GSoC 2026 Proposal Draft: Algorithmic Optimization of Sparse Polynomial GCD

Message has been deleted

Gaurav deshmukh

unread,

Mar 19, 2026, 7:06:08 AM (5 days ago) Mar 19

to sympy

Hi SymPy Development Team,

I am an active contributor preparing my GSoC 2026 proposal. I am targeting the sparse polynomial GCD bottleneck outlined in Issue #23131.

In my local profiling, the dense fallback dmp_rr_prs_gcd required ~1.62s for a 50-variable polynomial in ZZ_I, whereas a pure sparse dictionary traversal executed in 0.000023s (a structural optimization factor of over 70,000x).

To resolve this memory and execution bottleneck, my proposal outlines an architecture to replace the dense fallback with a native sparse implementation using Zippel’s Modular Algorithm to bypass intermediate expression swell, keeping polynomials natively within sympy.polys.rings.PolyElement.

Draft Link: https://docs.google.com/document/d/1PQJVOHLJnm_xLPo9y-JokQdOwdKzrfWuRFekOIDH4o4/edit?usp=sharing

I would greatly appreciate any brutal, technical feedback on the timeline and the mathematical architecture before the final submission window closes.

Best regards,

Gaurav Deshmukh

GitHub: Gaurav23-bit

Current PR: #28922: Add coverage for transitive inequality simplification

Gaurav deshmukh

unread,

Mar 21, 2026, 1:23:10 AM (3 days ago) Mar 21

to sympy

Hi everyone,

Apologies for the duplicate email yesterday—I had a slight browser glitch when submitting.

To make this bump worthwhile, I wanted to share that I have significantly updated my proposal draft. Specifically, I have integrated my recent codebase momentum into the "Contributor Background" section to demonstrate my familiarity with SymPy's architecture and strict CI pipeline. This includes:

PR #29544: Implementing a LaTeX printer for PuiseuxPoly to resolve a raw string fallback issue. (Open / CI Passing)
PR #28922: Adding transitive inequality simplification coverage to the logic module. (Open / CI Passing)

I am continuing to refine the architectural phases (specifically regarding the Normalization problem during sparse interpolation) and the execution timeline.

The updated draft is at the same link: https://docs.google.com/document/d/1PQJVOHLJnm_xLPo9y-JokQdOwdKzrfWuRFekOIDH4o4/edit?usp=sharing

I am actively seeking brutal technical critiques on the mathematical architecture before the final submission window closes. Thank you for your time and bandwidth.