Symmetries and phase diagrams with real-space mutual information neural estimation
arXiv:2103.16887 · doi:10.1103/PhysRevE.104.064106
Abstract
Real-space mutual information (RSMI) was shown to be an important quantity, formally and from a numerical standpoint, in finding coarse-grained descriptions of physical systems. It very generally quantifies spatial correlations, and can give rise to constructive algorithms extracting relevant degrees of freedom. Efficient and reliable estimation or maximization of RSMI is, however, numerically challenging. A recent breakthrough in theoretical machine learning has been the introduction of variational lower bounds for mutual information, parametrized by neural networks. Here we describe in detail how these results can be combined with differentiable coarse-graining operations to develop a single unsupervised neural-network based algorithm, the RSMI-NE, efficiently extracting the relevant degrees of freedom in the form of the operators of effective field theories, directly from real-space configurations. We study the information contained in the statistical ensemble of constructed coarse-graining transformations, and its recovery from partial input data using a secondary machine learning analysis applied to this ensemble. In particular, we show how symmetries, also emergent, can be identified. We demonstrate the extraction of the phase diagram and the order parameters for equilibrium systems, and consider also an example of a non-equilibrium problem.
arXiv commentAccepted version. Added new discussion of symmetries in the ensemble of coarse-graining rules. 27 pages, 13 figures. arXiv admin note: substantial text overlap with arXiv:2101.11633
References in corpus (17)
- Estimating Mutual Information
- Machine learning and the physical sciences
- Area laws in quantum systems: mutual information and correlations
- Regularity Properties and Pathologies of Position-Space Renormalization-Group Transformations
- Critical phenomena in atmospheric precipitation
- Strong disorder RG approach of random systems
- Mutual Information, Neural Networks and the Renormalization Group
- Interacting classical dimers on the square lattice
- Classical dimers with aligning interactions on the square lattice
- Exact Phase Diagram of a model with Aggregation and Chipping
- Mutual information in classical spin models
- Information theoretic aspects of the two-dimensional Ising model
- Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised Learning
- Machine-learning Iterative Calculation of Entropy for Physical Systems
- Optimal Renormalization Group Transformation from Information Theory
- Relevance in the Renormalization Group and in Information Theory
- Statistical physics through the lens of real-space mutual information
Cited by in corpus (11)
- A robust estimator of mutual information for deep learning interpretability
- Statistical physics through the lens of real-space mutual information
- Network Renormalization
- Machine Learning the Operator Content of the Critical Self-Dual Ising-Higgs Gauge Model
- Learning phase transitions from regression uncertainty: A new regression-based machine learning approach for automated detection of phases of matter
- Compression theory for inhomogeneous systems
- Bayesian RG Flow in Neural Network Field Theories
- Scanning-probe and information-concealing machine learning intermediate hexatic phase and critical scaling of solid-hexatic phase transition in deformable particles
- Reveal flocking phase transition of self-propelled active particles by machine learning regression uncertainty
- Minimizing couplings in renormalization by preserving short-range mutual information
- Characterizing nonlinear dynamics by contrastive cartography