Agents speed up DFT by 36x, sparking push to auto-optimize forcefields for MD
lihua_lei_stat · x · 2026-11-13
The thread highlights agents being deployed to peruse codebases and suggest fixes (e.g., a softmax/x-entropy bug in torch/jax), and envisions agents optimally reimplementing forcefields/potentials for MD and FEP. A recent example from discoveredmat sped up electronic DFT by 36x, suggesting large untapped speedups in computational chemistry.
Related event: AI agents speed up scientific computing, boosting DFT 36x(3 posts)→
More from coding & agent
- SGLang Summit: agents now write ~95% of merged PRs as v1.0 and Miles v0.2 ship — hsu_byron · 2026-11-13
- Real-time chess tracking from iPhone video with RF-DETR and Stockfish — measure_plan · 2026-11-13
- OmniMind: an open-source MCP server for handing off tasks between AI agents — IrfanCodesBTW · 2026-11-13
- Claude Design Crit in Canvas: Screenshots Every Component State, Fixes All Comments in Parallel — max__drake · 2026-11-13
- Billion-dollar Unity ships the same editor for 10 years; one user builds their own equivalent — gdechichi · 2026-11-13
- When anyone can vibecode, the moat is UX polish, says ex-Google VC — vivekhaldar · 2026-11-13