Short selection

• T.-V. Luong, N. Melab and E.-G. Talbi. GPU Computing for Parallel Local Search Metaheuristic Algorithms. IEEE Transactions on Computers, IEEE, Vol. 62(1), pages 173-185, 2013. DOI: https://doi.org/10.1109/TC.2011.206.

This is a pioneering contribution to the design and implementation of parallel local search metaheuristics on GPU accelerators. The paper proposes a unified methodology for mapping neighborhoods, managing memory, controlling threads and reducing CPU-GPU transfer overheads. This methodology was validated on several local search methods and optimization problems, and later implemented in ParadisEO-GPU, published at ACM GECCO 2013 and nominated for the Best Paper Award.

• J. Gmys, M. Mezmaz, N. Melab and D. Tuyttens. A computationally efficient Branch-and-Bound algorithm for the permutation Flow-shop scheduling problem. European Journal of Operational Research, Elsevier, Vol. 284(3), pages 814-833, 2020. DOI: https://doi.org/10.1016/j.ejor.2020.01.039.

This paper is one of the strongest representatives of parallel exact combinatorial optimization. It investigates how three key components of Branch-and-Bound algorithms — lower bounds, branching strategies and parallelism — interact when solving the permutation flow-shop scheduling problem. The work led to the optimal resolution of two difficult Taillard instances with 500 jobs and 20 machines, open for about 25 years, and improved the best known solutions for 89 other benchmark instances.

• T. Carneiro, J. Gmys, N. Melab and D. Tuyttens. Towards ultra-scale Branch-and-Bound using a high-productivity language. Future Generation Computer Systems, Elsevier, Vol. 105, pages 196-209, 2020. DOI: https://doi.org/10.1016/j.future.2019.11.011.

This article revisits the design and implementation of distributed Branch-and-Bound algorithms using Chapel, a high-productivity language based on the PGAS programming model. The objective is to prepare exact optimization algorithms for future ultra-scale and exascale supercomputers, while balancing performance, scalability and software productivity. This work also initiated a collaboration with HPE/Cray around Chapel-based optimization.

• A. Hebbal, L. Brevault, M. Balesdent, N. Melab and E.-G. Talbi. Deep Gaussian process for multi-objective Bayesian optimization. Optimization and Engineering, Springer, Vol. 24, pages 1809-1848, 2023. DOI: https://doi.org/10.1007/s11081-022-09753-0. HAL: https://hal.science/hal-03770763.

This paper represents the surrogate-assisted and Bayesian optimization axis. It is part of Ali Hebbal’s PhD thesis on Deep Gaussian Processes for the analysis and optimization of complex systems, with applications to aerospace system design. The PhD was jointly supervised with ONERA, with E.-G. Talbi and N. Melab as academic supervisors, and L. Brevault and M. Balesdent as ONERA co-supervisors. The thesis was awarded the ONERA-TIS 2020 Best PhD Thesis / Doctoral Student Award. The paper proposes a multi-objective Bayesian optimization approach based on Deep Gaussian Processes, uncommonly exploiting correlations between objectives to improve the exploration of the search space and accelerate convergence toward the Pareto front.

• J. Rouzé, N. Melab, J. Gmys and D. Tuyttens. A parallel memetic algorithm for qubit mapping on noisy intermediate-scale quantum machines. Engineering Applications of Artificial Intelligence, Elsevier, Vol. 161, article 112081, 2025. DOI: https://doi.org/10.1016/j.engappai.2025.112081. HAL: https://hal.science/hal-05264970.

This paper addresses the qubit mapping problem on noisy intermediate-scale quantum (NISQ) machines, including both the initial placement of logical qubits and the routing operations required to satisfy hardware connectivity constraints. The proposed parallel memetic algorithm combines evolutionary search and local improvement mechanisms, linking quantum compilation challenges with metaheuristics, combinatorial optimization and high-performance computing. The explicit use of parallelism is also an important and relatively rare feature in this domain, where many qubit mapping and routing approaches remain mainly sequential.