GSoC application

[krastano...@gmail.com]

Hi,

Some of you know me as I have a few minor contributions to sympy and I
helped a bit with GCI. I would like to know whether it is possible for
me to apply for a GSoC project this year?

About the project I have in mind: It is the tensor module. It is quite
bare-bones at the moment. I would like to make sufficiently general to
be useful in (for example):
- Einstein equations in general relativity
- Integrating Feynman diagrams (it has to do with a tensor product of
the space of Lorenz and the space of Dirac)
- nonlinear optics, linear elasticity, fluid mechanics, etc

The first two of those will be directly useful for me.
There is a package for Mathematica (called Ricci) that may serve as an
example but I have just started looking at it. I've also checked the
related issues in the issue tracker.

And about eligibility for the GSoC: I am still a student (4th year
physics in France).

So if you are interested I can prepare a more in-depth presentation of
the potential project.

Stefan

[Matthew Rocklin]

Hi Stefan, 

As someone who has followed and used many of your "minor" contributions (optics, plotting) I'll say that you should definitely apply for a GSoC project this year. I'm confident that your contributions will be very useful. 

My summer plans are up in the air but I may also end up applying again (and if not I'll probably end up mentoring). If so I will propose rebuilding the various Tensor and Matrix-like objects in SymPy in a more unified and sensible way. The focus here would be less on Tensors explicitly and more on how Tensors and the various kinds of Matrix objects can all live together elegantly. 

If you'd like to communicate about how these two ideas could work together I'd be happy to talk. I think that SymPy is well positioned to make a really awesome linear/tensor algebra module. I'd be happy if that work could happen this Summer. 

-Matt

Stefan

--
You received this message because you are subscribed to the Google Groups "sympy" group.
To post to this group, send email to sy...@googlegroups.com.
To unsubscribe from this group, send email to sympy+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/sympy?hl=en.

[Sean Vig]

Stefan,

Getting physics to utilize the tensor module sounds like a great idea. If you are looking for more things to flesh out a project, I know the spin module could be greatly improved with a well integrated tensor module, things like Wigner-Eckart theorem and irreducible spherical tensor operators immediately come to mind. From the work you've done so far, you definitely seem up to the task.

Sean

Stefan

[Aaron Meurer]

Hi.

This is great. I think we definitely need to improve that module.
You've already completed the patch requirement, so I would spend time
discussing your idea with the community. Once you've got a solid
idea, you can start writing an application (though it's still early,
so don't worry about this too much yet).

I know very little about the physics, but I have studied tensors in a
general setting in a course on differential geometry (general meaning
that we regularly look at things like T^{i_1, ..., i_n}_{j_n, ...,
j_m} without specifying n or m). So this is something to consider
when planning the module. Many tensor calculations can be computed
with a symbolic number of indices.

Of couse, this is not necessary if you don't want to do it. Just
giving you a little food for thought :)

Aaron Meurer

[krastano...@gmail.com]

Hi,

Thanks for the feedback. I will try to prepare some kind of a
blueprint (at least some imaginary code examples) and then we can
discuss them.

Matt has also sent me https://github.com/sympy/sympy/wiki/Linear-Algebra-Vision
There are some points from it that were not in my vision of the
potential project so I will pay some more attention to them in the
discussion.

Stefan

[Matthew Rocklin]

I should point out that the linear algebra vision page was just something I started after your e-mail. It reflects my thoughts and not necessarily those of the community in general. 

[Ondřej Čertík]

On Mon, Feb 6, 2012 at 4:04 AM, krastano...@gmail.com
<krastano...@gmail.com> wrote:
> Hi,
>
> Some of you know me as I have a few minor contributions to sympy and I
> helped a bit with GCI. I would like to know whether it is possible for
> me to apply for a GSoC project this year?
>
> About the project I have in mind: It is the tensor module. It is quite
> bare-bones at the moment. I would like to make sufficiently general to
> be useful in (for example):
> - Einstein equations in general relativity

This would be really cool. This was my very first "application" of sympy, to see
whether it can derive the Schwarzschild metric or not
(examples/advanced/relativity.py)

> - Integrating Feynman diagrams (it has to do with a tensor product of
> the space of Lorenz and the space of Dirac)

I tried to do this using brute force --- simply write the 4x4 matrices
explicitly,
multiply them out and then try to simplify the results, see
examples/advanced/qft.py

But it is a mess. I didn't figure out how to simplify this. So I think
a better approach
is to implement the "trace technology" using tensors.

Is this what you have in mind? There are packages for this in Mathematica,
so we should follow what they do.

> - nonlinear optics, linear elasticity, fluid mechanics, etc

Do you have something concrete in mind?

For example here I have derivation of Euler equations from the 4D
relativistic tensor:

http://theoretical-physics.net/dev/src/fluid-dynamics/general.html#perfect-fluids

it'd be nice to do this in sympy.

Anyway, any of the above would be really cool to have. These physical
projects really add value to SymPy, and typically show some deficiencies
in our simplification or manipulation.

>
> The first two of those will be directly useful for me.
> There is a package for Mathematica (called Ricci) that may serve as an
> example but I have just started looking at it. I've also checked the
> related issues in the issue tracker.
>
> And about eligibility for the GSoC: I am still a student (4th year
> physics in France).
>
> So if you are interested I can prepare a more in-depth presentation of
> the potential project.

You should definitely apply for GSoC as a student.

I am available to mentor projects like these.
Maybe Sean could become a mentor this year? :)

Wigner-Eckart theorem and irreducible spherical tensor operators are
also very useful. Here I
have an example how to use it to derive the formula for an integral of
three spherical
harmonics:

http://theoretical-physics.net/dev/src/math/spherical-harmonics.html#gaunt-coefficients

Ondrej

[krastano...@gmail.com]

Ondřej, I was thinking about studying those packages for Mathematica:

Ricci,
MathTensor,
FeynCalc

Were your referring to some other packages that I can look into? I
would like to check the established solution before I start designing
an API and so on.

2012/2/7 Ondřej Čertík <ondrej...@gmail.com>:

[Ondřej Čertík]

On Tue, Feb 7, 2012 at 1:13 PM, krastano...@gmail.com
<krastano...@gmail.com> wrote:
> Ondřej, I was thinking about studying those packages for Mathematica:
>
> Ricci,
> MathTensor,
> FeynCalc
>
> Were your referring to some other packages that I can look into? I

There is a good list here:

http://en.wikipedia.org/wiki/Tensor_software

I am aware of GRTensorII for Maple.

Cadabra is really good too for tensors, see here the examples:

http://cadabra.phi-sci.com/examples.html

(Search also this list for "Cadabra", you'll find a few threads.)

> would like to check the established solution before I start designing
> an API and so on.

Definitely.

Ondrej

[someone]

> Ondřej, I was thinking about studying those packages for Mathematica:
>
> Ricci,
> MathTensor,
> FeynCalc

Maybe you want also to look at xAct and related packages?

http://www.xact.es

The code is GPL, maybe this can be a benefit
even given that sympy uses a BSD license.

I have to admit that I never really used it though.