Post-Doctoral Research Visit F/M Graybox Tunneling for Multi-Objective Combinatorial Optimization: Algorithmic and HPC Perspectives

Inria; Villeneuve-d'Ascq; Sur place; EUR 32 000 - 42 000; Plein temps

Avantages offerts par ce poste

Partial transport reimbursement

Fixed timetable and annual leave

Teleworking after 6 months

Equipment provided

Social, cultural and sports events

Résumé du poste

Inria invites applications for a Post-Doctoral Research Visit focusing on graybox tunneling for multi-objective combinatorial optimization. The successful candidate will design, implement, and validate parallel tunneling algorithms and explore island-model hybrids on modern HPC platforms.

The role requires a strong background in combinatorial optimization, evolutionary computation, and parallel HPC methods, with experience in parallelization, scalability, and empirical evaluation.

Qualifications

• PhD or equivalent in combinatorial optimization, evolutionary computation, computational intelligence, or related areas.

Responsabilités

• Design graybox tunneling algorithms for multi-objective optimization.
• Develop parallel tunneling strategies and island-based optimization approaches.
• Implement and evaluate parallel tunneling algorithms on HPC platforms.
• Develop reusable software for parallel graybox tunneling and ensure reproducible experiments.

Connaissances

Combinatorial optimization

Evolutionary computation

Computational intelligence

Heuristic search

Parallel computing

High-performance computing

Description du poste

Post-Doctoral Research Visit F/M Graybox Tunneling for Multi-Objective Combinatorial Optimization: Algorithmic and HPC Perspectives

Fonction : Post-Doctorant

The postdoctoral researcher will contribute to the design, development, and experimental validation of novel graybox tunneling algorithms for multi-objective combinatorial optimization, with a particular focus on their parallelization and scalability on high-performance computing (HPC) platforms.

The main assignment will be to investigate how knowledge about the structure and local interactions of combinatorial optimization problems can be exploited to design efficient tunneling mechanisms capable of navigating large sets of Pareto local optima. In particular, the researcher will contribute to extending existing single-objective tunneling principles to the multi-objective setting and to developing new algorithmic strategies for efficiently exploring different regions of the Pareto front.

A second assignment will concern the identification and exploitation of different levels of parallelism arising within tunneling-based multi-objective search. This can include solution-level and iteration-level parallelism, as well as higher-level parallelism based on concurrent tunneling processes. For exemple, the researcher will investigate hybrid optimization strategies, notably based on distributed island models, in which different tunneling processes explore complementary regions of the search space and exchange selected Pareto local optimal solutions.

The researcher will also contribute to the implementation and empirical evaluation of the resulting algorithms on modern HPC platforms. The position is open to candidates with a strong background in combinatorial optimization, evolutionary computation, computational intelligence, heuristic search, or related areas, as well as to candidates with expertise in parallel and HPC computing combined with a solid understanding of combinatorial optimization. While prior HPC experience is an advantage for candidates from optimization and computational intelligence backgrounds, candidates primarily specialized in HPC are also encouraged to apply provided they have relevant expertise in combinatorial optimization.

The work will ultimately contribute to the development of prototype software libraries for parallel tunneling and to systematic computational studies demonstrating their effectiveness and scalability on multi-objective combinatorial optimization problems, such as, NK-landscapes, SAT, TSP.

The main activities of the postdoctoral researcher will include:

• Design of graybox tunneling algorithms for multi-objective combinatorial optimization, exploiting structural information about problem variables and their interactions to efficiently navigate sets of Pareto local optima.
• Development of parallel tunneling strategies, identifying and exploiting solution-level, iteration-level, and algorithm-level sources of parallelism to enhance the exploration of large combinatorial search spaces.
• Design of hybrid island-based optimization approaches, enabling multiple tunneling processes to concurrently explore complementary regions of the Pareto front and exchange selected Pareto local optima.
• Implementation and scalability analysis of parallel tunneling algorithms on HPC platforms, initially targeting multi-core and multi-CPU architectures andpossibly investigating the use of many-core architectures where appropriate.
• Experimental validation and comparative evaluationof the proposed methods on multi-objective combinatorial optimization problems.

The additional activities of the postdoctoral researcher will include:

• Development of reusable softwarefor parallel graybox tunneling, supporting large-scale and reproducible computational experiments.
• Scientific dissemination, including preparation of research papers, conference presentations, technical reports, etc
• Research collaboration and knowledge exchange withresearchers in combinatorial optimization, evolutionary computation, computational intelligence, and high-performance computing.

The position is open to candidates with a strong background in combinatorial optimization, evolutionary computation, computational intelligence, heuristic search, or related areas, as well as to candidates with expertise in parallel and HPC computing combined with a solid understanding of combinatorial optimization. While prior HPC experience is an advantage for candidates from optimization and computational intelligence backgrounds, candidates primarily specialized in HPC are also encouraged to apply provided they have relevant expertise in combinatorial optimization.

Avantages

• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities