arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.11022cond-mat.str-elcond-mat.dis-nncond-mat.stat-mechquant-ph
收藏

量子一维ANNNI模型的相图探索

Exploring the Phase Diagram of the quantum one-dimensional ANNNI model

Max-Plank-Institut für Quantenoptik(马克斯·普朗克量子光学研究所);Munich Center for Quantum Science and Technology (MCQST)(慕尼黑量子科学与技术中心);European Organization for Nuclear Research (CERN)(欧洲核子研究组织);RWTH Aachen University(亚琛工业大学);Deutsches Elektronen-Synchrotron (DESY)(德国电子同步辐射中心);University of the Basque Country UPV/EHU(巴斯克大学);Donostia International Physics Center(圣塞巴斯蒂安国际物理中心);IKERBASQUE, Basque Foundation for Science(伊克尔巴斯基克,巴斯克科学基金会);Institute of Fundamental Physics IFF-CSIC(基础物理研究所)

M. Cea, M. Grossi, S. Monaco, E. Rico, L. Tagliacozzo, S. Vallecorsa

AI总结:

该工作将张量网络与量子机器学习集成,通过经典态制备电路实现可扩展的量子-经典混合流水线,用于研究一维ANNNI模型的相图。

摘要:

这项工作在带横向场的一维轴向次近邻伊辛模型的背景下,提出了量子机器学习与张量网络的电路级集成。该研究具体地将这两种范式联系起来:模型的基态通过密度矩阵重整化群获得,表示为矩阵乘积态,并转换为量子电路,然后通过将态制备电路前置到可训练拟设上,用作混合量子-经典分类器和量子自编码器的输入。这种构造直接解决了纯量子工作流程的一个关键瓶颈,即每个输入态都需要在硬件上进行昂贵的基于梯度的优化,通过将态制备委托给可扩展的经典张量网络方法来解决。其结果是用于研究自旋系统的可扩展且与硬件兼容的流水线。

英文摘要:

This work presents a circuit-level integration of Quantum Machine Learning and Tensor Networks in the context of the one-dimensional Axial Next-Nearest-Neighbor Ising model with a transverse field. The study concretely connects these two paradigms: ground states of the model are obtained via the density matrix renormalization group as Matrix Product States and converted into quantum circuits, which are then used as input to a hybrid quantum-classical classifier and a Quantum Autoencoder by prepending the state-preparation circuit to the trainable ansatz. This construction directly addresses a key bottleneck of purely quantum workflows, namely the need for expensive, gradient-based optimization on hardware for every input state, by delegating state preparation to classical Tensor Network methods, which are scalable by construction. The result is a scalable and hardware-compatible pipeline for the task of studying spin systems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