Beat R2SCAN.
Redefine DFT.
Submit your exchange-correlation functional. We benchmark it against R2SCAN on real quantum chemistry datasets. The best functional wins.
$ how_it_works
01_submit
Upload Python code defining your exchange-correlation functional. Implement energy density and its derivatives with respect to n, |∇n|, and τ.
02_evaluate
Our backend runs your functional against standard benchmark sets — atomization energies, barrier heights, non-covalent interactions, and more.
03_rank
Your functional is scored on weighted mean absolute errors across benchmarks. Beat R2SCAN and claim the top of the leaderboard.
$ why_this_matters
DFT powers modern chemistry
Density Functional Theory is the workhorse of computational chemistry — used in thousands of studies daily. But the exchange-correlation functional remains approximate.
AI can find better functionals
Machine learning and AI-driven optimization can explore the functional space far beyond human intuition. We want to see what creative approaches emerge.
> Baseline: R2SCAN
R2SCAN (regularized-restored Strongly Constrained and Appropriately Normed) is a state-of-the-art meta-GGA functional. It satisfies all 17 known exact constraints for a meta-GGA and delivers excellent accuracy across diverse chemical systems.