NewtonRaphsonControl: Controls for Newton-like optimizer

In nspope/radish: Fast gradient-based optimization of resistance surfaces

Description

Control settings for Newton and quasi-Newton algorithms

Usage

NewtonRaphsonControl(
  maxit = 100,
  ctol = sqrt(.Machine$double.eps),
  ftol = sqrt(.Machine$double.eps),
  etol = 10 * .Machine$double.eps,
  verbose = FALSE,
  eps = 1e-08,
  del = 1,
  ls.control = HagerZhangControl()
)

Arguments

maxit	Maximum number of Newton steps
ctol	Convergence tolerance in the gradient (max(abs(grad)) < ctol)
ftol	Convergence tolerance in the objective (f - fold < ftol)
etol	Maximum spectral gap for a eigenvalue to be considered too close to singular
verbose	Print progress to stdout
eps	For box-constrained optimization, how close to a boundary can be considered "on the boundary"
del	For quasi-Newton methods, a typical step size used to initialize the approximate Hessian
ls.control	Control settings for the line search

nspope/radish documentation built on July 12, 2020, 11:50 a.m.