solve_fista_goertler: solve_fista_goertler

In MarianSchoen/DTD: Digital Tissue Deconvolution

solve_fista_goertler

Description

solve_fista_goertler

Usage

solve_fista_goertler(
  model,
  lambda = 0.01,
  maxiter = 100,
  stop.crit.threshold = 1e-05,
  save.all.tweaks = FALSE,
  learningrate = NA,
  linesearchspeed = 2,
  cycles = 5,
  restarts = TRUE,
  verbose = FALSE
)

Arguments

model	input model, as constructed by, e.g., empty_model()
lambda	regularization parameter lambda
maxiter	maximum number of iterations
stop.crit.threshold	stopping criterion: if loss functions falls by less than this value, stop optimization.
save.all.tweaks	if set to true, keep results after every iteration
learningrate	initial learning rate (if set to NA, learning rate is determined automatically)
linesearchspeed	rate of in- or decrease of learning rate
cycles	number of function evaluations per step.
restarts	if set to true, restart nesterov step width if loss fn is increased.

Value

list, containing an element "Tweak" with the result vector and, if save.all.tweaks is set to TRUE, "History", containing tweaks at every step.

MarianSchoen/DTD documentation built on April 29, 2022, 1:59 p.m.