TY - GEN

T1 - Fitness Landscapes of Buffer Allocation Problem in Production Lines and Genetic Algorithms Performance

AU - Dolgui, Alexandre

AU - Eremeev, Anton

AU - Sigaev, Vyatcheslav

N1 - A. Dolgui was supported by the ANR (French national agency for scientific research) project ANR-21-CE10-0019 “ReconfiDurable”. A.Eremeev was supported by the state task of the IM SB RAS project FWNF-2022-0020 and the RSF grant 21-41-09017.

PY - 2025/8/11

Y1 - 2025/8/11

N2 - We study structural properties of the buffer allocation problem from the fitness landscape perspective. We consider manufacturing flow lines with series-parallel network structure. The machines are supposed to be unreliable, their time to failure and repair time are exponentially distributed. We carry out computational experiments with local search and genetic algorithms in order to evaluate the fitness landscape properties of previously published instances and their modifications. We show that in many problem instances, several clusters of local optima can be identified. Besides that, the so-called ‘massif central’ or ‘big valley’ structure of the fitness landscape is present only partially. The performance of genetic algorithms is discussed with respect to population clustering. The crossover operator is shown to be useful on those problem instances, where the population clustering was observed and the permanent usage of crossover is recommended. This abstract for the Hot-off-the-Press track of GECCO 2025 summarizes work that has appeared in [6].

AB - We study structural properties of the buffer allocation problem from the fitness landscape perspective. We consider manufacturing flow lines with series-parallel network structure. The machines are supposed to be unreliable, their time to failure and repair time are exponentially distributed. We carry out computational experiments with local search and genetic algorithms in order to evaluate the fitness landscape properties of previously published instances and their modifications. We show that in many problem instances, several clusters of local optima can be identified. Besides that, the so-called ‘massif central’ or ‘big valley’ structure of the fitness landscape is present only partially. The performance of genetic algorithms is discussed with respect to population clustering. The crossover operator is shown to be useful on those problem instances, where the population clustering was observed and the permanent usage of crossover is recommended. This abstract for the Hot-off-the-Press track of GECCO 2025 summarizes work that has appeared in [6].

KW - Genetic algorithm

KW - buffer allocation

KW - local optima

KW - production line

KW - unreliable machines

UR - https://www.scopus.com/pages/publications/105014587400

UR - https://www.mendeley.com/catalogue/8708e9de-7723-3e98-828f-f1c17b502fd4/

U2 - 10.1145/3712255.3734243

DO - 10.1145/3712255.3734243

M3 - Conference contribution

SN - 979-8-4007-1464-1

T3 - GECCO 2025 Companion - Proceedings of the 2025 Genetic and Evolutionary Computation Conference Companion

SP - 27

EP - 28

BT - GECCO 2025 Companion - Proceedings of the 2025 Genetic and Evolutionary Computation Conference Companion

PB - Association for Computing Machinery

T2 - The Genetic and Evolutionary Computation Conference

Y2 - 14 July 2025 through 18 July 2025

ER -