Hi Jacek, Very interesting work! I encourage people to comment on Jacek's points. I know several people prototyping/using GSL. Any progress in the interface with this library or others? It would be nice if somebody could report on this topic at the coming ROOT workshop at FNAL. Rene Brun Jacek M. Holeczek wrote: > > > How are the GSL wrappers coming along? > Well, I did not say I intend to go with the GSL. > I admit, however, that I currently look at it (they provide RH i386.rpm). > The first step is to get a GSL-aware Complex class working. This is almost > done (the first known problem is that sizeof(long double) is equal 24 in > g++, and 16 in root/cint, but I can live with it). > If I succeed with the Complex, I will (possibly) try to get the Block, > Vector and Matrix working. No promises. > (BTW. There is a set of C++ wrappers to (some) GSL Special Functions > available : > http://www.fnal.gov/docs/working-groups/fpcltf/Pkg/SpecialFunctions/doc/html/0SpecialFunctions.html > http://cdfcodebrowser.fnal.gov/CdfCode/source/SpecialFunctions/ > .) > The question of a linear algebra package in ROOT was already raised a > couple of times without any conclusion. I asked the original question > about Complex Matrixes and Complex Vectors in order to find someone who > already made some progress in it, but ... there is nobody. > I need to provide a working (complex) solution to some people in the next > couple of days. If I get bored with GSL I will try to take another, > already in C++, library (on the cost of no root/cint interactive support). > I considered TNT (http://math.nist.gov/tnt/) which is a nice link to > LAPACK + BLAST (again there are RH rpms for LAPACK and BLAST). Both LAPACK > and BLAST can be compiled as shared libraries, but they link to libg2c.a > which does not exist in "shared" form, so I gave it up. > I think, I will choose Blitz++ (http://oonumerics.org/blitz/), unless > someone gives me a better hint. > Best regards, > Jacek. > P.S. A list of available packages can be found on : > http://www.openscience.org/links.php?section=103 > or another list on : > http://oonumerics.org/oon/ > Jacek.
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:45 MET