A simple tweak to a neural network's wiring allows it to simulate complex quantum physics that usually requires supercomputers.
April 14, 2026
Original Paper
Geometry-Induced Long-Range Correlations in Recurrent Neural Network Quantum States
arXiv · 2604.08661
The Takeaway
By adding "dilated" gaps in how neurons connect, researchers unlocked the ability to model long-range quantum behaviors that standard AI couldn't touch. This turns simple, cheap models into high-powered tools for discovering new quantum materials with 100% accuracy.
From the abstract
Neural Quantum States based on autoregressive recurrent neural network (RNN) wave functions enable efficient sampling without Markov-chain autocorrelation, but standard RNN architectures are biased toward finite-length correlations and can fail on states with long-range dependencies. A common response is to adopt transformer-style self-attention, but this typically comes with substantially higher computational and memory overhead. Here we introduce dilated RNN wave functions, where recurrent uni