Adiabatic Fine-Tuning of Neural Quantum States Enables Detection of Phase Transitions in Weight Space
This work addresses the interpretability problem for researchers using machine learning in physics, offering a method to detect phase transitions from network weights, though it is incremental as it builds on existing neural quantum state frameworks.
The authors tackled the challenge of interpreting how neural quantum states encode physical information by introducing adiabatic fine-tuning, which trains models across a phase diagram and reveals phase transitions through correlated weight representations, validated on the transverse field Ising and J1-J2 Heisenberg models with distinct structures in weight space.
Neural quantum states (NQS) have emerged as a powerful tool for approximating quantum wavefunctions using deep learning. While these models achieve remarkable accuracy, understanding how they encode physical information remains an open challenge. In this work, we introduce adiabatic fine-tuning, a scheme that trains NQS across a phase diagram, leading to strongly correlated weight representations across different models. This correlation in weight space enables the detection of phase transitions in quantum systems by analyzing the trained network weights alone. We validate our approach on the transverse field Ising model and the J1-J2 Heisenberg model, demonstrating that phase transitions manifest as distinct structures in weight space. Our results establish a connection between physical phase transitions and the geometry of neural network parameters, opening new directions for the interpretability of machine learning models in physics.