时间：2025年8月27日（星期三）16:00-17:30
地点：西湖大学云谷校区E10-212
主讲人：Junwei Liu, Hong Kong University of Science and Technology
报告题目：Effective-field neural networks for many-body interactions
报告摘要：We introduce effective-field neural networks (EFNNs), a new architecture based on continued functions---mathematical tools used in renormalization to handle divergent perturbative series. Our key insight is that neural networks can implement these continued functions directly, providing a principled approach to many-body interactions. Testing on three systems (a classical 3-spin model, a continuous Heisenberg system, and a quantum double exchange model), we find that EFNN outperforms standard deep networks, ResNet, and DenseNet. Most striking is EFNN's generalization: trained on 10×10 lattices, it accurately predicts behavior on systems up to 40×40 with no additional training---and the accuracy improves with system size. This demonstrates that EFNN captures the underlying physics rather than merely fitting data, making it valuable beyond many-body problems to any field where renormalization ideas apply.
