GSOC project listings

42 views
Skip to first unread message

Oscar Benjamin

unread,
Nov 16, 2025, 5:36:23 PM (5 days ago) Nov 16
to sympy
Hi all,

We are at the time of year where it seems that candidates are
interested in GSOC. Every year the same list of projects is rolled
over and I think that needs to change. Most of the ideas in the list
are effectively impossible for GSOC or don't have anyone who would
mentor them. Most of the ideas are not described in enough up to date
detail to be implementable even if they are otherwise suitable.

This year we should not carry over any project ideas from previous
years and make a completely new GSOC ideas page with only ideas that
are fresh this year or that someone confirmed that they would at least
potentially be interested in supervising this year. Each idea should
clearly describe the current state and what the most immediate work to
be done is.

Oscar

Jason Moore

unread,
Nov 16, 2025, 11:50:03 PM (5 days ago) Nov 16
to sy...@googlegroups.com
Dear all,

In the past, it seemed our approach was to try to take the best students regardless of what they proposed (whether their own idea or one extracted from our list). We would then accept as many as we could mentor. But this approach is often awkward, or has gotten more so over the years, because it doesn't align with the mentorship resources (quantity and desire).

I would support an approach where we first start with a number of mentors, then determine the number of projects, and those mentors draft the N ideas for that year that can be applied for. We don't even have to consider any other ideas.

It is also true that we get more applicants for some projects than others. If we take the best applicant for that project, then we aren't necessarily taking the best applications from the whole pool. Should we accept the best applicants and assign them the projects or select the best per project?

The LLM written proposals will grow again. I'm not sure what should be done about that.

Jason


--
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 sympy+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sympy/CAHVvXxTwk%3DY0H02agwu6CN1PEzzC_uNm%3DJ2a%2BBH1QBs1Tk3F%3DQ%40mail.gmail.com.

Akash

unread,
Nov 17, 2025, 2:23:43 AM (5 days ago) Nov 17
to sympy
Hello all , 

I was reviewing the open projects and issues in this organisation that I can contribute to. I was interested to know whether the projects under the list - https://github.com/sympy/sympy/wiki/GSoC-Ideas will be considered for evaluation for upcoming GSOC .


Regards,


Akash Saha 

Jason Moore

unread,
Nov 17, 2025, 2:25:37 AM (5 days ago) Nov 17
to sy...@googlegroups.com
Dear Akash,

We are discussing whether that list will be considered or not in this thread. So there is no current answer to your question. Note that GSoC opens in March or April, so please be patient.

Jason


Aaron Meurer

unread,
Nov 17, 2025, 12:43:24 PM (5 days ago) Nov 17
to sy...@googlegroups.com
OK, so we should probably start out by clearing the current page and
starting from scratch (we can keep the structure). Some of the ideas
can be carried over but we can grab them from the page history.

I do also think we should keep track of the "moonshot" ideas, maybe on
a separate page. These are the things that would be pretty hard for
most students, typically due to the mathematical level required, but
are things we would like to see (like CAD, Karr, algebraic Risch,
etc.). Maybe we could call the page "algorithms we'd like to have" or
something like that so it isn't necessarily tied to GSoC.

Aaron Meurer

Aaron Meurer

unread,
Nov 17, 2025, 12:52:28 PM (5 days ago) Nov 17
to sy...@googlegroups.com
On Sun, Nov 16, 2025 at 9:49 PM Jason Moore <moore...@gmail.com> wrote:
>
> Dear all,
>
> In the past, it seemed our approach was to try to take the best students regardless of what they proposed (whether their own idea or one extracted from our list). We would then accept as many as we could mentor. But this approach is often awkward, or has gotten more so over the years, because it doesn't align with the mentorship resources (quantity and desire).
>
> I would support an approach where we first start with a number of mentors, then determine the number of projects, and those mentors draft the N ideas for that year that can be applied for. We don't even have to consider any other ideas.

I agree. I always try to do this but it's hard because we always seem
to end up with more people we want to accept than mentors. It's also
not as easy as just saying we have N mentors because different people
can mentor different things. I cannot mentor people on mechanics
projects, for instance (except as a backup mentor), so if we get only
students applying for mechanics projects then I cannot really help
even if I am able to mentor.

>
> It is also true that we get more applicants for some projects than others. If we take the best applicant for that project, then we aren't necessarily taking the best applications from the whole pool. Should we accept the best applicants and assign them the projects or select the best per project?

I think we should just continue to accept the best per project. The
application process is semi-open so people should be aware if they are
applying to a more "competitive" idea or not. People have to apply to
what they think they will be able to do, and what they are interested
in (and anyways, GSoC rules don't allow us to just completely change
people's project from what they applied for).

However, we should be more open about how many mentors we have for
each type of project. I guess one issue is that the potential mentors
list has a lot more people than actually mentor, and they aren't split
by project type
https://github.com/sympy/sympy/wiki/GSoC-Ideas#potential-mentors. We
should fix that. If we only have, say, 4 people who are able to mentor
mechanics projects, then ideally students should be aware going in
that we will only be able to accept 2 (or at most 3) projects.

It's challenging because some people will mentor any project to help
out with mentoring, but a lot of people aren't fully committed to
mentoring, or will only mentor a project if it's one they are
interested in. I think a simple thing we can do is to replace that
list on the ideas page with a table, with "name", "email", "types of
projects willing to mentor", "can be primary mentor or just backup
mentor" (anything else?). That would be more informative to the
students and it would also help me as org admin a lot too.

Aaron Meurer
> To view this discussion visit https://groups.google.com/d/msgid/sympy/CAP7f1AhzW7Wcht7Ydh2O6eM_qoFYcfo8Ly_mAjaum7usq7tDmQ%40mail.gmail.com.
Reply all
Reply to author
Forward
0 new messages