lik_GaussianPIC {ebTobit}

Helper Function - generate likelihood for pair (L,R) and mean gr

Description

Compute P(L_i, R_i | theta = t_k) for observations (L_i, R_i) and grid of mean t_k.

Usage

lik_GaussianPIC(L, R, gr, s1)

Arguments

L

numeric vector of lower bounds

R

numeric vector of upper bounds

gr

numeric vector of means

s1

numeric vector of standard deviations

Value

the likelihood under partial interval censoring

Examples

# set-up
p = 15
gr = stats::rnorm(p)
L = R = stats::rnorm(p)
missing.idx = sample.int(n = p, size = p/5)
L[missing.idx] = L[missing.idx] - stats::runif(length(missing.idx), 0, 1)
R[missing.idx] = R[missing.idx] + stats::runif(length(missing.idx), 0, 1)

# R solution
lik = prod(ifelse(
           L == R,
           stats::dnorm(L-gr),
           stats::pnorm(R-gr) - stats::pnorm(L-gr)))

# Compare R to RcppParallel method
all.equal(lik, lik_GaussianPIC(L, R, gr, rep(1,p)))