score.fin: Gradient of the variational negative log-likelihood with...

In wrbrooks/cox: Estimate Cox process regression models

Description

Gives the gradient of the variational negative log-likelihood with respect to coefficients beta, precision component ltau, and mean vector of the variational approximation, M. Within the cox package, this function is used to calculate gradients within a call to optim after the variational approximation has converged, as part of the process of calculating the Hessian.

Usage

score.fin(par, y, X, S, wt, V)

Arguments

par	vector composed of beta, M, and ltau
y	vector of response values
X	matrix of fixed effect covariates
S	design matrix for the spatial random effects
wt	vector of observation weights
V	converged covariance matrix for the random effects, estimated by a variational approximation

Value

vector of gradients of the variational negative log-likelihood with respect to par

wrbrooks/cox documentation built on May 4, 2019, 11:58 a.m.