R/mlnormal_eval_priors_derivative2.R

In LAM: Some Latent Variable Models

## File Name: mlnormal_eval_priors_derivative2.R
## File Version: 0.06

mlnormal_eval_priors_derivative2 <- function( pars, prior, h )
{
    #--- adjust numerical differentiation parameter
    hvec <- mlnormal_adjust_numdiff_parameter( h=h, pars=pars  )
    #--- evaluate pars
    x0 <- mlnormal_eval_priors( pars=pars, prior=prior )
    x1 <- mlnormal_eval_priors( pars=pars + hvec, prior=prior )
    x2 <- mlnormal_eval_priors( pars=pars - hvec, prior=prior )
    #--- compute gradient and second derivative
    der <- (x1-x2)/ (2*hvec)
    info <- ( x1 + x2 - 2*x0) / hvec^2
    infomat <- diag(info)
    res <- list( der=der, info=info,  infomat=infomat )
    return(res)
}