nmm (version 0.9)

grad_hess_eval: grad_hess_eval forms function of gradient and Hessian of log-likelihood produced by f_create.

Description

grad_hess_eval forms function of gradient and Hessian of log-likelihood produced by f_create.

Usage

grad_hess_eval(mn, parnl, hessian = FALSE, fixed = 0, data = "", cheqs0 = NULL)

Arguments

mn

Expression, can be a list of equations.

parnl

Names of parameters.

hessian

if TRUE, returns hessian function, otherwise gradient.

fixed

Integer, which parameter is fixed to be 0.

data

Name of the data frame with which the function will be evaluated.

cheqs0

If continuous are supplied, include the expressions of errors.

Value

A function for evaluation of gradient or Hessian.

Examples

# NOT RUN {
eq_d <- c("ASC1 * 1 + B11_dur * dur_1" , "ASC2 * 1 + B12_dur * dur_2",
"ASC3 * 1 + B13_dur * dur_3 + B20_cost * cost_3 + B53_parkman * PbAvl_3",
"ASC4 * 1 + B14_dur * dur_4 + B20_cost * cost_4 + B34_serv * servIdx_4 + B44_stop * stopUs1R1_4")
parl <- c(paste0("ASC", 1:4), paste0("B1", 1:4, "_dur"), "B20_cost", "B53_parkman", "B34_serv",
 "B44_stop")
disc_par <- get_par(parl, eq_d)
ffor <- disc_par$cheqs0
parld <- disc_par$parld
res <- MNlogitf(ffor, separatenmm=FALSE, transform=FALSE)
parnl <- paste0("par", 1:length(parld))
gf <- grad_hess_eval (res, parnl, data="data", fixed=1)
hf <- grad_hess_eval (res, parnl, data="data", fixed=1, hessian=TRUE)
# }