momentuHMM (version 1.5.5)

n2w: Scaling function: natural to working parameters.

Description

Scales each data stream probability distribution parameter from its natural interval to the set of real numbers, to allow for unconstrained optimization. Used during the optimization of the log-likelihood. Parameters of any data streams for which a design matrix is specified are ignored.

Usage

n2w(par, bounds, beta, delta = NULL, nbStates, estAngleMean, DM, Bndind, dist)

Value

A vector of unconstrained parameters.

Arguments

par

Named list of vectors containing the initial parameter values for each data stream.

bounds

Named list of 2-column matrices specifying bounds on the natural (i.e, real) scale of the probability distribution parameters for each data stream.

beta

List of regression coefficients for the transition probabilities.

delta

Initial distribution. Default: NULL ; if the initial distribution is not estimated.

nbStates

The number of states of the HMM.

estAngleMean

Named list indicating whether or not to estimate the angle mean for data streams with angular distributions ('vm' and 'wrpcauchy').

DM

An optional named list indicating the design matrices to be used for the probability distribution parameters of each data stream. Each element of DM can either be a named list of linear regression formulas or a matrix.

Bndind

Named list indicating whether DM is NULL with default parameter bounds for each data stream.

dist

A named list indicating the probability distributions of the data streams.

Examples

if (FALSE) {
m<-example$m
nbStates <- 2
nbCovs <- 2
parSize <- list(step=2,angle=2)
par <- list(step=c(t(m$mle$step)),angle=c(t(m$mle$angle)))
bounds <- m$conditions$bounds
beta <- matrix(rnorm(6),ncol=2,nrow=3)
delta <- c(0.6,0.4)

#working parameters
wpar <- momentuHMM:::n2w(par,bounds,list(beta=beta),log(delta[-1]/delta[1]),nbStates,
m$conditions$estAngleMean,NULL,m$conditions$Bndind,
m$conditions$dist)

#natural parameter
p <-   momentuHMM:::w2n(wpar,bounds,parSize,nbStates,nbCovs,m$conditions$estAngleMean,
m$conditions$circularAngleMean,lapply(m$conditions$dist,function(x) x=="vmConsensus"),
m$conditions$stationary,m$conditions$fullDM,
m$conditions$DMind,1,m$conditions$dist,m$conditions$Bndind,
matrix(1,nrow=length(unique(m$data$ID)),ncol=1),covsDelta=m$covsDelta,
workBounds=m$conditions$workBounds)
}