Displaying 1 result from an estimated 1 matches for "lik1b".
2007 Oct 24
1
vectorized mle / optim
.../ optim lead to an
annoying bug: the "minuslogl" function cannot have "computed" default
parameters, because all of them must be specified either as "fixed" or
"start".
-------------------------------- example 3 ------------ 8<
----------------------
lik1b <- function(m, v=var(data), data) {
N <- length(data)
lik.mean <- dnorm(mean(data), m, sqrt(v/N), log=T)
lik.var <- dchisq(N*var(data)/v, N-1, log=T)
return(-lik.mean - lik.var)
}
ml.result.1 <- mle(lik1b, start=list(m=2, v=2), fixed=list(data=data))
ml.result.2 <- mle(lik...