Neural network setups for a precise detection of the many-body localization transition: finite-size scaling and limitations

Abstract

Determining phase diagrams and phase transitions semi-automatically using machine learning has received a lot of attention recently, with results in good agreement with more conventional approaches in most cases. When it comes to more quantitative predictions, such as the identification of universality class or precise determination of critical points, the task is more challenging. As an exacting test-bed, we study the Heisenberg spin-1/2 chain in a random external field that is known to display a transition from a many-body localized to a thermalizing regime, which nature is not entirely characterized. We introduce different neural network structures and dataset setups to achieve a finite-size scaling analysis with the least possible physical bias (no assumed knowledge on the phase transition and directly inputing wave-function coefficients), using state-of-the-art input data simulating chains of sizes up to L=24. In particular, we use domain adversarial techniques to ensure that the network learns scale-invariant features. We find a variability of the output results with respect to network and training parameters, resulting in relatively large uncertainties on final estimates of critical point and correlation length exponent which tend to be larger than the values obtained from conventional approaches. We put the emphasis on interpretability throughout the paper and discuss what the network appears to learn for the various used architectures. Our findings show that a it quantitative analysis of phase transitions of unknown nature remains a difficult task with neural networks when using the minimally engineered physical input.

v2: published version

References in corpus (22)

• Many body localization and thermalization in quantum statistical mechanics
• Many-body localization edge in the random-field Heisenberg chain
• Many body localization in Heisenberg XXZ magnet in a random field
• Phenomenology of fully many-body-localized systems
• Learning phase transitions by confusion
• Discovering Phases, Phase Transitions and Crossovers through Unsupervised Machine Learning: A critical examination
• Recent progress in many-body localization
• Restricted-Boltzmann-Machine Learning for Solving Strongly Correlated Quantum Systems
• Multifractal Scalings across the Many-Body Localization Transition
• Ergodicity breaking in a model showing many-body localization
• Self-Learning Monte Carlo Method
• How a small quantum bath can thermalize long localized chains
• Machine learning vortices at the Kosterlitz-Thouless transition
• Machine learning meets quantum physics
• Kosterlitz-Thouless scaling at many-body localization phase transitions
• Solving the Bose-Hubbard model with machine learning
• Deep Learning the Quantum Phase Transitions in Random Two-Dimensional Electron Systems
• Renormalization-group study of the many-body localization transition in one dimension
• Super-resolving the Ising model with convolutional neural networks
• Quantum phase recognition via unsupervised machine learning
• Unsupervised Learning Eigenstate Phases of Matter
• Multi-faceted machine learning of competing orders in disordered interacting systems

Cited by in corpus (25)

• Many-Body Localization in the Age of Classical Computing
• Unsupervised phase discovery with deep anomaly detection
• Drawing Phase Diagrams of Random Quantum Systems by Deep Learning the Wave Functions
• Transition to a many-body localized regime in a two-dimensional disordered quantum dimer model
• Hybrid convolutional neural network and PEPS wave functions for quantum many-particle states
• Phase diagram study of a two-dimensional frustrated antiferromagnet via unsupervised machine learning
• Phase Detection with Neural Networks: Interpreting the Black Box
• Detecting ergodic bubbles at the crossover to many-body localization using neural networks
• Distinguishing an Anderson Insulator from a Many-Body Localized phase through space-time snapshots with Neural Networks
• Hessian-based toolbox for reliable and interpretable machine learning in physics
• Unsupervised mapping of phase diagrams of 2D systems from infinite projected entangled-pair states via deep anomaly detection
• Unsupervised Machine Learning of Quenched Gauge Symmetries: A Proof-of-Concept Demonstration
• Machine learning of phases and structures for model systems in physics
• Phase determination with and without deep learning
• A simple framework for contrastive learning phases of matter
• A learning algorithm with emergent scaling behavior for classifying phase transitions
• Supervised and unsupervised learning of the many-body critical phase, phase transitions, and critical exponents in disordered quantum systems
• Characterizing out-of-distribution generalization of neural networks: application to the disordered Su-Schrieffer-Heeger model
• Many-body mobility edges in 1D and 2D revealed by convolutional neural networks
• Supervised learning of an interacting 2D hard-core boson model of a weak topological insulator using correlation functions
• Neural Networks as Universal Probes of Many-Body Localization in Quantum Graphs
• Engineering many-body quantum Hamiltonians with non-ergodic properties using quantum Monte Carlo
• Global exploration of phase behavior in frustrated Ising models using unsupervised learning techniques
• Autonomous Discovery of the Ising Model's Critical Parameters with Reinforcement Learning
• Unsupervised Techniques to Detect Quantum Chaos