accept_leftmost_improvement: Alternative acceptance function for multi-dimensional scores...

In designit: Blocking and Randomization for Experimental Design

Alternative acceptance function for multi-dimensional scores in which order (left to right, e.g. first to last) denotes relevance.

Description

Alternative acceptance function for multi-dimensional scores in which order (left to right, e.g. first to last) denotes relevance.

Usage

accept_leftmost_improvement(current_score, best_score, ..., tolerance = 0)

Arguments

current_score	One- or multi-dimensional score from the current optimizing iteration (double or vector of doubles)
best_score	Best one- or multi-dimensional score found so far (double or vector of doubles)
...	Ignored arguments that may be used by alternative acceptance functions
tolerance	Tolerance value: When comparing score vectors from left to right, differences within +/- tol won't immediately shortcut the comparison at this point, allowing improvement in a less important score to exhibit some influence

Value

Boolean, TRUE if current score should be taken as the new optimal score, FALSE otherwise

designit documentation built on May 29, 2024, 12:04 p.m.