Jiaxiang Li

A new Rust-based DFT code achieves sub-meV accuracy against Quantum ESPRESSO via a novel, fixture-driven validation system.

Hybrid functional DFT calculations can now be accurately learned by neural networks using compact AO Hamiltonians derived from PW data, bypassing the need for explicit nonlocal operator reconstruction.

By intelligently filtering out irrelevant visual noise, TPS-Drive lets VLMs focus on what *really* matters for safe self-driving: dynamic agents, not static scenery.