Sampling backends

Many optimization algorithms require the gradient of the objective function. These are estimated using evaluations of perturbed controls, which requires (pseudo-)stochastic sampling. By default these perturbations are generated using sampling code from the SciPy package. SciPy provides sampling from common distributions such as Gaussian, Uniform and Bernoulli, and some additional methods such as Sobol and Latin hypercube sampling.

The sampling method and options are specified in the sampler subsection of the controls configuration settings. The sampling method is selected using the method keyword. If the method keyword is missing, a normal distribution is used.

The sampling methods in the SciPy backend support several options that can be passed using the options keyword. Consult the online SciPy manual on Statistical functions for details. To find the algorithms and options that are supported in Everest, consult the relevant section of the ropt manual: SciPy Sampler Plugin.

Example

optimization:
    sampler:
        method: sobol
        options:
            scramble: False