ebTobit (version 1.0.2)

likMat: Helper Function - generate likelihood matrix

Description

Compute a matrix L whose entries are L[i,k] = P(L_i, R_i | theta = t_k) for observations (L_i, R_i) and grid of means t_k.

Usage

likMat(L, R, gr, s1)

Value

the n x m likelihood matrix under partial interval censoring

Arguments

L

n x p matrix of lower bounds

R

n x p matrix of upper bounds

gr

m x p matrix of candidate means

s1

n x p matrix of standard deviations

Examples

# set-up
n = 100; m = 50; p = 5
gr = matrix(stats::rnorm(m*p), m, p)
L = R = matrix(stats::rnorm(n*p), n, p)
s1 = matrix(1, n, p)
missing.idx = sample.int(n = n*p, size = p*p)
L[missing.idx] = L[missing.idx] - stats::runif(p, 0, 1)

# R solution
lik = matrix(nrow = n, ncol = m)
for (i in 1:n) {
    for(k in 1:m) {
        lik[i,k] = prod(ifelse(
            L[i,] == R[i,],
            stats::dnorm(L[i,]-gr[k,], sd = s1[i,]),
            stats::pnorm(R[i,]-gr[k,], sd = s1[i,]) - stats::pnorm(L[i,]-gr[k,], sd = s1[i,])
        ))
    }
}

# Compare R to RcppParallel method
all.equal(lik, likMat(L, R, gr, s1))