This is the mail archive of the gsl-discuss@sources.redhat.com mailing list for the GSL project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: gsl_multimin question

Hi.

Could you show the code? Are you sure the derivatives are correct? One of my collegue is using the multimin code for neural network learning and he has sometime more than one hundred parameters, some I almost sure the number of parameter is not an issue.

Fabrice

Rajarshi Guha wrote:

Hi,
  I've been trying to use the gsl_multimin* functions to optimize a set
of coordinates. I'm using the BFGS method but after the first iteration
the argument vector  to the function to be minimized becomes NAN. The
function that I am minimizing takes 17 arguments in my current case but
it could take more. Would a large number of parameters affect the
functioning of the minimization function?

The values that are being passed to the function are in the range of 0 -
20 so I'm not sure where a NAN is being generated.

Any pointers would be appreciated



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]