Emergence of global receptive fields capturing multipartite quantum correlations
arXiv:2408.13033 · doi:10.1103/t6sy-r7l5
Abstract
In quantum physics, even simple data with a well-defined structure at the wave function level can be characterized by extremely complex correlations between its constituent elements. The inherent non-locality of the quantum correlations generally prevents one from providing their simple and transparent interpretation, which also remains a challenging problem for advanced classical techniques that approximate quantum states with neural networks. Here we show that monitoring the neural network weight space while learning quantum statistics from measurements allows to develop physical intuition about complex multipartite patterns and thus helps to construct more effective classical representations of the wave functions. Particularly, we observe the formation of distinct global convolutional structures, receptive fields in the hidden layer of the Restricted Boltzmann Machine (RBM) within the neural quantum tomography of the highly-entangled Dicke states. On this basis we propose an exact two-parameter classical representation not only for a specific quantum wave function, but for the whole family of the N-qubit Dicke states of different entanglement. Our findings suggest a fresh look at constructing convolutional neural networks for processing data with non-local patterns and pave the way for developing exact learning-based representations of entangled quantum states.
The second version of the manuscript with new results on symmetrical quantum states
References in corpus (32)
- Solving the Quantum Many-Body Problem with Artificial Neural Networks
- Machine learning phases of matter
- Entanglement versus Correlations in Spin Systems
- Quantum Entanglement in Neural Network States
- Efficient Representation of Quantum Many-body States with Deep Neural Networks
- Symmetries and many-body excited states with neural-network quantum states
- Study of the Two-Dimensional Frustrated J1-J2 Model with Neural Network Quantum States
- Reinforcement Learning with Neural Networks for Quantum Feedback
- Neural-Network Quantum States, String-Bond States, and Chiral Topological States
- Constructing exact representations of quantum many-body systems with deep neural networks
- Finite automata for caching in matrix product algorithms
- Neural Quantum States of frustrated magnets: generalization and sign structure
- Dirac-type nodal spin liquid revealed by refined quantum many-body solver using neural-network wave function, correlation ratio, and level spectroscopy
- Towards a Theory of Evolution as Multilevel Learning
- Approximating strongly correlated spin and fermion wavefunctions with correlator product states
- Transformer variational wave functions for frustrated quantum spin systems
- Using Sequences of Life-events to Predict Human Lives
- Unifying Neural-network Quantum States and Correlator Product States via Tensor Networks
- Experimental quantum homodyne tomography via machine learning
- Quantifying genuine multipartite correlations and their pattern complexity
- Entanglement and the Sign Structure of Quantum States
- Investigating Topological Order using Recurrent Neural Networks
- Data-driven emergence of convolutional structure in neural networks
- Wavefunction positivization via automatic differentiation
- Transformer Wave Function for two dimensional frustrated magnets: emergence of a Spin-Liquid Phase in the Shastry-Sutherland Model
- Neural Network Evolution Strategy for Solving Quantum Sign Structures
- 'Place-cell' emergence and learning of invariant data with restricted Boltzmann machines: breaking and dynamical restoration of continuous symmetries in the weight space
- Compact Neural-network Quantum State representations of Jastrow and Stabilizer states
- Many-body quantum sign structures as non-glassy Ising models
- Dicke states as matrix product states
- Investigating Network Parameters in Neural-Network Quantum States
- Achieving the volume-law entropy regime with random-sign Dicke states