ts_change: Time series change stopping rule

In noisyCE2: Cross-Entropy Optimisation of Noisy Functions

Description

Deterministic stopping rule based on the last change in the value of γ_n. Changes smaller than tol, or relative changes smaller than reltol stop the algorithm. This criterion is suitable only in case of deterministic objective functions.

Usage

ts_change(x, reltol = 1e-04, tol = 1e-12)

Arguments

x	numeric vector of last γ_n values, as selected by the function passed to noisyCE2() through the argument stopwindow.
reltol	relative changes smaller than tol stop the algorithm.
tol	changes smaller than tol stop the algorithm.

Value

A numeric indicating whether the algorithm has converged:

0	the algorithm has converged.
1	the algorithm has not converged.

See Also

Other stopping rules: geweke()

noisyCE2 documentation built on Nov. 9, 2020, 5:13 p.m.