InitialEvaluationHistory
is an option for functions such as BayesianMinimization that gives an initial set of configurations and values.
Details
- In the case of BayesianMinimization and BayesianMaximization, the configurations and values given by InitialEvaluationHistory correspond to inputs and outputs of the objective function.
- Possible settings for InitialEvaluationHistory include:
-
None no evaluation history {conf1val1,conf2val2,…} list of configurations associated to their values {conf1,conf2,…}{val1,val2,…} rule associating configurations to their respective values {<"Configuration"conf1, "Value"val1 >,… } list of associations with configurations and values - InitialEvaluationHistory can also be given as a Dataset object.
Examples
Basic Examples (1)
Minimize a function using BayesianMinimization over a domain defined by a random generator by specifying an initial list of configurations associated to their values:
Specify initial evaluation history as a rule associating configurations to their respective values:
Specify initial evaluation history as a list of associations containing configurations and their values:
The initial evaluation history can also be specified as a Dataset of any of the above formats:
Text
Wolfram Research (2016), InitialEvaluationHistory, Wolfram Language function, https://reference.wolfram.com/language/ref/InitialEvaluationHistory.html.
CMS
Wolfram Language. 2016. "InitialEvaluationHistory." Wolfram Language & System Documentation Center. Wolfram Research. https://reference.wolfram.com/language/ref/InitialEvaluationHistory.html.
APA
Wolfram Language. (2016). InitialEvaluationHistory. Wolfram Language & System Documentation Center. Retrieved from https://reference.wolfram.com/language/ref/InitialEvaluationHistory.html