GraphED: graph-based AI learns how solids deform by sharing law structure across materials
bravo_abad · x · 2026-09-12
Xu et al. propose GraphED for scientific equation discovery. Instead of fitting a separate symbolic expression per dataset, it targets the part of the equation that stays constant across many experiments.
Key points:
- Candidate equations are graphs: nodes are math operations, edges carry material-specific parameters;
- The structure of the law is shared; only coefficients vary per material;
- Each candidate is tested across multiple datasets, fitting parameters per material and scoring the shared structure.
For rate-dependent deformation of steels, the resulting explicit model reduces prediction error (post truncated). A notable shift toward 'shared structure, variable parameters' in equation discovery.
More from Research
- Whole-Brain Fly Simulation Hits 91% Behavior Accuracy With No Training — alexcovo_eth · 2026-09-12
- AI bio capability hype vs wet lab reality: models fail over half the time — nlarusstone · 2026-09-12
- smolbenchmark ranks sub-8GB models by speed, tok/J and heat on your own hardware — East-Muffin-6472 · 2026-09-12
- Clay Math Institute says the Navier-Stokes problem 'has apparently been settled' — badumtsssst · 2026-09-12
- Decagon shares 19+ ablations on using GEPA for test-driven prompt optimization in production — kastnerkyle · 2026-09-12
- GeoGuessr as an RL env: 4B VLM trained with OpenEnv and TRL to play the game — SergioPaniego · 2026-09-12