GIST: GIST

In jhorzek/psydiff: panel stochastic differential equations

GIST

Description

General Iterative Shrinkage and Thresholding Algorithm based on Gong, P., Zhang, C., Lu, Z., Huang, J., & Ye, J. (2013). A General Iterative Shrinkage and Thresholding Algorithm for Non-convex Regularized Optimization Problems. In S. Dasgupta & D. McAllester (Eds.), Proceedings of Machine Learning Research (PMLR; Vol. 28, Issue 2, pp. 37–45). PMLR. http://proceedings.mlr.press

Usage

GIST(
  model,
  startingValues,
  lambda,
  adaptiveLassoWeights,
  regularizedParameters,
  eta = 1.5,
  sig = 0.2,
  initialStepsize = 1,
  stepsizeMin = 0,
  stepsizeMax = 999999999,
  GISTLinesearchCriterion = "monotone",
  GISTNonMonotoneNBack = 5,
  maxIter_out = 100,
  maxIter_in = 100,
  break_outer = 1e-08,
  numDeriv = FALSE,
  verbose = 0,
  silent = FALSE
)

Arguments

model	model
startingValues	named vector with starting values
lambda	penalty value
adaptiveLassoWeights	named vector with adaptive lasso weights
regularizedParameters	named vector of regularized parameters
eta	if the current step size fails, eta will decrease the step size. Must be > 1
sig	GIST: sigma value in Gong et al. (2013). Sigma controls the inner stopping criterion and must be in (0,1). Generally, a larger sigma enforce a steeper decrease in the regularized likelihood while a smaller sigma will result in faster acceptance of the inner iteration.
initialStepsize	initial stepsize to be tried in the outer iteration
stepsizeMin	Minimal acceptable step size. Must be > 0. A larger number corresponds to a smaller step from one to the next iteration. All step sizes will be computed as described by Gong et al. (2013)
stepsizeMax	Maximal acceptable step size. Must be > stepsizeMin. A larger number corresponds to a smaller step from one to the next iteration. All step sizes will be computed as described by Gong et al. (2013)
GISTLinesearchCriterion	criterion for accepting a step. Possible are 'monotone' which enforces a monotone decrease in the objective function or 'non-monotone' which also accepts some increase.
GISTNonMonotoneNBack	in case of non-monotone line search: Number of preceding regM2LL values to consider
maxIter_out	maximal number of outer iterations
maxIter_in	maximal number of inner iterations
break_outer	stopping criterion for the outer iteration.
numDeriv	boolean should numDeriv package be used for derivatives?
verbose	set to 1 to print additional information and plot the convergence

Details

GIST minimizes a function of form f(theta) = l(theta) + g(theta), where l is the likelihood and g is a penalty function. Various penalties are supported, however currently only lasso and adaptive lasso are implemented.

jhorzek/psydiff documentation built on Sept. 15, 2022, 6:26 a.m.