Quantum

Scalable Neural Wave Functions via Information Mixing and Hierarchical Network Architectures

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 23/10/2026, 04:13:24 UTC.

Briefings

A new research paper proposes a hierarchical neural architecture combining multi-layer perceptron (MLP) mixing with transformers to solve quantum many-body simulation challenges. The study benchmarks this approach against RNNs and transformer models on 2D quantum systems, demonstrating superior performance for large-scale simulations while maintaining computational efficiency. The authors report that the hierarchical combination of MLP and transformer models outperforms other state-of-the-art architectures for investigating quantum phases of matter.

Key points

• The proposed architecture combines multi-layer perceptron (MLP) mixing with transformer models to capture long-range correlations in quantum many-body systems.
• Benchmarking on 2D quantum many-body systems shows the hierarchical MLP+transformer model outperforms other state-of-the-art architectures for large-scale simulations.

Context

Quantum many-body systems are central to modern physics, but simulating them is intractable due to exponential Hilbert space growth. Recurrent neural networks (RNNs) are efficient but lack long-range correlation capacity, while transformers excel at capturing these correlations but require high computational resources. This work addresses these limitations by proposing a hybrid MLP+transformer architecture.

Limitations

The study benchmarks the model on 2D quantum many-body systems across various phases of matter. Results are limited to ground state representations and representational accuracy for different system sizes. No additional background on the physical systems or specific experimental constraints is supplied.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 23/10/2026, 04:30:04 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeidvfwlbfmjrpqxrc7rg62tk3x2akxvrfmp63nxswwl5cqw3fuyuta