Engines¶
The execution of the multistarts can be parallelized in different ways, e.g. multi-threaded or cluster-based. Note that it is not checked whether a single task itself is internally parallelized.
-
class
pypesto.engine.Engine¶ Bases:
abc.ABCAbstract engine base class.
-
__abstractmethods__= frozenset({'execute'})¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.engine.base', '__doc__': '\n Abstract engine base class.\n ', '__init__': <function Engine.__init__>, 'execute': <function Engine.execute>, '__abstractmethods__': frozenset({'execute'}), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.engine.base'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
execute(tasks: List[pypesto.engine.task.Task])¶ Execute tasks.
Parameters: tasks – List of tasks to execute.
-
-
class
pypesto.engine.MultiProcessEngine(n_procs: int = None)¶ Bases:
pypesto.engine.base.EngineParallelize the task execution using multiprocessing.
Parameters: n_procs – The maximum number of processes to use in parallel. Defaults to the number of CPUs available on the system according to os.cpu_count(). The effectively used number of processes will be the minimum of n_procs and the number of tasks submitted. -
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.engine.multi_process', '__doc__': '\n Parallelize the task execution using multiprocessing.\n\n Parameters\n ----------\n n_procs:\n The maximum number of processes to use in parallel.\n Defaults to the number of CPUs available on the system according to\n `os.cpu_count()`.\n The effectively used number of processes will be the minimum of\n `n_procs` and the number of tasks submitted.\n ', '__init__': <function MultiProcessEngine.__init__>, 'execute': <function MultiProcessEngine.execute>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(n_procs: int = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.engine.multi_process'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
execute(tasks: List[pypesto.engine.task.Task])¶ Pickle tasks and distribute work over parallel processes.
-
-
class
pypesto.engine.MultiThreadEngine(n_threads: int = None)¶ Bases:
pypesto.engine.base.EngineParallelize the task execution using multithreading.
Parameters: n_threads – The maximum number of threads to use in parallel. Defaults to the number of CPUs available on the system according to os.cpu_count(). The effectively used number of threads will be the minimum of n_threads and the number of tasks submitted. -
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.engine.multi_thread', '__doc__': '\n Parallelize the task execution using multithreading.\n\n Parameters\n ----------\n n_threads:\n The maximum number of threads to use in parallel.\n Defaults to the number of CPUs available on the system according to\n `os.cpu_count()`.\n The effectively used number of threads will be the minimum of\n `n_threads` and the number of tasks submitted.\n ', '__init__': <function MultiThreadEngine.__init__>, 'execute': <function MultiThreadEngine.execute>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(n_threads: int = None)¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.engine.multi_thread'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
execute(tasks: List[pypesto.engine.task.Task])¶ Deepcopy tasks and distribute work over parallel threads.
-
-
class
pypesto.engine.OptimizerTask(optimizer: pypesto.optimize.optimizer.Optimizer, problem: pypesto.problem.Problem, x0: numpy.ndarray, id: str, options: pypesto.optimize.options.OptimizeOptions, history_options: pypesto.objective.history.HistoryOptions, handle_exception: Callable)¶ Bases:
pypesto.engine.task.TaskA multistart optimization task, performed in pypesto.minimize.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.engine.task', '__doc__': '\n A multistart optimization task, performed in `pypesto.minimize`.\n ', '__init__': <function OptimizerTask.__init__>, 'execute': <function OptimizerTask.execute>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__(optimizer: pypesto.optimize.optimizer.Optimizer, problem: pypesto.problem.Problem, x0: numpy.ndarray, id: str, options: pypesto.optimize.options.OptimizeOptions, history_options: pypesto.objective.history.HistoryOptions, handle_exception: Callable)¶ Create the task object.
Parameters: - optimizer – The optimizer to use.
- problem – The problem to solve.
- x0 – The point from which to start.
- id – The multistart id.
- options – Options object applying to optimization.
- history_options – Optimizer history options.
- handle_exception – Callable to apply when the optimization fails.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.engine.task'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
execute() → pypesto.optimize.result.OptimizerResult¶ Execute the task and return its results.
-
-
class
pypesto.engine.SingleCoreEngine¶ Bases:
pypesto.engine.base.EngineDummy engine for sequential execution on one core. Note that the objective itself may be multithreaded.
