Actually.. my original idea is to parse matlab/r code and generate legal C++ code with this library http://eigen.tuxfamily.org/index.php?title=Main_Page#Documentation
<http://eigen.tuxfamily.org/index.php?title=Main_Page#Documentation>and use pypy to do some kind of type/shape inference on matrix object On Thu, Dec 16, 2010 at 5:21 PM, William ML Leslie < [email protected]> wrote: > On 17 December 2010 09:43, René Dudfield <[email protected]> wrote: > > [Numexpr] approach, could be applied in the pypy jit I guess. > > There was a lot of discussion in this direction during micronumpy > experiments, but I'm not sure how well ideas were documented. > > A similar class of optimisation is polyhedral loop refactoring; > implementing it within the framework of a tracing JIT would be new and > exciting research if anyone is looking for a topic. > > -- > William Leslie >
_______________________________________________ [email protected] http://codespeak.net/mailman/listinfo/pypy-dev
