Machine Learning Predicts Hamiltonian Parameters From Phase Diagrams
Yasinskaya et al. (submitted Dec 3, 2025) present a machine-learning method to infer Hamiltonian parameters of a cuprate superconductor from computed phase diagrams. They compare VGG, ResNet and a regression-adapted U-Net, training on mean-field phase diagrams and validating on semiclassical heat-bath Monte Carlo data; U-Net performed best and accurately recovered parameters, highlighting parameter-insensitive regions and aiding experimental parameter selection.
Key Points
- 1Demonstrates U-Net regression predicts Hamiltonian parameters from simulated cuprate phase diagrams accurately.
- 2Validates model on semiclassical heat-bath Monte Carlo data, showing robustness across simulation methods.
- 3Enables identification of parameter-insensitive regions, guiding experimental parameter selection and model interpretation.
Scoring Rationale
Demonstrates practical ML inversion with cross-method validation, limited by specialized cuprate focus and preprint-stage (non-peer-reviewed) results.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
