GSOC 2012 idea

[Saurabh Jha]

My name is Saurabh Jha. I am currently pursuing B.Tech in computer
science and engineering. I am very much interested in working in sympy
project, more specifically on the idea "implement Karr algorithm, a
decision procedure for symbolic summation". It's rated "very hard", so
I wanted to know whether this idea is feasible for summer of code or
it is too ambitious.

Regards,
-Saurabh Jha

[Sergiu Ivanov]

Hello,

In my understanding, if the idea is listed on the GSoC ideas list, it
can be done in a summer.

Ideas which are rated as "very hard" will almost certainly require
solid mathematical background to understand the algorithms behind the
would-be implementation.

Could you please describe your background in mathematics?

Sergiu

[Saurabh Jha]

I went through the paper "Symbolic summation with radical expression"
and I found myself unable to understand many points due to my
insufficient mathematical background which is summarized below:

1. Calculus(Single and Multivariable)
2. Coordinate Geometry
3. Sequences Series (General properties of series)
4. Complex numbers
5. Basic Linear Algebra
6. Binomial Theorem, Permutation and Combination

I haven't done any abstract algebra.
However, I went through the idea list and found the idea "SYMBOLIC
COMPUTATION OF INTEGRALS BY RECURRENCE" interesting. I went through
Michael Barnett's paper and it seemed approachable to me with my
mathematical background. Also I saw that integration is currently done
using Risch-Norman Algorithm, and it needs to be improved.


Is the implementation of this algorithm appropriate as a Summer of
code project? Is my mathematical background enough for the successful
execution of this project?

Regards

-Saurabh Jha

On Mar 23, 7:50 pm, Sergiu Ivanov <unlimitedscol...@gmail.com> wrote:
> Hello,
>

[Sergiu Ivanov]

On Fri, Mar 23, 2012 at 10:55 PM, Saurabh Jha <saurab...@gmail.com> wrote:
>
> I went through the paper "Symbolic summation with radical expression"
> and I found myself unable to understand many points due to my
> insufficient mathematical background which is summarized below:
>
> 1. Calculus(Single and Multivariable)
> 2. Coordinate Geometry
> 3. Sequences Series (General properties of series)
> 4. Complex numbers
> 5. Basic Linear Algebra
> 6. Binomial Theorem, Permutation and Combination
>
> I haven't done any abstract algebra.
> However, I went through the idea list and found the idea "SYMBOLIC
> COMPUTATION OF INTEGRALS BY RECURRENCE" interesting. I went through
> Michael Barnett's paper and it seemed approachable to me with my
> mathematical background. Also I saw that integration is currently done
> using Risch-Norman Algorithm, and it needs to be improved.

I guess this last detail is the essential one. In my opinion, if you
understand the article sufficiently well to write a good proposal, you
are quite fit to actually implement the necessary functionality.

However, you should wait for Aaron's opinion on this matter.

> Is the implementation of this algorithm appropriate as a Summer of
> code project?

It is mentioned on [0], so, I guess it's appropriate.

Sergiu

[0] https://github.com/sympy/sympy/wiki/GSoC-2012-Ideas

[someone]

Hi,

> > I went through the paper "Symbolic summation with radical
> > expression" and I found myself unable to understand many points due
> > to my insufficient mathematical background

The whole topic of symbolic summation requires a
very strong background on (abstract) algebra. That's
one reason why it is rated as "very hard".

BTW this paper contains only additional information
describing extensions to the Karr algorithm.

Maybe we should group the references according to importance.

> > I haven't done any abstract algebra.
> > However, I went through the idea list and found the idea "SYMBOLIC
> > COMPUTATION OF INTEGRALS BY RECURRENCE" interesting. I went through
> > Michael Barnett's paper and it seemed approachable to me with my
> > mathematical background. Also I saw that integration is currently
> > done using Risch-Norman Algorithm, and it needs to be improved.
>
> I guess this last detail is the essential one. In my opinion, if you
> understand the article sufficiently well to write a good proposal, you
> are quite fit to actually implement the necessary functionality.

I don't think that this algorithm is enough work for a whole
gsoc project. It is only a single technique applicable to some
(rather special) integration problems.

Never the less it would be nice to have it to extend the integration
facilities of sympy.

Also there is at least one other gsoc thread about this topic.
(But I could not find these mails right now.)

[Saurabh Jha]

Hi,
Since the earlier idea(Symbolic Computation of integral by recurrence)
is not enough for a whole GSoC project, I went through the idea list
once again, and focused on "Series Expansions". I then went through
the link "current situation", but was unable to figure out whether the
implementation of algorithms listed in references is the main goal or
something else.

Can the implementation of the algorithms listed in references be a
good project for SymPy?

Regards,

-Saurabh Jha