-
__abstractmethods__= frozenset()¶
-
__class__¶ alias of
abc.ABCMeta
-
__delattr__¶ Implement delattr(self, name).
-
__dict__= mappingproxy({'__module__': 'pypesto.engine.single_core', '__doc__': '\n Dummy engine for sequential execution on one core. Note that the\n objective itself may be multithreaded.\n ', '__init__': <function SingleCoreEngine.__init__>, 'execute': <function SingleCoreEngine.execute>, '__abstractmethods__': frozenset(), '_abc_registry': <_weakrefset.WeakSet object>, '_abc_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache': <_weakrefset.WeakSet object>, '_abc_negative_cache_version': 48})¶
-
__dir__() → list¶ default dir() implementation
-
__eq__¶ Return self==value.
-
__format__()¶ default object formatter
-
__ge__¶ Return self>=value.
-
__getattribute__¶ Return getattr(self, name).
-
__gt__¶ Return self>value.
-
__hash__¶ Return hash(self).
-
__init__()¶ Initialize self. See help(type(self)) for accurate signature.
-
__init_subclass__()¶ This method is called when a class is subclassed.
The default implementation does nothing. It may be overridden to extend subclasses.
-
__le__¶ Return self<=value.
-
__lt__¶ Return self<value.
-
__module__= 'pypesto.engine.single_core'¶
-
__ne__¶ Return self!=value.
-
__new__()¶ Create and return a new object. See help(type) for accurate signature.
-
__reduce__()¶ helper for pickle
-
__reduce_ex__()¶ helper for pickle
-
__repr__¶ Return repr(self).
-
__setattr__¶ Implement setattr(self, name, value).
-
__sizeof__() → int¶ size of object in memory, in bytes
-
__str__¶ Return str(self).
-
__subclasshook__()¶ Abstract classes can override this to customize issubclass().
This is invoked early on by abc.ABCMeta.__subclasscheck__(). It should return True, False or NotImplemented. If it returns NotImplemented, the normal algorithm is used. Otherwise, it overrides the normal algorithm (and the outcome is cached).
-
__weakref__¶ list of weak references to the object (if defined)
-
execute(tasks: List[pypesto.engine.task.Task])¶ Execute all tasks in a simple for loop sequentially.
-
Startpoint¶
Methods for selecting points that can be used as start points for multistart optimization. All methods have the form
method(**kwargs) -> startpoints
where the kwargs can/should include the following parameters, which are passed by pypesto:
- n_starts: int
- Number of points to generate.
- lb, ub: ndarray
- Lower and upper bound, may for most methods not contain nan or inf values.
- x_guesses: ndarray, shape=(g, dim), optional
- Parameter guesses by the user, where g denotes the number of guesses. Note that these are only possibly taken as reference points to generate new start points (e.g. to maximize some distance) depending on the method, but regardless of g, there are always n_starts points generated and returned.
- objective: pypesto.Objective, optional
- The objective can be used to evaluate the goodness of start points.
- max_n_fval: int, optional
- The maximum number of evaluations of the objective function allowed.
-
pypesto.startpoint.assign_startpoints(n_starts: int, startpoint_method: Callable, problem: pypesto.problem.Problem, options: pypesto.optimize.options.OptimizeOptions) → numpy.ndarray¶ Assign startpoints.
-
pypesto.startpoint.latin_hypercube(**kwargs) → numpy.ndarray¶ Generate latin hypercube points.
-
pypesto.startpoint.uniform(**kwargs) → numpy.ndarray¶ Generate uniform points.
Logging¶
Logging convenience functions.
-
pypesto.logging.log_to_console(level=None)¶ Log to console.
Parameters: level (int) – The output level to use. Default: logging.DEBUG.
-
pypesto.logging.log_to_file(level=None, filename=None)¶ Log to file.
Parameters: - level (int) – The output level to use. Default: logging.DEBUG.
- filename (str) – The name of the file to append to. Default: .pypesto_logging.log.
Release notes¶
0.0 series¶
0.0.13 (2020-05-03)¶
- Tidy up and speed up tests (#265 and others).
- Basic self-implemented Adaptive Metropolis and Adaptive Parallel Tempering sampling routines (#268).
- Fix namespace sample -> sampling (#275).
- Fix covariance matrix regularization (#275).
- Fix circular dependency PetabImporter - PetabAmiciObjective via AmiciObjectBuilder, PetabAmiciObjective becomes obsolete (#274).
- Define AmiciCalculator to separate the AMICI call logic (required for hierarchical optimization) (#277).
