pith. sign in

arxiv: 2605.18351 · v1 · pith:WARVR67Nnew · submitted 2026-05-18 · 💻 cs.NE

Mapping the Fitness Landscape: A Structure-Guided Approach to Multi-Modal Optimization

Meng Xiang , Pei Yan This is my paper

Pith reviewed 2026-05-19 23:31 UTC · model grok-4.3

classification 💻 cs.NE
keywords multimodal optimizationniching evolutionary algorithmsfitness landscapebasin identificationpersistence-guided mergingk-nearest neighbor graphchaotic map exploration
0
0 comments X

The pith

Chaotic decoding and persistence-guided merging on a kNN graph recovers the basin structure of multimodal landscapes to locate more distinct peaks.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Most niching evolutionary algorithms rely on distance or density rules that allow many individuals to collapse into the same basin, producing pseudo-multimodality instead of true coverage of multiple optima. The paper introduces CLDE, which converts multimodal search into an explicit decode-value-allocate-refine loop: a logistic chaotic map drives global exploration with decaying steps, a k-nearest-neighbor graph is built on the decoded points, and persistence-guided basin growing merges peaks only when they lack deep valleys between them. An adaptive persistence threshold continuously adjusts the resolution to limit both over-fragmentation and over-merging. The recovered basin structure then directs basin-wise selection and refinement, preserving coverage while raising solution quality inside each basin.

Core claim

CLDE is a decision-space-centric framework that turns multimodal optimization into a closed decode–value–allocate–refine loop. It injects controlled global exploration via a logistic chaotic map with decaying step size, constructs a k-nearest-neighbor graph on a decoding canvas, and applies persistence-guided basin growing that merges peaks only when they are not separated by deep valleys. An adaptive persistence threshold tunes the decoding resolution online. Guided by the decoded structure, the method performs basin-wise selection and refinement to improve solution quality while preserving basin coverage.

What carries the argument

Persistence-guided basin growing on the k-nearest-neighbor graph, which merges peaks only across shallow valleys and uses an adaptive persistence threshold to recover the underlying peak-basin organization without over-fragmentation or over-merging.

If this is right

  • CLDE-S attains higher peak ratios than prior niching methods on the 20 CEC2013 multimodal functions under the same evaluation budget.
  • CLDE-M achieves competitive or superior IGD and IGDx values on the DTLZ and MMMOP suites, with larger gains on problems that have strongly multimodal structure.
  • Basin-wise selection and refinement improves local solution quality inside each recovered basin while the decoded structure prevents loss of coverage across basins.
  • The adaptive persistence threshold prevents both excessive splitting of single basins and erroneous merging of distinct basins during online search.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same decode-value-allocate-refine loop could be inserted into other population-based methods to reduce pseudo-multimodality without redesigning their core operators.
  • If the kNN graph and persistence computation remain tractable, the approach may extend to higher-dimensional decision spaces where density estimators alone become unreliable.
  • Real-world expensive multimodal problems could benefit from fewer total evaluations once the basin organization is decoded early in the run.

Load-bearing premise

The k-nearest-neighbor graph combined with persistence-guided merging accurately recovers the true underlying peak-basin organization of the decision space without over-fragmentation or over-merging.

What would settle it

Apply CLDE to a synthetic multimodal test function whose exact peak locations and basin boundaries are known in advance, then measure whether the recovered basins match the ground-truth partition and whether the achieved peak ratio exceeds that of standard distance-based niching methods under identical evaluation budgets.

Figures

Figures reproduced from arXiv: 2605.18351 by Meng Xiang, Pei Yan.

