Awarded: Trelis AI Grant
Third Matter has been awarded a Trelis AI Grant from Trelis Research — compute support for the GPU-heavy part of the project.
The grant backs a concrete, falsifiable goal: a DFT validation of machine-learning-predicted proton-transfer barriers in iron sulfides. We use the MACE interatomic potential to predict diffusion barriers across three minerals — pentlandite, mackinawite, and pyrite — and then check those predictions against first-principles DFT reference calculations (GPAW), targeting agreement within 0.15 eV. The outcome, benchmarks and all, goes into an open preprint.
This is exactly the kind of work where compute is the bottleneck for an independent researcher: no institutional cluster, no allocation, just the question and the calculations needed to answer it honestly. The Trelis grant closes that gap.
Work is in progress, with a target completion date of 30 June 2026. We are grateful to Trelis Research for supporting curiosity-driven, open research from outside the usual institutions.