Based on a surrogate model, the acquisition function encodes the preference to evaluate a new point for evaluation.

## Format

R6::R6Class object.

## Construction

acqf = AcquisitionFunction(id, opt_dir, settings, requirements)

• id :: character(1)
Short name of the criterion.

• opt_dir :: character(1)
Should the criterion be minimized of maximized? Can be "min", "max" or "obj", where the latter means that the same direction

• settings :: named list
Control settings and constants. as specified in the objective function is taken.

• requirements :: named list

## Methods

(list of mlr3::Measure, logical(1), logical(1), logical(1), logical(1)) -> data.table::data.table()

• eval_batch(dt)
data.table::data.table() -> numeric(1)
Evaluates all design points in dt with the acquisition function where each points is a row, and columns are scalar parameters.