- Define initialize function for resetting steady states in AmiciObjective (#281).
- Fix scipy least squares options (#283).
- Allow failed starts by default (#280).
- Always copy parameter vector in objective to avoid side effects (#291).
- Add Dockerfile (#288).
- Fix header names in CSV history (#299).
Documentation:
- Use imported members in autodoc (#270).
- Enable python syntax highlighting in notebooks (#271).
0.0.12 (2020-04-06)¶
- Add typehints to global functions and classes.
- Add PetabImporter.rdatas_to_simulation_df function (all #235).
- Adapt y scale in waterfall plot if convergence was too good (#236).
- Clarify that Objective is of type negative log-posterior, for minimization (#243).
- Tidy up AmiciObjective.parameter_mapping as implemented in AMICI now (#247).
- Add MultiThreadEngine implementing multi-threading aside the MultiProcessEngine implementing multi-processing (#254).
- Fix copying and pickling of AmiciObjective (#252, #257).
- Remove circular dependence history-objective (#254).
- Fix problem of visualizing results with failed starts (#249).
- Rework history: make thread-safe, use factory methods, make context-specific (#256).
- Improve PEtab usage example (#258).
- Define history base contract, enabling different backends (#260).
- Store optimization results to HDF5 (#261).
- Simplify tests (#263).
Breaking changes:
- HistoryOptions passed to pypesto.minimize instead of Objective (#256).
- GlobalOptimizer renamed to PyswarmOptimizer (#235).
0.0.11 (2020-03-17)¶
- Rewrite AmiciObjective and PetabAmiciObjective simulation routine to directly use amici.petab_objective routines (#209, #219, #225).
- Implement petab test suite checks (#228).
- Various error fixes, in particular regarding PEtab and visualization.
- Improve trace structure.
- Fix conversion between fval and chi2, fix FIM (all #223).
0.0.10 (2019-12-04)¶
- Only compute FIM when sensitivities are available (#194).
- Fix documentation build (#197).
- Add support for pyswarm optimizer (#198).
- Run travis tests for documentation and notebooks only on pull requests (#199).
0.0.9 (2019-10-11)¶
- Update to AMICI 0.10.13, fix API changes (#185).
- Start using PEtab import from AMICI to be able to import constant species (#184, #185)
- Require PEtab>=0.0.0a16 (#183)
0.0.8 (2019-09-01)¶
- Add logo (#178).
- Fix petab API changes (#179).
- Some minor bugfixes (#168).
0.0.7 (2019-03-21)¶
- Support noise models in Petab and Amici.
- Minor Petab update bug fixes.
0.0.6 (2019-03-13)¶
- Several minor error fixes, in particular on tests and steady state.
0.0.5 (2019-03-11)¶
- Introduce AggregatedObjective to use multiple objectives at once.
- Estimate steady state in AmiciObjective.
- Check amici model build version in PetabImporter.
- Use Amici multithreading in AmiciObjective.
- Allow to sort multistarts by initial value.
- Show usage of visualization routines in notebooks.
- Various fixes, in particular to visualization.
0.0.4 (2019-02-25)¶
- Implement multi process parallelization engine for optimization.
- Introduce PrePostProcessor to more reliably handle pre- and post-processing.
- Fix problems with simulating for multiple conditions.
- Add more visualization routines and options for those (colors, reference points, plotting of lists of result obejcts)
0.0.3 (2019-01-30)¶
- Import amici models and the petab data format automatically using pypesto.PetabImporter.
- Basic profiling routines.
0.0.2 (2018-10-18)¶
- Fix parameter values
- Record trace of function values
- Amici objective to directly handle amici models
0.0.1 (2018-07-25)¶
- Basic framework and implementation of the optimization
Authors¶
This package was mainly developed by:
- Jan Hasenauer
- Yannik Schälte
- Fabian Fröhlich
- Daniel Weindl
- Paul Stapor
- Leonard Schmiester
- Dantong Wang
- Leonard Schmiester
- Caro Loos
Contact¶
Discovered an error? Need help? Not sure if something works as intended? Please contact us!
- Yannik Schälte: yannik.schaelte@gmail.com
License¶
Copyright (c) 2018, Jan Hasenauer
All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Logo¶
pyPESTO’s logo can be found in multiple variants in the doc/logo directory on github, in svg and png format. It is made available under a creative commons CC0 license. You are encouraged to use it e.g. in presentations and posters.
We thank Patrick Beart for his contribution to the logo.