I am trying to optimize a likelihood function using NLMINB. After running without a problem for quite a few iterations (enough that my intermediate output extends further than I can scroll back), it tries a vector of parameter values NaN. This has happened with multiple Monte Carlo datasets, and a few different (but very similar) likelihood functions. (They are complicated, but I can send them to someone if desired.) Is this something that can happen with NLMINB, perhaps because of a 0/0 in the gradient calculations? Or is it unique to my code? Thanks in advance! Rebecca -- Rebecca Sela Statistics Department/IOMS Stern School of Business New York University
NaN as a parameter in NLMINB optimization
2 messages · Rebecca Sela, Duncan Murdoch
On 21/12/2007 12:07 AM, Rebecca Sela wrote:
I am trying to optimize a likelihood function using NLMINB. After running without a problem for quite a few iterations (enough that my intermediate output extends further than I can scroll back), it tries a vector of parameter values NaN. This has happened with multiple Monte Carlo datasets, and a few different (but very similar) likelihood functions. (They are complicated, but I can send them to someone if desired.) Is this something that can happen with NLMINB, perhaps because of a 0/0 in the gradient calculations? Or is it unique to my code?
I would say it's a bug. I can't tell if it's a bug in your code or in the nlminb code, but even if it's a problem with your code, it would be nice if nlminb could recognize it and give a error report rather than passing garbage to your function. Can you put together a small enough example to post here? If not, you could send me the full example, but I may not get to it for a couple of weeks. In either case, try to make sure the error occurs on every run by using set.seed() and/or a fixed dataset. If the problem is intermittent even with that, please let us know. Duncan Murdoch