Re: GSoC 2014

[Burcin Erocal]

Hi,

(copying pynac-devel in from a lmnd-devel thread about GSoC 2014)

> On Monday, December 2, 2013 12:54:35 PM UTC, Burcin Erocal wrote:
> >
> > * pattern matching in Pynac
> >
> > Try to port Rubi [3] to Sage, see what enhancements are needed
> > in our pattern matching interface and implement them.
> >
> > [3] http://www.apmaths.uwo.ca/~arich/
> >

On Mon, 2 Dec 2013 05:48:44 -0800 (PST)
Volker Braun <vbrau...@gmail.com> wrote:

> That sounds neat! I guess there needs to be some sort of caching when
> matching large numbers of rules. E.g. so that you don't always have
> to walk subtrees to figure out if it is independent of a variable.

Even though our pattern matching interface is probably the most capable
open source / free software implementation available, it still lacks
some features in comparison to MMA.

Rubi can be a great test case, both to determine how to improve the
pattern matching capabilities and to test performance. However, I would
prefer to concentrate on improving the functionality before thinking
about optimization.


Cheers,
Burcin

[Volker Braun]

On Monday, December 2, 2013 4:00:42 PM UTC, Burcin Erocal wrote:

Rubi can be a great test case, both to determine how to improve the
pattern matching capabilities and to test performance. However, I would
prefer to concentrate on improving the functionality before thinking
about optimization. 

Agree, but one should design the interface to be compatible with future optimizations ;-)