Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2303.14877 · Communications Physics

Quantum approximate optimization via learning-based adaptive optimization

Lixue Cheng, Yuqin Chen, Shi‐Xin Zhang, Shengyu Zhang

Published Mar 6, 2024• 48 citations
arXivHTMLPDF

Abstract

Abstract Combinatorial optimization problems are ubiquitous and computationally hard to solve in general. Quantum approximate optimization algorithm (QAOA), one of the most representative quantum-classical hybrid algorithms, is designed to solve combinatorial optimization problems by transforming the discrete optimization problem into a classical optimization problem over continuous circuit parameters. QAOA objective landscape is notorious for pervasive local minima, and its viability significantly relies on the efficacy of the classical optimizer. In this work, we design double adaptive-region Bayesian optimization (DARBO) for QAOA. Our numerical results demonstrate that the algorithm greatly outperforms conventional optimizers in terms of speed, accuracy, and stability. We also address the issues of measurement efficiency and the suppression of quantum noise by conducting the full optimization loop on a superconducting quantum processor as a proof of concept. This work helps to unlock the full power of QAOA and paves the way toward achieving quantum advantage in practical classical tasks.

DOI
10.1038/s42005-024-01577-x

Related papers

Hybrid Quantum–Classical Optimization: Enhancing QAOA Performance via Classical Optimizer Selection

Xuezhu Chen

Hybrid Quantum–Classical Optimization combines the advantages of quantum computing and classical optimization to address complex combinatorial optimization problems. The Quantum Approximate Optimization Algorithm (QAOA), as an outstanding Variational Quantum Algorithm (VQA), shows great potential in solving combinatorial optimization problems. Classical optimization methods play a crucial role in tuning the parameters of quantum circuits, affecting the algorithm’s convergence speed and solution quality. This paper designs and simulates QAOA circuits based on the Qiskit framework, focusing on comparing the performance of three classical optimizers— Adaptive Moment Estimation (Adam), Constrained Optimization BY Linear Approximations (COBYLA), and Simultaneous Perturbation Stochastic Approximation (SPSA)—in parameter optimization. Adam and SPSA perform more stably in deeper circuits, making them suitable for high-dimensional parameter spaces, whereas COBYLA tends to get trapped in local optima and exhibits greater fluctuations under limited iterations. The results further highlight the importance of classical optimizer selection and hyperparameter tuning in enhancing the performance of hybrid quantum–classical optimization algorithms.

International Conference on Machine Learning, Computational Intelligence and Pattern Recognition (MLCIPR)•Dec 19, 2025•0 citations

Solution of SAT problems with the adaptive-bias quantum approximate optimization algorithm

Yunlong Yu, Chenfeng Cao, Xiang‐Bin Wang +2 more

The quantum approximate optimization algorithm (QAOA) is a promising method for solving certain classical combinatorial optimization problems on near-term quantum devices. When employing the QAOA to 3-SAT and Max-3-SAT problems, the quantum cost exhibits an easy-hard-easy or easy-hard pattern, respectively, as the clause density is changed. The quantum resources needed in the hard-region problems are out of reach for current noisy intermediate-scale quantum (NISQ) devices. We show by numerical simulations with up to 14 variables and analytical arguments that the adaptive-bias QAOA (ab-QAOA) greatly improves performance in the hard region of the 3-SAT problems and hard region of the Max-3-SAT problems. For similar accuracy, on average, ab-QAOA needs 3 levels for 10-variable 3-SAT problems as compared to 22 for QAOA. For 10-variable Max-3-SAT problems, the numbers are 7 levels and 62 levels. The improvement comes from a more targeted and more limited generation of entanglement during the evolution. We demonstrate that classical optimization is not strictly necessary in the ab-QAOA since local fields are used to guide the evolution. This leads us to propose an optimization-free ab-QAOA that can solve the hard-region 3-SAT and Max-3-SAT problems effectively with significantly fewer quantum gates as compared to the original ab-QAOA. Our work paves the way for realizing quantum advantages for optimization problems on NISQ devices.

