sim: Monte Carlo simulation for the filter estimation

In be5tan/rlystop: Early stopping for statistical inverse problems

Description

Performs N simulation runs of the filter estimation procedure for a given design matrix lambda, a given signal mu and a given noise level delta and evaluates the loss and oracle quantities.

Usage

sim(N, lambda, mu, delta, alpha, kappa, filt = c("cutoff", "landw"),
  m0 = 0, alphaLoss = -1)

Arguments

N	Integer number of simulation runs.
lambda	Numeric vector of decreasing, strictly positive entries of the diagonal design matrix.
mu	Numeric vector valued input signal.
delta	Numeric noise level.
alpha	Numeric smoothing index for the residuals.
kappa	Numeric critical stopping value.
filt	Character string designating the filter to be used. filt should be one of "cutoff" or "landw".
m0	Integer cut-off index for the two-step procudure.
alphaLoss	Numeric smoothing parameter for the loss. alpha = -1 gives the strong loss. alpha = 0 gives the weak loss.

Value

Returns a data frame containing N observations of the residual stopping rule and loss.

be5tan/rlystop documentation built on Nov. 22, 2019, 12:05 a.m.