[GiNaC-devel] Volunteer to continue the work on NTL factorisation bindings

[Jens Vollinga]

Hi,

Remco Bloemen schrieb:
> An NTL based implementation would be very powerfull and feasible, as said in 
> [1]. I would like to volunteer to continue this work.
> 
> I have the following questions:
> 
> 1) Where can I obtain the source of this binding?
>    (or should I start from the code in gTybalt, or from scratch?)

the binding is part of the gtybalt sources. So you have to look at that 
code. IIRC the relevant code consists of only few lines of code. It is 
just doing data conversions back and forth and calling NTL in between. I 
am not so sure whether this approach gives a good performance. NTL is 
fast, but the conversions ... Maybe your code is already better.

> 2) Is there any chance such a patch will be included?
>     (this would imply dependecy on NTL)

I don't like the idea of adding another dependency very much. But maybe 
that is the price to pay. For me it depends on the additional 
functionality beyond "simple" Q[x] factorization that would be added to 
GiNaC. Multi-variate factorization, integration, ... that would justify 
the _non-optional_ inclusion, I think.

> 2a) And if it is compile-time optional?

Yes, then I would include such a patch even if it is just a simple 
mono-variate factorization.

> 2b) And if it also included integration of rational functions?
> 2c) And if it also included integration of trancedental functions?

See above.

Regards,
Jens