arXiv:2210.02822•Physical Review Research•Jun 2, 2023•17 citations

Learning to Optimize Variational Quantum Circuits to Solve Combinatorial Problems

Sami Khairy, Ruslan Shaydulin, Lukasz Cincio +2 more

Quantum computing is a computational paradigm with the potential to outperform classical methods for a variety of problems. Proposed recently, the Quantum Approximate Optimization Algorithm (QAOA) is considered as one of the leading candidates for demonstrating quantum advantage in the near term. QAOA is a variational hybrid quantum-classical algorithm for approximately solving combinatorial optimization problems. The quality of the solution obtained by QAOA for a given problem instance depends on the performance of the classical optimizer used to optimize the variational parameters. In this paper, we formulate the problem of finding optimal QAOA parameters as a learning task in which the knowledge gained from solving training instances can be leveraged to find high-quality solutions for unseen test instances. To this end, we develop two machine-learning-based approaches. Our first approach adopts a reinforcement learning (RL) framework to learn a policy network to optimize QAOA circuits. Our second approach adopts a kernel density estimation (KDE) technique to learn a generative model of optimal QAOA parameters. In both approaches, the training procedure is performed on small-sized problem instances that can be simulated on a classical computer; yet the learned RL policy and the generative model can be used to efficiently solve larger problems. Extensive simulations using the IBM Qiskit Aer quantum circuit simulator demonstrate that our proposed RL- and KDE-based approaches reduce the optimality gap by factors up to 30.15 when compared with other commonly used off-the-shelf optimizers.

arXiv:1911.11071•Proceedings of the AAAI Conference on Artificial Intelligence•Apr 3, 2020•117 citations

Scalable Quantum Approximate Optimisation with Adaptive Bayesian Dimensionality Reduction

Amir Alizadeh, Amir Pourabdollah, Ahmad Lotfi

Quantum Approximate Optimisation Algorithm (QAOA) has emerged as a promising hybrid quantum-classical approach for tackling NP-hard combinatorial optimisation problems by combining quantum and classical computational paradigms. However, efficiently optimising QAOA parameters is notoriously challenging due to the complex, high-dimensional landscapes involved, characterised by numerous local minima and barren plateaus. Bayesian Optimisation (BO) has shown promise in addressing these challenges, yet its effectiveness diminishes significantly with increasing problem dimensionality. In this work, we adopt the Minimum Average Variance Estimation for Bayesian Optimisation (MAVE-BO) method to address this scalability bottleneck. By leveraging the low-dimensional structure inherent in QAOA parameter spaces, our approach improves optimisation efficiency and robustness. We applied our model to a practical power systems problem, optimising eigenvector centrality (EC) in an IEEE 14-bus network. The proposed method demonstrates superior performance in both efficiency and solution quality compared to classical optimisers such as standard Bayesian Optimisation (Vanilla BO), COBYLA, L-BFGS-B, and genetic algorithms. These results underscore the potential of dimensionality-aware BO frameworks for realworld quantum applications in noisy, resource-constrained environments.

IEEE International Conference on Quantum Artificial Intelligence (QAI)•Nov 2, 2025•0 citations

An adaptive quantum approximate optimization algorithm for solving combinatorial problems on a quantum computer

Zhu, Linghua, Ho Lun Tang, George S. Barron +4 more

The quantum approximate optimization algorithm (QAOA) is a hybrid variational quantum-classical algorithm that solves combinatorial optimization problems. While there is evidence suggesting that the fixed form of the standard QAOA ansatz is not optimal, there is no systematic approach for finding better ansätze. We address this problem by developing an iterative version of QAOA that is problem-tailored, and which can also be adapted to specific hardware constraints. We simulate the algorithm on a class of Max-Cut graph problems and show that it converges much faster than the standard QAOA, while simultaneously reducing the required number of CNOT gates and optimization parameters. We provide evidence that this speedup is connected to the concept of shortcuts to adiabaticity.

