Updates on GSoC 2026 Ideas

[kaddy]

Hello community,

I am considering applying to GSoC 2026 to work under the sympy org. An idea which I've taken an interest in and explored a bit is extending manualintegrate. I have gone through the code and the run the test cases to see where manualintegrate is used and I believe I have a decent idea of the flow of control which leads to calls to manualintegrate. 

In particular, I noticed that solving issues like Issue 16396 which are addressed in test_failing_integrals.py, which should be easy for a student to calculate using change of variables and integration identities could be a target for the project. An ambitious target could also be to be able to compute the antiderivative analytically like it is done over here. I believe they use a combination of manual identities and a complete implementation of Risch algorithm, which is also something I would be interesting in going for in case there are ideas around it. 

I have my dev environment set up and I have submitted 2 bug-fixing PRs to sympy in the past. I don't have a college math background to be able to understand stuff like Risch algorithm well, but I am open to learning. Due to time constraints, I would prefer to take on a 90-hour project. 

I would appreciate guidance from the community on how best to proceed from here.

Thanks and regards,

ButteryPaws (on github).

[kaddy]

Hello all,

Just following up on the message above in case it got lost.

I’d appreciate any thoughts on whether the direction I’m exploring around manualintegrate makes sense, and what a good next step might be for someone interested in contributing in this area (for GSoC or otherwise). The Ideas page mentions that potential contributors should "Discuss with us" in case someone is interested in smaller projects to just improve manualintegrate, so this is the discussion I would like to initiate.

Thanks, and apologies for the nudge.

Best,
ButteryPaws.

[AB]

The risch algorithm can be studied from bronsteins book on transcendental equations, it also has the parametric version which is very helpful in many cases

Apart from that, there are cases where u substitutions in the manner required (like rationalising subs or cases where we solve for x in terms of u) are not yet implemented, work on that could be done, however I don't know if it's expansive enough to be considered a full fledged gsoc project 

Ayush

[kaddy]

Hey Ayush,

I have gone over some of the Risch code as well, there's clearly a lot of work to be done there. I am open to reading up on any literature required to implement the missing parts. Even the RUBI project looks quite interesting to me. The dedicated project for Risch implementation is a 350 hour GSoC project. I am afraid I can't commit to 175 or 350 hour projects. 

@Aaron Meurer is there scope for this idea to be broken down into smaller projects or some sidecar project which would be doable in 90 hours?

Thanks and Regards. 

[Aaron Meurer]

This does look like a good start. I would run through the code of
manualintegrate() on this integral and see what steps are missing. It
looks like this integral makes use of a straightforward trig
substitution which turns the integral into a rational integral.
Rational integrals are easy to solve (the ratint() function can solve
all of them).

Aaron Meurer

> --
> 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/829a739c-4828-40d5-aece-d5f041242f94n%40googlegroups.com.