Skip to content

Briefly

Research Digest — 2026-10-11

deng-group/briefly

Research Digest — 2026-10-11¶

ML Interatomic Potentials & AI Methods¶

1. AtomWorld-Mirror: Macro-Step World Modeling of Critical Evolution Backbones for Materials Dynamics ¶

Source: arXiv (cond-mat.mtrl-sci) · 📅 2026-10-08 · ↗ Open paper

Proposes a 'world model' approach that predicts macro-steps of critical evolution backbones in materials dynamics instead of simulating event-by-event. On a nitrogen-based anti-perovskite solid electrolyte, macro-step inference delivers a 10^3-10^4x speedup over conventional event-by-event simulation.

Relevance to DENG.Group

An emerging AI-native alternative to MD sampling, demonstrated on a solid electrolyte system — a watch-list item for Yanhao Deng's ML potential work and for Jerry's grant strategy on where atomistic simulation is heading (learned dynamics vs. learned potentials).