Highly entangled tensor networks

Jack Y. Araz, Michael Spannowsky

Tensor networks (TN) are approximations of high-dimensional tensors designed to represent locally entangled quantum many-body systems efficiently. This paper provides a comprehensive comparison between classical TNs and TN-inspired quantum circuits in the context of machine learning on highly complex, simulated Large Hadron Collider data. We show that classical TNs require exponentially large bond dimensions and higher Hilbert-space mapping to perform comparably to their quantum counterparts. While such an expansion in the dimensionality allows better performance, we observe that, with increased dimensionality, classical TNs lead to a highly flat loss landscape, rendering the usage of gradient-based optimization methods highly challenging. Furthermore, by employing quantitative metrics, such as the Fisher information and effective dimensions, we show that classical TNs require a more extensive training sample to represent the data as efficiently as TN-inspired quantum circuits. We also engage with the idea of hybrid classical-quantum TNs and show possible architectures to employ a larger phase space from the data. We offer our results using three main TN Ans\"atze: Tree tensor networks, matrix product states, and multiscale entanglement renormalization Ans\"atze.

Samuel Desrosiers, Glen Evenbly, Thomas E. Baker

Philip Blagoveschensky, Anh Huy Phan

Neural networks have achieved state of the art results in many areas, supposedly due to parameter sharing, locality, and depth. Tensor networks (TNs) are linear algebraic representations of quantum many-body states based on their entanglement structure. TNs have found use in machine learning. We devise a novel TN based model called Deep convolutional tensor network (DCTN) for image classification, which has parameter sharing, locality, and depth. It is based on the Entangled plaquette states (EPS) TN. We show how EPS can be implemented as a backpropagatable layer. We test DCTN on MNIST, FashionMNIST, and CIFAR10 datasets. A shallow DCTN performs well on MNIST and FashionMNIST and has a small parameter count. Unfortunately, depth increases overfitting and thus decreases test accuracy. Also, DCTN of any depth performs badly on CIFAR10 due to overfitting. It is to be determined why. We discuss how the hyperparameters of DCTN affect its training and overfitting.

Qiang Wen, Mingshuai Xu, Haocheng Zhong

Holographic tensor networks serve as toy models for the Anti-de Sitter/Conformal Field Theory (AdS/CFT) correspondence, capturing many of its essential features in a concrete manner. However, existing holographic tensor network models remain far from a complete theory of quantum gravity. A key obstacle is their discrete structure, which only approximates the semi-classical geometry of gravity in a qualitative sense. In \cite{Lin:2024dho}, it was shown that a network of partial-entanglement-entropy (PEE) threads, which are bulk geodesics with a specific density distribution, generates a perfect tessellation of AdS space. Moreover, such PEE-network tessellations can be constructed for more highly symmetric geometries using the Crofton formula. In this paper, we assign a quantum state to each vertex in the PEE network and develop several holographic tensor network models: (1) the factorized PEE tensor network, which takes the form of a tensor product of EPR pairs; (2) the HaPPY-like PEE tensor network constructed from perfect tensors; and (3) the random PEE tensor network. In all these models, we reproduce the exact Ryu-Takayanagi formula by showing that the minimal number of cuts along a homologous surface in the network exactly equals the area of that surface.

Simone Montangero

Thibaut Lacroix