[論文レビュー] Probabilistic Design of Parametrized Quantum Circuits through Local Gate Modifications

Grier M. Jones, Aviraj Newatia+4 more|arXiv (Cornell University)|Feb 12, 2026

Quantum Computing Algorithms and Architecture被引用数 0

ひとこと要約

The paper introduces Local Quantum Architecture Search (LQAS), an evolution-inspired heuristic that refines parametrized quantum circuits by local gate-level modifications to improve task-specific performance, evaluated on synthetic and quantum chemistry datasets with state-vector simulations and some real-device experiments.

ABSTRACT

Within quantum machine learning, parametrized quantum circuits provide flexible quantum models, but their performance is often highly task-dependent, making manual circuit design challenging. Alternatively, quantum architecture search algorithms have been proposed to automate the discovery of task-specific parametrized quantum circuits using systematic frameworks. In this work, we propose an evolution-inspired heuristic quantum architecture search algorithm, which we refer to as the local quantum architecture search. The goal of the local quantum architecture search algorithm is to optimize parametrized quantum circuit architectures through a local, probabilistic search over a fixed set of gate-level actions applied to existing circuits. We evaluate the local quantum architecture search algorithm on two synthetic function-fitting regression tasks and two quantum chemistry regression datasets, including the BSE49 dataset of bond separation energies for first- and second-row elements and a dataset of water conformers generated using the data-driven coupled-cluster approach. Using state-vector simulation, our results highlight the applicability of local quantum architecture search algorithm for identifying competitive circuit architectures with desirable performance metrics. Lastly, we analyze the properties of the discovered circuits and demonstrate the deployment of the best-performing model on state-of-the-art quantum hardware.

研究の動機と目的

• Motivate the design bottleneck in task-specific PQCs and the need for scalable automation.
• Propose an evolution-inspired, local search framework (LQAS) that refines existing circuit templates through probabilistic gate-level modifications.
• Evaluate LQAS on synthetic function-fitting tasks and quantum chemistry datasets; analyze performance and circuit properties.
• Demonstrate deployment of best-performing models on state-vector simulations and IBM quantum hardware as a hardware baseline.

提案手法

• Start from a hardware-efficient ansatz (HEA) baseline.
• Define a local modification vocabulary (gate addition, removal, switch, move) with associated probabilities padd, premove, pswitch, pmove.
• Iteratively sample and train modified circuits in small neighborhoods around the baseline, selecting top performers to form the next generation.
• Represent modifications probabilistically as Bernoulli variables to control exploration extent.
• Evaluate candidate ansätze on a regression task using metrics like mean-squared error (MSE) and R^2.
• Analyze the evolution of performance across iterations and study results on synthetic and chemistry datasets.

実験結果

リサーチクエスチョン

• RQ1 Can localized, probabilistic gate-level modifications discover task-specific PQCs more efficiently than global searches?
• RQ2 How does LQAS influence expressibility and performance on function-fitting and quantum chemistry regression tasks?
• RQ3 Do locally modified PQCs retain advantages when transferred from state-vector simulations to real quantum hardware?

主な発見

• LQAS yields PQCs with notably better regression performance than the baseline HEA on synthetic datasets.
• On quantum chemistry tasks (DDCC and BSE49), models improve over iterations, indicating effective local refinement.
• Best-performing circuits identified by LQAS can be deployed on current quantum hardware as a performance baseline, despite hardware noise.
• Minimal yet targeted circuit modifications (e.g., gate removals or replacements) can produce large performance gains.
• The first iteration often yields substantial gains, with further improvements in subsequent iterations.