Figure 1
Figure 1. Figure 1: Overview of the CLDE evolutionary cycle with three synergistic modules. (A) Persistence-guided region discovery: [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Convergence speed and success statistics in a worst [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Persistence-guided basin saliency and resource al [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Runtime comparison on CEC2013 (F1–F20). We report the wall-clock time of CLDE_full and three ablation variants [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Average runtime (mean ± std) on MMMOP1– MMMOP6 under the same evaluation budget. The compara￾ble runtime indicates that the decision-space gains of CLDE￾M are not achieved by extra computational overhead. Taken together, the DTLZ and MMMOP results demonstrate that CLDE-M can achieve competitive objective-space convergence (IGD) while better preserving multiple decision-space Pareto fam￾ilies (IGD𝑥 ), valid… view at source ↗
Figure 6
Figure 6. Figure 6: (Figure 6) Population snapshots of CLDE-S on four representative 2-D CEC2013 functions. The contour background [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Bifurcation diagram of the logistic map 𝑥𝑡+1 = 𝜇𝑥𝑡 (1 − 𝑥𝑡 ). As 𝜇 increases, the system transitions from stabil￾ity (𝜇 < 3) to period-doubling (3 ≤ 𝜇 < 3.57) and then to fully developed chaos (𝜇 ≳ 3.57). The fully chaotic regime provides quasi-random and decorrelated perturbations, motivating the use of 𝜇 = 4 as a strong exploration setting in the CE module. To justify the default chaotic intensity adopte… view at source ↗
Figure 8
Figure 8. Figure 8: (i) the final best objective value (lower is better) and (ii) [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 8
Figure 8. Figure 8: Effect of the chaos parameter 𝜇 on optimization performance and diversity. The left panel reports the final best objective value (lower is better) and the right panel reports the median pairwise distance (higher is better), averaged over repeated runs under the same evaluation budget. The period-doubling range yields unstable performance and reduced diversity, whereas the fully chaotic regime improves both… view at source ↗
read the original abstract

Multimodal optimization requires finding many optima rather than merely keeping a diverse population. Yet most niching-based evolutionary algorithms rely on distances or density estimators without explicitly recovering the underlying peak--basin organization in the decision space, which can lead to pseudo-multimodality: many distinct individuals ultimately collapse into only a few basins. We introduce Chaotic Landscape-Decoding Evolution (CLDE), a decision-space-centric framework that turns multimodal search into a closed loop of decode--value--allocate--refine. CLDE injects controlled global exploration via a logistic chaotic map with a decaying step size, then builds a $k$-nearest-neighbor graph on a decoding canvas and performs persistence-guided basin growing that merges peaks only when they are not separated by deep valleys. An adaptive persistence threshold continuously tunes the decoding resolution online to avoid over-fragmentation and over-merging. Guided by the decoded structure, CLDE carries out basin-wise selection and refinement to improve solution quality while preserving basin coverage. We instantiate CLDE as CLDE-S and CLDE-M for single- and multi-objective multimodal optimization. Experiments on 20 CEC2013 functions show that CLDE-S achieves strong peak ratio under the same evaluation budget, while on DTLZ and MMMOP suites CLDE-M attains competitive IGD/IGDx, with pronounced gains on strongly multimodal problems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces Chaotic Landscape-Decoding Evolution (CLDE), a decision-space-centric framework for multimodal optimization. It frames the problem as a closed decode–value–allocate–refine loop that injects exploration via a logistic chaotic map, constructs a k-nearest-neighbor graph on a decoding canvas, and applies persistence-guided basin growing with an adaptive threshold to recover peak–basin structure. Basin-wise selection and refinement then improve solution quality while preserving coverage. Instantiations CLDE-S and CLDE-M are evaluated on 20 CEC2013 functions (peak ratio) and on DTLZ/MMMOP suites (IGD/IGDx), with reported gains on strongly multimodal instances.

Significance. If the persistence-guided merging step reliably recovers the underlying basin organization, the approach could improve upon distance- or density-based niching by explicitly mapping decision-space structure, potentially reducing pseudo-multimodality. The combination of chaotic exploration and online adaptive resolution is a concrete technical contribution that could be tested on other multimodal benchmarks.

