Atom search optimization: a systematic review of current variants and applications
Authors:
- Sylvère MugemanyiSMSylvère Mugemanyi
- Zhaoyang QuZQZhaoyang Qu
- François Xavier RugemaFRFrançois Xavier Rugema
- Yunchang DongYDYunchang Dong
- Lei WangLWLei Wang
- Emmanuel MutabaziEMEmmanuel Mutabazi
- Providence HabumuremyiProvidence Habumuremyi
- Rita Clémence MutabaziRMRita Clémence Mutabazi
- Alexis MuhirwaAMAlexis Muhirwa
- Christophe BananezaCBChristophe Bananeza
- 12 April 2025
- journal article
- review article
- Published by Springer Nature in Knowledge and Information Systems
Abstract
Atom search optimization (ASO) is a renowned physics-based metaheuristic algorithm which takes inspiration from the fundamental theory of molecular dynamics and imitates the natural atomic movement characteristics in conformity with classical mechanics. ASO has demonstrated its preeminence over many well-known algorithms in solving real-world optimization problems and flexible implementation; hence, it has great recognition in academic communities. Due to this rising attraction, numerous modifications and improved variants of ASO have been designed and ASO finds applications in many fields including engineering, computer science, medicine, health sector, etc. Therefore, this study aims at reviewing and analyzing the ASO along with its variants. In this review, many reported studies based on ASO have been identified, collected, discussed and summed up. We have identified and selected 141 ASO-related papers published in recognized journals. The strengths as well as weaknesses of ASO have been discussed. The performance of ASO and its selected variants in convergence analysis was evaluated on 23 benchmark test functions with dimension 30. Furthermore, statistical analysis was performed to rank and validate the performance of ASO and its variants. The results reveal that QASO exhibits better overall effectiveness of 91.304% and ranks first in dimension 30 based on both Friedman and Kruskal–Wallis tests. The results also illustrate that the conventional ASO is the fastest algorithm amidst the compared algorithms in terms of computational time. In addition, the results show that the ESA-ASO achieves better results regarding the performance index and mean absolute error test. According to analysis of variance (ANOVA) test, Tukey HSD (honestly significant difference) post hoc test and Kruskal–Wallis test, the findings demonstrate that there are no statistically significant differences between ASO and its variants. Finally, the prospects for future trends are suggested.Keywords
This publication has 209 references indexed in Scilit:
- A hybrid machine learning technique for early prediction of lung nodules from medical images using a learning‐based neural network classifierConcurrency and Computation: Practice and Experience, 2022
- 8–10-Gene Expression-Based Atom Search for Aquaponic Lettuce Evapotranspiration Optimization Based on Photosynthetic Light Properties2022
- Multi‐objective quantum atom search optimization algorithm for electric vehicle charging station planningInternational Journal of Energy Research, 2022
- Optimum cycle length models using atom search optimization and grasshopper optimization algorithmsConcurrency and Computation: Practice and Experience, 2021
- Atom search sunflower optimization for trust‐based routing in internet of thingsInternational Journal of Numerical Modelling: Electronic Networks, Devices and Fields, 2020
- Thermal Exchange Optimization Algorithm2019
- The Plant Propagation Algorithm for Discrete Optimisation: The Case of the Travelling Salesman Problem2016
- Multi-Start Methods2006
- Iterated Local Search2006
- Thinking Capability of Saplings Growing Up Algorithm2006