Thanks for the reply Ondrej.

On Thu, Mar 6, 2014 at 5:46 AM, Ondřej Čertík <[email protected]>wrote:

> On Wed, Mar 5, 2014 at 4:25 PM, Thilina Rathnayake
> <[email protected]> wrote:
> > Hi All,
> >
> > I wish to apply for in this year's GSoC as well and I am interested in
> doing
> > a project with CSymPy. I worked with Ondrej on implementing basic number
> > theoretic functionalities for CSymPy during the last few months. Below
> are
> > some PRs related to above work.
> >
> > https://github.com/certik/csympy/pull/99
> > https://github.com/certik/csympy/pull/103
> >
> > IMO what we need to do with CSymPy right now is to follow the 80-20
> > principle, i.e. implement that 20% of functionality that will be used in
> 80%
> > of the use cases. After that we can gradually implement the additional
>
> Precisely.
>
> > functionalities as we wish. Ondrej has already done a great job in
> > implementing the required foundation to do this.
> >
> > With this idea in mind, the first two projects (Implementing elementary
> > functions and Implementing series expansion) listed in the ideas page
> under
> > CSymPy looked interesting to me. I hope to study these two projects in
> > detail
> > during the next few days and discuss with the group on how these should
> be
> > carried out.
> >
> > Also, What functionalities listed here (except for the two mentioned
> > earlier)
> > can be considered appropriate to be implemented in CSymPy at the present?
>
> Great question.
>
> So for example, I don't think we need to implement limits, because
> I don't know any application where one would need to call limit() in
> some loop for lots of different things,
> in other words, SymPy is fast enough for this.
>

I agree.


> I think that were people find SymPy slow is:
>
> * general (long) expression manipulation
>
> * sparse multivariate polynomials (though the one in the latest sympy
> are pretty fast already), but
> I think there would be a benefit of having very efficient sparse
> multivariate poly implementation
>
> * Matrix manipulation with symbolic expressions, things like linear
> solve and similar.
>
> Most applications that I've seen where people tried SymPy and it was
> too slow have the above.
>
> So from this, the Matrix class and related algorithms is another great
> GSoC project I think. The algorithms
> are in SymPy, so I think it should be clear what to do.


Yes, CSymPy can hugely benefit from a Matrix class and related algorithms.
As you have correctly pointed out, especially when solving linear equations.
I was reading 
this<http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.47.2737>to
implement the solutions for a linear diophantine equation
system for SymPy/CSymPy. If we had a Matrix class in CSymPy, implementing
this would be really easy.

Ondrej
>
> --
> You received this message because you are subscribed to the Google Groups
> "sympy" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To post to this group, send email to [email protected].
> Visit this group at http://groups.google.com/group/sympy.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sympy/CADDwiVBHRcSoLsz5wyB5XV8rZXor68p2q-fzcQKQF5bpikq_Yw%40mail.gmail.com
> .
> For more options, visit https://groups.google.com/groups/opt_out.
>

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sympy.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sympy/CAPnoRoqq6DsoAZa1uDQULruhYMRoeaejE5vxZ_J2h1ke%3D_NUUg%40mail.gmail.com.
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to