automultinomial (version 1.0.0)

centeredLogLikGrad: Centered model gradient

Description

Computes a gradient for the centered model, taking coefficients, a design matrix, a response matrix, and a list of adjacency matrices

Usage

centeredLogLikGrad(theta, X, z, A, constraint, thetaInds, innerIndices = NULL)

Arguments

theta

a matrix of coefficients with dimension (p+(k-1)*length(A)) by k; frequently A will have length 1

X

nxp design matrix

z

nxk response matrix

A

a list of adjacency matrices

constraint

constraint on eta

thetaInds

relevant indices

innerIndices

only here to match argument list with uncentered case

Value

a gradient vector

Examples

# NOT RUN {
#g=centeredLogLikGrad(theta,X,z,A)
# }