AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-23
Scalable Neural Wave Functions via Information Mixing and Hierarchical Network Architectures
Simulating quantum many-body systems is hard because the Hilbert space grows exponentially with system size. The authors test an MLP mixer as a neural wave function ansatz on ground states of two-dimensional quantum many-body systems, comparing it with RNNs and Transformers. The MLP mixer performs competitively with state-of-the-art models, and a hierarchical combination with a Transformer outperforms other state-of-the-art architectures for large-scale quantum systems.