nidaarem: Nesterov Initialized DAAREM

In nchenderson/nidaarem: Nesterov Initialized Damped Anderson Acceleration with Epsilon Monotonicity

Description

Performs the DAAREM acceleration scheme with Nesterov initialization and alternating monotonicity

Usage

nidaarem(par, fixptfn, objfn, ..., control = list())

Arguments

par	A vector of starting values.
fixptfn	A vector function that performs the fixed-point mapping.
objfn	A scalar function that evaluates the objective function to maximized.
...	Arguments passed to fixptfn and objfn
control	A list of control parameters specifying any changes to default values of algorithm control parameters. Full names of control list elements must be specified, otherwise, user-specifications are ignored. See *Details*.

Value

A list with the following components:

par	Parameter, x* that are the fixed-point of G such that x* = G(x*), if convergence is successful.
value.objfn	The value of the objective function $L$ at termination.
fpevals	Number of times the fixed-point function fixptfn was evaluated.
objfevals	Number of times the objective function objfn was evaluated.
convergence	An integer code indicating type of convergence. 0 indicates successful convergence, whereas 1 denotes failure to converge.
objfn.track	A vector containing the value of the objective function at each iteration.

Author(s)

Nicholas Henderson

Examples

PP <- matrix(rnorm(9), nrow=3, ncol=3)
PP[c(3,4,7,8)] <- NA
ind <- which(!is.na(PP))

A <- PP[ind]
PA <- PP
PA[-ind] <- 0.0

par0 <- rnorm(9)

matcompl.test <- nidaarem(par=par0, fixptfn=MatrixCompleteUpdate, objfn=MatrixCompleteObj, 
                          lambda=2, A=A, PA=PA, ind=ind)

nchenderson/nidaarem documentation built on Feb. 19, 2020, 12:45 p.m.