A quantum inspired approach to learning dynamical laws from data -- block-sparsity and gauge-mediated weight sharing
arXiv:2208.01591 · doi:10.1088/2632-2153/ad4f4e
Abstract
Recent years have witnessed an increased interest in recovering dynamical laws of complex systems in a largely data-driven fashion under meaningful hypotheses. In this work, we propose a scalable and numerically robust method for this task, utilizing efficient block-sparse tensor train representations of dynamical laws, inspired by similar approaches in quantum many-body systems. Low-rank tensor train representations have been previously derived for dynamical laws of one-dimensional systems. We extend this result to efficient representations of systems with -mode interactions and controlled approximations of systems with decaying interactions. We further argue that natural structure assumptions on dynamical laws, such as bounded polynomial degrees, can be exploited in the form of block-sparse support patterns of tensor-train cores. Additional structural similarities between interactions of certain modes can be accounted for by weight sharing within the ansatz. To make use of these structure assumptions, we propose a novel optimization algorithm, block-sparsity restricted alternating least squares with gauge-mediated weight sharing. The algorithm is inspired by similar notions in machine learning and achieves a significant improvement in performance over previous approaches. We demonstrate the performance of the method numerically on three one-dimensional systems -- the Fermi-Pasta-Ulam-Tsingou system, rotating magnetic dipoles and point particles interacting via modified Lennard-Jones potentials, observing a highly accurate and noise-robust recovery.
arXiv comment19 pages, 5 figures. New analytical results added, block sparsity proof included, major revisions of the text
References in corpus (22)
- Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
- Area laws for the entanglement entropy - a review
- A Practical Introduction to Tensor Networks: Matrix Product States and Projected Entangled Pair States
- Matrix Product States, Projected Entangled Pair States, and variational renormalization group methods for quantum spin systems
- Data-driven discovery of coordinates and governing equations
- Matrix Product States and Projected Entangled Pair States: Concepts, Symmetries, and Theorems
- Unifying time evolution and optimization with matrix product states
- Matrix product states represent ground states faithfully
- Renormalization algorithms for Quantum-Many Body Systems in two and higher dimensions
- Discovering physical concepts with neural networks
- Entropy scaling and simulability by Matrix Product States
- Hand-waving and Interpretive Dance: An Introductory Course on Tensor Networks
- Tensor network decompositions in the presence of a global symmetry
- Neural-Network Quantum States, String-Bond States, and Chiral Topological States
- On the Expressive Power of Deep Learning: A Tensor Analysis
- Geometry of Matrix Product States: metric, parallel transport and curvature
- Multidimensional approximation of nonlinear dynamical systems
- SINDy-BVP: Sparse Identification of Nonlinear Dynamics for Boundary Value Problems
- Tensor-based algorithms for image classification
- Expressive power of tensor-network factorizations for probabilistic modeling, with applications from hidden Markov models to quantum machine learning
- Tensor network approaches for learning non-linear dynamical laws
- CINDy: Conditional gradient-based Identification of Non-linear Dynamics -- Noise-robust recovery