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, startpoint_resample: bool)numpy.ndarray[source]

Assign start points.

pypesto.startpoint.latin_hypercube(**kwargs)numpy.ndarray[source]

Generate latin hypercube points.

Parameters
  • n_starts – number of starting points to be sampled.

  • lb – lower bound.

  • ub – upper bound.

  • smooth – indicates if a (uniformly chosen) random starting point within the hypercube [i/n_starts, (i+1)/n_starts] should be chosen (True) or the midpoint of the interval (False). Default is True.

pypesto.startpoint.uniform(**kwargs)numpy.ndarray[source]

Generate uniform points.