AssumeDeterministic

AssumeDeterministic

is an option for functions such as BayesianMinimization that specifies whether or not the function being considered should be assumed to be deterministic.

Details

  • With AssumeDeterministic->True, the function f being considered is assumed to be deterministic, so that f[conf] always gives the same value for a given configuration conf.
  • With AssumeDeterministic->False, the function f being considered is assumed to be not necessarily deterministic, so that f[conf] can give different values on different evaluations when the configuration conf is the same.
  • The typical default setting is AssumeDeterministic->False.

Examples

open allclose all

Basic Examples  (1)

Minimize a function over a region using BayesianMinimization with the specification that the function is deterministic:

Get the estimated minimum configuration:

Get the estimated minimum value:

For a deterministic function, the model value agrees to good precision with the function value for evaluated configurations:

Scope  (1)

Define a function over a domain defined by a random generator with no specifications:

Minimize the function using BayesianMinimization without any specifications:

Get the minimum configuration and minimum value:

The function is assumed to be stochastic: the model value at the minimum configuration may differ from the function value:

Include information that the function is deterministic:

Get the minimum configuration and minimum value for the deterministic case:

Now the model value and function value at the minimum configuration agree to a good precision:

Introduced in 2016
 (11.0)