Approximately-symmetric neural networks for quantum spin liquids
2405.17541
0
0
Abstract
We propose and analyze a family of approximately-symmetric neural networks for quantum spin liquid problems. These tailored architectures are parameter-efficient, scalable, and significantly out-perform existing symmetry-unaware neural network architectures. Utilizing the mixed-field toric code model, we demonstrate that our approach is competitive with the state-of-the-art tensor network and quantum Monte Carlo methods. Moreover, at the largest system sizes (N=480), our method allows us to explore Hamiltonians with sign problems beyond the reach of both quantum Monte Carlo and finite-size matrix-product states. The network comprises an exactly symmetric block following a non-symmetric block, which we argue learns a transformation of the ground state analogous to quasiadiabatic continuation. Our work paves the way toward investigating quantum spin liquid problems within interpretable neural network architectures
Create account to get full access
Overview
- This paper presents a new approach to designing neural networks that can capture the symmetries of many-body quantum systems, particularly quantum spin liquids.
- The researchers introduce "approximately-symmetric neural networks" (ASNNs), which can learn to respect the symmetries of a quantum system without being explicitly programmed with those symmetries.
- ASNNs are shown to outperform traditional neural networks on tasks related to modeling quantum spin liquids, which are exotic states of matter with highly correlated, fluctuating spins.
Plain English Explanation
The paper is about a new type of neural network that can learn the symmetries of quantum systems, particularly quantum spin liquids. Quantum spin liquids are strange states of matter where the magnetic moments, or "spins," of electrons are highly correlated and in constant flux, unlike in more conventional magnets.
Typically, neural networks need to be explicitly programmed with the symmetries of the system they are trying to model. But the authors have developed a new approach called "approximately-symmetric neural networks" (ASNNs) that can learn these symmetries on their own. This allows the neural network to better capture the complex behavior of quantum spin liquids compared to traditional methods.
The key insight is that by designing the neural network architecture in a certain way, it can automatically respect the symmetries of the quantum system without needing to be told what those symmetries are ahead of time. This makes the ASNNs more flexible and powerful for modeling exotic quantum states of matter.
Technical Explanation
The researchers propose a new neural network architecture called "approximately-symmetric neural networks" (ASNNs) that can learn to respect the symmetries of many-body quantum systems, such as quantum spin liquids, without being explicitly programmed with those symmetries.
ASNNs build on recent advances in equivariant neural networks, which can learn to transform their outputs in a way that respects the symmetries of the input data. However, the authors argue that existing equivariant approaches may be too rigid, as they require the network architecture to be carefully designed for each specific symmetry group.
In contrast, ASNNs are designed to be more flexible and approximate the desired symmetries, rather than enforcing them exactly. This is achieved by incorporating multiple "symmetry streams" into the network, each of which learns to transform the data in a slightly different way. The final output is then computed as a weighted average of these symmetry-transformed streams.
The authors demonstrate the effectiveness of ASNNs on several tasks related to modeling quantum spin liquids, including predicting ground state properties and learning to generate realistic spin configurations. They show that ASNNs outperform traditional neural networks and even some specialized equivariant architectures on these benchmarks.
Critical Analysis
The authors present a compelling approach to designing neural networks that can more effectively capture the symmetries of many-body quantum systems. By allowing the network to approximately learn the relevant symmetries, rather than enforcing them exactly, the ASNNs gain flexibility and improved performance on tasks related to quantum spin liquids.
However, the paper does not provide a thorough analysis of the limitations of this approach. For example, it is unclear how well ASNNs would scale to larger or more complex quantum systems, or how sensitive they are to hyperparameter choices and other architectural details.
Additionally, the authors do not discuss potential issues with the interpretability or "explainability" of the learned symmetry transformations within the network. As these models become more complex, it may become increasingly difficult to understand and trust their inner workings.
Further research could explore the generalization of ASNNs to other types of quantum many-body problems, as well as investigate ways to improve the transparency and interpretability of these models. Novel convolutional neural network architectures for continuous symmetries may also provide useful insights for advancing the state-of-the-art in this area.
Conclusion
This paper introduces a new approach to designing neural networks that can more effectively capture the symmetries of many-body quantum systems, particularly quantum spin liquids. The proposed "approximately-symmetric neural networks" (ASNNs) are shown to outperform traditional neural networks and some specialized equivariant architectures on tasks related to modeling these exotic quantum states of matter.
