stepsize_schedule: Stepsize schedule
In ngme2: Linear Latent Non-Gaussian Models with Flexible Distributions

stepsize_schedule

R Documentation

Stepsize schedule

Description

Stepsize schedule

Usage

stepsize_schedule(
  method = c("constant", "poly"),
  alpha = 0.501,
  t0 = 1,
  burnin_iter = 0
)

Arguments

`method`	schedule type. "constant" keeps `\eta_t = \eta_0` and "poly" uses polynomial decay.
`alpha`	polynomial exponent used when `method = "poly"`. Must satisfy `1/2 < \alpha < 1`.
`t0`	non-negative offset in iteration index.
`burnin_iter`	non-negative integer. During these initial iterations, schedule scaling is fixed to 1. Afterward, polynomial decay is applied with reset local time index.

Details

Define an iteration-dependent stepsize schedule for use in control_opt(). This schedule is independent of stepsize_decay() and applies a direct multiplicative scaling by iteration index:

\eta_t = \eta_0 (t + t_0)^{-\alpha}