Training-Free Quantum Architecture Search

Authors: Zhimin He, Maijie Deng, Shenggen Zheng, Lvzhou Li, Haozhen Situ

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Simulations on three VQE tasks demonstrate that TF-QAS achieves a substantial enhancement of sampling efficiency ranging from 5 to 57 times compared to state-of-the-art QAS, while also being 6 to 17 times faster. Simulation Results We evaluate the performance of TF-QAS across three variational quantum eigensolver (VQE) tasks for finding the ground states of the transverse field Ising model (TFIM), Heisenberg model, and Be H2 molecule.
Researcher Affiliation Academia 1School of Electronic and Information Engineering, Foshan University, Foshan 528000, China 2School of Mechatronic Engineering and Automation, Foshan University, Foshan 528000, China 3Peng Cheng Laboratory, Shenzhen, 518055 China 4Institute of Quantum Computing and Computer Theory, School of Computer Science and Engineering, Sun Yat-Sen University, Guangzhou 510006, China 5College of Mathematics and Informatics, South China Agricultural University, Guangzhou 510642, China
Pseudocode Yes Algorithm 1: A two-stage progressive TF-QAS
Open Source Code No The paper does not provide an explicit statement about releasing its own source code or a link to a code repository for the methodology described.
Open Datasets No The paper discusses VQE tasks for 'TFIM, Heisenberg model, and Be H2 molecule' and refers to prior work for constructing Hamiltonians, but does not provide concrete access information (links, DOIs, or specific citations for data download) for a publicly available dataset used in training.
Dataset Splits No The paper refers to a 'training set' for a baseline method (PQAS) and 'final validation' for TF-QAS outputs, but it does not specify explicit dataset split percentages or sample counts for training, validation, or testing data in its own experimental setup.
Hardware Specification Yes Simulations are executed on a computer equipped with an R9 7950X CPU @4.5GHz and a GeForce RTX 4090 GPU.
Software Dependencies No The paper states 'All numerical simulations are implemented with the Tensorcircuit package (Zhang et al. 2023) in Python.' and mentions 'Adam optimizer', but it does not provide specific version numbers for Python, Tensorcircuit, or any other software dependencies.
Experiment Setup Yes The parameters of quantum circuits are initialized with random values within the range of [ 2π, 2π], and subsequently optimized using the Adam optimizer with a learning rate of 0.01 until convergence to calculate the ground-truth performance. For expressibility evaluation, 2000 state fidelities are sampled using random gate parameters to estimate the fidelity distribution of the circuit. The number of remaining circuits filtered by the path-based proxy R is set to 5000.