By allowing the network to learn the relevant symmetries rather than enforcing them explicitly, ASNNs gain flexibility and improved performance. This work represents an important step forward in the effort to develop neural network models that can accurately simulate and understand the complex behavior of quantum many-body systems, with potential applications in materials science, quantum computing, and beyond.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Papers
🧠
Theory for Equivariant Quantum Neural Networks
Quynh T. Nguyen, Louis Schatzki, Paolo Braccia, Michael Ragone, Patrick J. Coles, Frederic Sauvage, Martin Larocca, M. Cerezo
0
0
Quantum neural network architectures that have little-to-no inductive biases are known to face trainability and generalization issues. Inspired by a similar problem, recent breakthroughs in machine learning address this challenge by creating models encoding the symmetries of the learning task. This is materialized through the usage of equivariant neural networks whose action commutes with that of the symmetry. In this work, we import these ideas to the quantum realm by presenting a comprehensive theoretical framework to design equivariant quantum neural networks (EQNN) for essentially any relevant symmetry group. We develop multiple methods to construct equivariant layers for EQNNs and analyze their advantages and drawbacks. Our methods can find unitary or general equivariant quantum channels efficiently even when the symmetry group is exponentially large or continuous. As a special implementation, we show how standard quantum convolutional neural networks (QCNN) can be generalized to group-equivariant QCNNs where both the convolution and pooling layers are equivariant to the symmetry group. We then numerically demonstrate the effectiveness of a SU(2)-equivariant QCNN over symmetry-agnostic QCNN on a classification task of phases of matter in the bond-alternating Heisenberg model. Our framework can be readily applied to virtually all areas of quantum machine learning. Lastly, we discuss about how symmetry-informed models such as EQNNs provide hopes to alleviate central challenges such as barren plateaus, poor local minima, and sample complexity.
5/14/2024
Quantum consistent neural/tensor networks for photonic circuits with strongly/weakly entangled states
Nicolas Allegra
0
0
Modern quantum optical systems such as photonic quantum computers and quantum imaging devices require great precision in their designs and implementations in the hope to realistically exploit entanglement and reach a real quantum advantage. The theoretical and experimental explorations and validations of these systems are greatly dependent on the precision of our classical simulations. However, as Hilbert spaces increases, traditional computational methods used to design and optimize these systems encounter hard limitations due to the quantum curse of dimensionally. To address this challenge, we propose an approach based on neural and tensor networks to approximate the exact unitary evolution of closed entangled systems in a precise, efficient and quantum consistent manner. By training the networks with a reasonably small number of examples of quantum dynamics, we enable efficient parameter estimation in larger Hilbert spaces, offering an interesting solution for a great deal of quantum metrology problems.
6/5/2024
🧠
Unifying O(3) Equivariant Neural Networks Design with Tensor-Network Formalism
Zimu Li, Zihan Pengmei, Han Zheng, Erik Thiede, Junyu Liu, Risi Kondor
0
0
Many learning tasks, including learning potential energy surfaces from ab initio calculations, involve global spatial symmetries and permutational symmetry between atoms or general particles. Equivariant graph neural networks are a standard approach to such problems, with one of the most successful methods employing tensor products between various tensors that transform under the spatial group. However, as the number of different tensors and the complexity of relationships between them increase, maintaining parsimony and equivariance becomes increasingly challenging. In this paper, we propose using fusion diagrams, a technique widely employed in simulating SU($2$)-symmetric quantum many-body problems, to design new equivariant components for equivariant neural networks. This results in a diagrammatic approach to constructing novel neural network architectures. When applied to particles within a given local neighborhood, the resulting components, which we term fusion blocks, serve as universal approximators of any continuous equivariant function defined in the neighborhood. We incorporate a fusion block into pre-existing equivariant architectures (Cormorant and MACE), leading to improved performance with fewer parameters on a range of challenging chemical problems. Furthermore, we apply group-equivariant neural networks to study non-adiabatic molecular dynamics of stilbene cis-trans isomerization. Our approach, which combines tensor networks with equivariant neural networks, suggests a potentially fruitful direction for designing more expressive equivariant neural networks.
5/24/2024
Training-efficient density quantum machine learning
Brian Coyle, El Amine Cherrat, Nishant Jain, Natansh Mathur, Snehal Raj, Skander Kazdaghli, Iordanis Kerenidis
0
0
Quantum machine learning requires powerful, flexible and efficiently trainable models to be successful in solving challenging problems. In this work, we present density quantum neural networks, a learning model incorporating randomisation over a set of trainable unitaries. These models generalise quantum neural networks using parameterised quantum circuits, and allow a trade-off between expressibility and efficient trainability, particularly on quantum hardware. We demonstrate the flexibility of the formalism by applying it to two recently proposed model families. The first are commuting-block quantum neural networks (QNNs) which are efficiently trainable but may be limited in expressibility. The second are orthogonal (Hamming-weight preserving) quantum neural networks which provide well-defined and interpretable transformations on data but are challenging to train at scale on quantum devices. Density commuting QNNs improve capacity with minimal gradient complexity overhead, and density orthogonal neural networks admit a quadratic-to-constant gradient query advantage with minimal to no performance loss. We conduct numerical experiments on synthetic translationally invariant data and MNIST image data with hyperparameter optimisation to support our findings. Finally, we discuss the connection to post-variational quantum neural networks, measurement-based quantum machine learning and the dropout mechanism.
5/31/2024