arXiv:2005.10258•May 20, 2020•25 citations

Adaptive quantum approximate optimization algorithm for solving combinatorial problems on a quantum computer

Linghua Zhu, Ho Lun Tang, George S. Barron +4 more

The quantum approximate optimization algorithm (QAOA) is a hybrid variational quantum-classical algorithm that solves combinatorial optimization problems. While there is evidence suggesting that the fixed form of the standard QAOA Ansatz is not optimal, there is no systematic approach for finding better Ans\"atze. We address this problem by developing an iterative version of QAOA that is problem tailored, and which can also be adapted to specific hardware constraints. We simulate the algorithm on a class of Max-Cut graph problems and show that it converges much faster than the standard QAOA, while simultaneously reducing the required number of CNOT gates and optimization parameters. We provide evidence that this speedup is connected to the concept of shortcuts to adiabaticity.

Physical Review Research•Jul 11, 2022•209 citations
Abstract Combinatorial optimization problems are ubiquitous and computationally hard to solve in general. Quantum approximate optimization algorithm (QAOA), one of the most representative quantum-classical hybrid algorithms, is designed to solve combinatorial optimization problems by transforming the discrete optimization problem into a classical optimization problem over continuous circuit parameters. QAOA objective landscape is notorious for pervasive local minima, and its viability significantly relies on the efficacy of the classical optimizer. In this work, we design double adaptive-region Bayesian optimization (DARBO) for QAOA. Our numerical results demonstrate that the algorithm greatly outperforms conventional optimizers in terms of speed, accuracy, and stability. We also address the issues of measurement efficiency and the suppression of quantum noise by conducting the full optimization loop on a superconducting quantum processor as a proof of concept. This work helps to unlock the full power of QAOA and paves the way toward achieving quantum advantage in practical classical tasks.
Hybrid Quantum–Classical Optimization combines the advantages of quantum computing and classical optimization to address complex combinatorial optimization problems. The Quantum Approximate Optimization Algorithm (QAOA), as an outstanding Variational Quantum Algorithm (VQA), shows great potential in solving combinatorial optimization problems. Classical optimization methods play a crucial role in tuning the parameters of quantum circuits, affecting the algorithm’s convergence speed and solution quality. This paper designs and simulates QAOA circuits based on the Qiskit framework, focusing on comparing the performance of three classical optimizers— Adaptive Moment Estimation (Adam), Constrained Optimization BY Linear Approximations (COBYLA), and Simultaneous Perturbation Stochastic Approximation (SPSA)—in parameter optimization. Adam and SPSA perform more stably in deeper circuits, making them suitable for high-dimensional parameter spaces, whereas COBYLA tends to get trapped in local optima and exhibits greater fluctuations under limited iterations. The results further highlight the importance of classical optimizer selection and hyperparameter tuning in enhancing the performance of hybrid quantum–classical optimization algorithms.
The quantum approximate optimization algorithm (QAOA) is a promising method for solving certain classical combinatorial optimization problems on near-term quantum devices. When employing the QAOA to 3-SAT and Max-3-SAT problems, the quantum cost exhibits an easy-hard-easy or easy-hard pattern, respectively, as the clause density is changed. The quantum resources needed in the hard-region problems are out of reach for current noisy intermediate-scale quantum (NISQ) devices. We show by numerical simulations with up to 14 variables and analytical arguments that the adaptive-bias QAOA (ab-QAOA) greatly improves performance in the hard region of the 3-SAT problems and hard region of the Max-3-SAT problems. For similar accuracy, on average, ab-QAOA needs 3 levels for 10-variable 3-SAT problems as compared to 22 for QAOA. For 10-variable Max-3-SAT problems, the numbers are 7 levels and 62 levels. The improvement comes from a more targeted and more limited generation of entanglement during the evolution. We demonstrate that classical optimization is not strictly necessary in the ab-QAOA since local fields are used to guide the evolution. This leads us to propose an optimization-free ab-QAOA that can solve the hard-region 3-SAT and Max-3-SAT problems effectively with significantly fewer quantum gates as compared to the original ab-QAOA. Our work paves the way for realizing quantum advantages for optimization problems on NISQ devices.
