MSTEP: Execute a Newton-Raphson step within M-step of the EM...

In jasonxu90/bdsem: Analysis of discretely observed Birth-Death-Shift processes

Description

MSTEP executes one iteration of a Newton-Raphson algorithm as part of the maximization (M-step) of the EM algorithm. Given the matrix of expected sufficient statistics returned by ESTEP, this function uses closed form gradient and hessian expressions to efficiently optimize the current settings of the coefficients beta. This is called up to 10 times per M-step within EM.run

Usage

  MSTEP(matrix, betaVec, num.patients, patients.design)

Arguments

matrix	A matrix in the format returned by ESTEP or ESTEP.slow
betaVec	A vector of regression coefficients
num.patients	An integer, the number of unique patients
patients.design	A design matrix in the format generated by PatientDesignExample

Value

An updated coefficient vector after one Newton-Raphson step

jasonxu90/bdsem documentation built on May 18, 2019, 5:54 p.m.