Optim() maximum number of parameters

Hello,

I'm trying to maximize a likelihood function for a HMM with the optim() 
function using the Nelder-Mead Method. The LLF has 20 Parameters which 
are to be estimated. We found out that R changes some variables but not 
all of them, especially the last 2 parameters aren't changed. I already 
checked if I assign this parameters vales within the function but this 
is not the case.

Is there a upper limit for the number parameters in optim()? Do you have 
any experience with this kind of problem?

Many thanks

Best regards
Rainer Haidinger