Quantum computing is a computational paradigm with the potential to outperform classical methods for a variety of problems. Proposed recently, the Quantum Approximate Optimization Algorithm (QAOA) is considered as one of the leading candidates for demonstrating quantum advantage in the near term. QAOA is a variational hybrid quantum-classical algorithm for approximately solving combinatorial optimization problems. The quality of the solution obtained by QAOA for a given problem instance depends on the performance of the classical optimizer used to optimize the variational parameters. In this paper, we formulate the problem of finding optimal QAOA parameters as a learning task in which the knowledge gained from solving training instances can be leveraged to find high-quality solutions for unseen test instances. To this end, we develop two machine-learning-based approaches. Our first approach adopts a reinforcement learning (RL) framework to learn a policy network to optimize QAOA circuits. Our second approach adopts a kernel density estimation (KDE) technique to learn a generative model of optimal QAOA parameters. In both approaches, the training procedure is performed on small-sized problem instances that can be simulated on a classical computer; yet the learned RL policy and the generative model can be used to efficiently solve larger problems. Extensive simulations using the IBM Qiskit Aer quantum circuit simulator demonstrate that our proposed RL- and KDE-based approaches reduce the optimality gap by factors up to 30.15 when compared with other commonly used off-the-shelf optimizers.
Quantum Approximate Optimisation Algorithm (QAOA) has emerged as a promising hybrid quantum-classical approach for tackling NP-hard combinatorial optimisation problems by combining quantum and classical computational paradigms. However, efficiently optimising QAOA parameters is notoriously challenging due to the complex, high-dimensional landscapes involved, characterised by numerous local minima and barren plateaus. Bayesian Optimisation (BO) has shown promise in addressing these challenges, yet its effectiveness diminishes significantly with increasing problem dimensionality. In this work, we adopt the Minimum Average Variance Estimation for Bayesian Optimisation (MAVE-BO) method to address this scalability bottleneck. By leveraging the low-dimensional structure inherent in QAOA parameter spaces, our approach improves optimisation efficiency and robustness. We applied our model to a practical power systems problem, optimising eigenvector centrality (EC) in an IEEE 14-bus network. The proposed method demonstrates superior performance in both efficiency and solution quality compared to classical optimisers such as standard Bayesian Optimisation (Vanilla BO), COBYLA, L-BFGS-B, and genetic algorithms. These results underscore the potential of dimensionality-aware BO frameworks for realworld quantum applications in noisy, resource-constrained environments.
The quantum approximate optimization algorithm (QAOA) is a hybrid variational quantum-classical algorithm that solves combinatorial optimization problems. While there is evidence suggesting that the fixed form of the standard QAOA ansatz is not optimal, there is no systematic approach for finding better ansätze. We address this problem by developing an iterative version of QAOA that is problem-tailored, and which can also be adapted to specific hardware constraints. We simulate the algorithm on a class of Max-Cut graph problems and show that it converges much faster than the standard QAOA, while simultaneously reducing the required number of CNOT gates and optimization parameters. We provide evidence that this speedup is connected to the concept of shortcuts to adiabaticity.
The quantum approximate optimization algorithm (QAOA) is a hybrid variational quantum-classical algorithm that solves combinatorial optimization problems. While there is evidence suggesting that the fixed form of the standard QAOA Ansatz is not optimal, there is no systematic approach for finding better Ans\"atze. We address this problem by developing an iterative version of QAOA that is problem tailored, and which can also be adapted to specific hardware constraints. We simulate the algorithm on a class of Max-Cut graph problems and show that it converges much faster than the standard QAOA, while simultaneously reducing the required number of CNOT gates and optimization parameters. We provide evidence that this speedup is connected to the concept of shortcuts to adiabaticity.