> Hi, > > there seems to be a limit of 50 parameters that can be fitted at a time. > Is there any way to circumvent this? Is this a limit of MINUIT or the > PDL implementation? > > Thank you! > > Cheers, > > Rahman
This is a limitation of the fortran version of Minuit; I ran into a similar problem with the Minuit included in the XSPEC fitting package. Specifically, there are a number of arrays which have hard-coded dimensions. These arrays are passed around via FORTRAN common blocks, so they have to have static dimensions. In minuit.f from the http://www.eso.org/~ajordan/PDL-Minuit-1.0.tgz tarball, these are defined as parameter (mne=100 , mni=50) parameter (mnihl=mni*(mni+1)/2) A number of arrays in common blocks mn7, mn7ext, mn7err, etc., are dimensioned to have size mni (or mne or mnihl). Note that in minuit.f, these declarations are in all the subroutines which use the relevant common blocks, so you can't change the declaration in just one place. (In the version of Minuit in XSPEC, the mni, mne, and mnihl declarations and the common blocks are in their own include file, so I was able to make a single patch to increase the array sizes.) There is a C++ version of Minuit at http://seal.web.cern.ch/seal/work-packages/mathlibs/minuit/ Apparently version 1.7.9 is the last version with the old C++ API. Minuit2 (also C++) is part of the CERN ROOT package. There seems to be a standalone version at http://seal.web.cern.ch/seal/work-packages/mathlibs/minuit/ Presumably the C++ versions uses dynamic memory allocation to build arrays of the required size. ROOT is generally under the GNU LESSER GENERAL PUBLIC LICENSE, Version 2.1, February 1999. However, in the Minuit2 distribution, I don't see a LICENSE file, and the individual files contain a notice "Copyright (c) 2005 LCG ROOT Math team, CERN/PH-SFT" Terry Gaetz > On Thu, 2007-06-21 at 14:48 +0200, Andres Jordan wrote: > > I've included an interface to Minuit as a new module PDL::Minuit in the > > PDL distribution, so if you get fresh source from the cvs you can get it. > > > > Alternatively for those who would like to try this without recompiling > > PDL at this point, I have put a standalone tarball at: > > > > http://www.eso.org/~ajordan/PDL-Minuit-1.0.tgz > > > > Enjoy (assuming the compilation goes smoothly....). > > > > -Andres > > [...] _______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
