sbgo

Use the surrogate based optimization method

Specification

• Alias: None

• Arguments: None

Child Keywords:

Required/Optional	Description of Group	Dakota Keyword	Dakota Keyword Description
Optional		gaussian_process	Gaussian Process surrogate model
Optional		use_derivatives	Use derivative data to construct surrogate models
Optional		import_build_points_file	File containing points you wish to use to build a surrogate
Optional		export_approx_points_file	Output file for surrogate model value evaluations

Description

A surrogate-based optimization method will be used. The surrogate employed in sbo is a Gaussian process surrogate.

The main difference between ego and the sbo approach is the objective function being optimized. ego relies on an expected improvement function, while in sbo, the optimization proceeds using an evolutionary algorithm ( coliny_ea) on the Gaussian process surrogate: it is a standard surrogate-based optimization. Also note that the sbo option can support optimization over discrete variables (the discrete variables are relaxed) while ego cannot.

This is not the same as surrogate_based_global.