Sorry to have been slow responding here - I just noticed this question wi th your reply, Chris!
The svd routine in matrixops is an iterative-fit algorithm originally written by Bryant Marks (is I remember right - it is in the POD). It is not particularly threadsafe. MatrixOps in general is intended as a lightweight, minimal implementation of some commonly useful algorithms - but for serious work (involving, say, 10^6 or more 2x2 matrices -- or one matrix that is more than about 15x15) it is well worth the effort to use the slatec or gsl modules. Sent from my iPad > On Dec 29, 2014, at 8:09 AM, Chris Marshall <[email protected]> wrote: > > svd() calls a C routine SVD() which is defined in the svd.c file in > PDL/Basic/MatrixOps/ > I don't know about the status w.r.t. automatic parallelization of the thread > loops. > > --Chris > >> On Sat, Dec 27, 2014 at 11:33 PM, mraptor <[email protected]> wrote: >> I meant Singular Vector Decomposition >> -------| http://ifni.co >> >> >> On Sat, Dec 27, 2014 at 11:24 PM, mraptor <[email protected]> wrote: >> > Hi, >> > >> > Where can I find how exactly svd() works (PDL::MatrixOps), cant seem >> > to find it ? >> > it is just alias to PDL::svn() >> > >> > Also it does not seem to support ParallelCPU, or I'm doing something wrong >> > !! >> > >> > thanks >> > >> > -------| http://ifni.co >> >> _______________________________________________ >> Perldl mailing list >> [email protected] >> http://mailman.jach.hawaii.edu/mailman/listinfo/perldl > > _______________________________________________ > Perldl mailing list > [email protected] > http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
