mspathDistributedCalculator {mspath}R Documentation

Constructors for mspathCalculator Objects

Description

Generic function to compute the likelihood and or counts of the model described in the calculator under the given parameters. It does so by updating the calculator itself.

Usage

mspathDistributedCalculator(do.what, params, allinits, misc, subject, time, state, qvector, evector, covvec,
                    constrvec, misccovvec, miscconstrvec, baseconstrvec, basemiscconstrvec, 
                    pathvars, pathoffset, pathconstrvec,
                    initprobs, nstates, nintens, nintenseffs, nmisc, nmisceffs, nobs, npts,
                    ncovs, ncoveffs, nmisccovs, nmisccoveffs,
                    npatheffs,
                    isexact, fixedpars, stepnumerator, stepdenominator,
comm=0,
profile = FALSE)

Arguments

See mspath and the documentation on the C++ code for the full details.

do.what Type of computation to perform, as interpreted by the C++ code. This may be ignored, or only apply to the non-distributed calculators created by this one (need to check).
params Ordinarily, the precise values here may be irrelevant, since they can be reset on specific calls to calculate. However, the size of this argument must be correct.
allinits The “initial” values for all fixed and free parameters. Only the fixed values are used; the rest come from params, which may be reset during the life of the calculator.
misc 0 = no misclassification; 1 = full misclassification; 2 = simple, fixed misclassification.
subject The ID, which must be an integer, for each row of data. The ID's should be sorted in ascending order.
time time of each observation, in ascending order within cases.
state The state of each observation. States should be numbered 1 through nstates.
qvector Vectorized matrix of allowed transitions, 1 for allowed, 0 for not allowed.
evector Vectorized matrix of allowed misclassifications (on relevant if do.what is not 0).
covvec vectorized matrix of covariate values.
constrvec constraints for each covariate
misccovvec Vectorized matrix of covariate values for misclassification (only relevant if misc = 1).
miscconstrvec list of constraints for each misclassification covariate
baseconstrvec constraints on baseline transition intensities
basemiscconstrvec constraints on baseline misclassification probabilities
pathvars character vector of the names of history-dependent variables.
pathoffset add this double to every time 0 in the paths
pathconstrvec constraints on path effects on intensities
initprobs initial state occupancy probabilities
nstates number of states
nintens number of intensity parameters
nintenseffs number of distinct intensity parameters
nmisc number of misclassification rates
nmisceffs number of distinct misclassification rates
nobs number of observations in the data set
npts number of individuals/cases in the data set
ncovs number of covariates on transition rates
ncoveffs number of distinct covariate effect parameters
nmisccovs number of distinct misclassification parameters
nmisccoveffs number of distinct misclassification effects
npatheffs number of distinct path (history) effects on transitions
isexact non-0 if we observed time of entry to absorbing states exactly (e.g., death)
fixedpars which parameters to fix
stepnumerator integer This divided by stepdenominator give the maximum step size in the discrete approximation, in the same scale as time.
stepdenominator integer see stepnumerator
comm The MPI communicator to use. The default ordinarily suffices.
profile TRUE to enable profiling.

Value

Returns a new mspathDistributedCalculator object.

Note

The Rmpi library is loaded on object creation, and must be accessible.

Author(s)

Ross Boylan

See Also

mspathDistributedCalculator, mspath, Rmpi


[Package mspath version 0.9-9 Index]