bilby.core.sampler.proposal.DifferentialEvolution

class bilby.core.sampler.proposal.DifferentialEvolution(sigma=0.0001, mu=1.0, priors=None)

Bases: JumpProposal

__init__(sigma=0.0001, mu=1.0, priors=None)

Differential evolution step. Takes two elements from the existing coordinates and differentially evolves the old sample based on them using some Gaussian randomisation in the step.

Parameters:

sigma: float, optional

Random spread in the evolution step. Default is 1e-4

mu: float, optional

Scale of the randomization. Default is 1.0

__call__(sample, **kwargs)

A generic wrapper for the jump proposal function

Parameters:

args: Arguments that are going to be passed into the proposal function

kwargs: Keyword arguments that are going to be passed into the proposal function

Returns:

dict: A dictionary with the new samples. Boundary conditions are applied.

Methods

__init__([sigma, mu, priors])

Differential evolution step.