# [R] a nlm() question

Jiang Peng jp021 at sjtu.edu.cn
Thu Oct 30 09:49:00 CET 2008

  Dear R listers,
I have a very annoying problem using nlm().
I want to find the minimizer of my target function, if written in
\LaTeX is
f(\mu1,\mu2,\sigma1,\sigma2) = \sum_i^n( w_ig_t(z_i) ), where
g_t(z) is a pdf of  bivariate normal distribution and z_i is my samples.
I cannot get the estimation result generated by nlm(), and I got
the following errors

"
Error in nlm(foo, theta.start, hessian = TRUE) :
non-finite value supplied by 'nlm'
"

from time to time( If I ran the code second time , it does not have
the error .) I am new to R so I don't know if there is any debug
method of R code. i hope somebody can help me out .

------------------------
Jiang Peng, Ph.D. Candidate
Department of mathematics &
Antai college of Economics and management
Shanghai Jiao Tong University
Address: Room 127, #3 Building, Xuhui Campus
E-mail: jp021 at sjtu.edu.cn
jp021 at 126.com
MSN: jp021 at hotmail.com
iChat: mathfrog at mac.com
Mobile: +86-138 168 780 95