smooth_dec: Decreasing first-order smoothing rule

In noisyCE2: Cross-Entropy Optimisation of Noisy Functions

Description

Decreasing smoothing rule

x_{t+1}:=a_t\,x_t + (1-a_t)\,x_{t-1}

where

a_t:= b\,≤ft(1-≤ft(1-\frac{1}{t}\right)^q\right)

for some 0.7≤q b≤q1 and some 5≤q q≤q10.

Usage

smooth_dec(x, xt, b, qu)

Arguments

x	numeric value of the last value of the parameter.
xt	numeric vector of past values of the parameter (time series).
b	smoothing parameter b.
qu	smoothing parameter q.

Value

A numeric vector of updated parameters.

See Also

Other smoothing rules: smooth_lin()

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