Machine learning of Kondo physics using variational autoencoders and symbolic regression

Abstract

We employ variational autoencoders to extract physical insight from a dataset of one-particle Anderson impurity model spectral functions. Autoencoders are trained to find a low-dimensional, latent space representation that faithfully characterizes each element of the training set, as measured by a reconstruction error. Variational autoencoders, a probabilistic generalization of standard autoencoders, further condition the learned latent space to promote highly interpretable features. In our study, we find that the learned latent variables strongly correlate with well known, but nontrivial, parameters that characterize emergent behaviors in the Anderson impurity model. In particular, one latent variable correlates with particle-hole asymmetry, while another is in near one-to-one correspondence with the Kondo temperature, a dynamically generated low-energy scale in the impurity model. Using symbolic regression, we model this variable as a function of the known bare physical input parameters and "rediscover" the non-perturbative formula for the Kondo temperature. The machine learning pipeline we develop suggests a general purpose approach which opens opportunities to discover new domain knowledge in other physical systems.

arXiv commentUpdate to match PRB publication + typo fixes + minor edits

References in corpus (15)

• Angle-resolved photoemission spectroscopy of the cuprate superconductors
• The numerical renormalization group method for quantum impurity systems
• Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders
• Sum-rule Conserving Spectral Functions from the Numerical Renormalization Group
• Many-body Effects in Angle-resolved Photoemission: Quasiparticle Energy and Lifetime of a Mo(110) Surface State
• Searching for New Physics with Deep Autoencoders
• Ultrafast energy and momentum resolved dynamics of magnetic correlations in photo-doped Mott insulator Sr${}_2$IrO${}_4$
• Robust learning from noisy, incomplete, high-dimensional experimental data via physically constrained symbolic regression
• Variational Autoencoders for New Physics Mining at the Large Hadron Collider
• Unsupervised phase discovery with deep anomaly detection
• Machine Learning Assisted Insight to Spin Ice Dy${}_2$Ti${}_2$O${}_7$
• Tensor networks and the numerical renormalization group
• Exploring order parameters and dynamic processes in disordered systems via variational autoencoders
• Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised Learning
• Predicting impurity spectral functions using machine learning

Cited by in corpus (6)

• Decoding Structure-Spectrum Relationships with Physically Organized Latent Spaces
• Explainable Representation Learning of Small Quantum States
• Machine-learning the spectral function of a hole in a quantum antiferromagnet
• Physically interpretable approximations of many-body spectral functions
• Autoencoder-based analytic continuation method for strongly correlated quantum systems
• Hamiltonian parameter inference from resonant inelastic x-ray scattering with active learning