major comments (2)
  1. [Method (decode-value-allocate-refine loop)] Method section (decode–value–allocate–refine loop and persistence-guided basin growing): the claim that the k-NN graph plus persistence-guided merging recovers the true peak–basin organization without systematic over-fragmentation or over-merging is load-bearing for attributing performance gains to structure recovery. No quantitative validation (e.g., basin-count error, boundary fidelity, or recovery rate) is supplied against ground-truth multimodal landscapes whose basins are known analytically or by exhaustive sampling.
  2. [Experiments] Experimental results (CEC2013, DTLZ, MMMOP): the abstract and method description assert strong peak ratios and competitive IGD/IGDx with pronounced gains on multimodal problems, yet the supplied text contains no tables, statistical tests, error bars, or explicit baseline comparisons. Without these, the empirical claims cannot be verified and the contribution of the structure-recovery component versus the chaotic map or basin-wise selection alone remains unclear.
minor comments (2)
  1. [Title/Abstract] Title refers to 'Mapping the Fitness Landscape' while the abstract and method center on CLDE; ensure the full manuscript maintains consistent framing between title and content.
  2. [Method] Notation for the adaptive persistence threshold and its decay schedule should be defined explicitly with a single equation or pseudocode block to avoid ambiguity in the online tuning procedure.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback on our manuscript. The comments highlight important aspects regarding validation of the basin recovery mechanism and the presentation of experimental results. We address each major comment below and describe the revisions we intend to make to strengthen the paper.

read point-by-point responses

  1. Referee: [Method (decode-value-allocate-refine loop)] Method section (decode–value–allocate–refine loop and persistence-guided basin growing): the claim that the k-NN graph plus persistence-guided merging recovers the true peak–basin organization without systematic over-fragmentation or over-merging is load-bearing for attributing performance gains to structure recovery. No quantitative validation (e.g., basin-count error, boundary fidelity, or recovery rate) is supplied against ground-truth multimodal landscapes whose basins are known analytically or by exhaustive sampling.

    Authors: We agree that direct quantitative validation of the persistence-guided basin recovery against analytically known ground-truth structures would strengthen the attribution of performance gains specifically to the structure-mapping component. The current manuscript provides indirect support through superior peak ratios on the CEC2013 suite, whose functions have known numbers of global optima. To address this directly, we will add a dedicated subsection with experiments on low-dimensional synthetic landscapes (e.g., 2D and 3D Rastrigin, Griewank, and custom multimodal functions) where basins can be exhaustively enumerated. We will report basin-count error, boundary fidelity (measured via overlap with ground-truth partitions), and recovery rate. These additions will be placed in the experimental section and will help isolate the contribution of the k-NN plus persistence-guided merging step. revision: yes

  2. Referee: [Experiments] Experimental results (CEC2013, DTLZ, MMMOP): the abstract and method description assert strong peak ratios and competitive IGD/IGDx with pronounced gains on multimodal problems, yet the supplied text contains no tables, statistical tests, error bars, or explicit baseline comparisons. Without these, the empirical claims cannot be verified and the contribution of the structure-recovery component versus the chaotic map or basin-wise selection alone remains unclear.

    Authors: We apologize that the version supplied to the referee appears to have omitted the full experimental tables and statistical details. The complete manuscript contains tables reporting peak ratios for CLDE-S versus established multimodal optimizers (including NEA2, RSPSO, and others) across all 20 CEC2013 functions, as well as IGD and IGDx results for CLDE-M on DTLZ and MMMOP suites, with explicit baseline comparisons. Statistical significance is evaluated via the Wilcoxon rank-sum test (p-values reported), and figures display error bars over 30 independent runs. To further clarify the contribution of the structure-recovery component, we will add an ablation study comparing full CLDE against variants that disable the persistence-guided merging or the adaptive threshold. All tables, statistical results, and ablation analyses will be prominently included and cross-referenced in the revised manuscript. revision: yes

Circularity Check

0 steps flagged

No circularity: algorithmic description is self-contained with no equations or self-referential derivations

full rationale

The paper describes a procedural framework (decode-value-allocate-refine loop using k-NN graph and persistence-guided basin growing with adaptive threshold) but presents no mathematical equations, fitted parameters, or derivation chain that reduces outputs to inputs by construction. Performance metrics (peak ratio on CEC2013, IGD on DTLZ/MMMOP) are reported as empirical results on external benchmarks rather than predictions forced by the method's own parameters. No self-citations appear in the abstract or method summary as load-bearing justifications, and the central claim of structure recovery is presented as an algorithmic choice without uniqueness theorems or ansatzes imported from prior author work. This is a standard non-circular empirical methods paper.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

Abstract-only review limits visibility into exact parameter counts and background assumptions; the ledger reflects only elements explicitly named or implied by the method description.

free parameters (2)
  • k for nearest-neighbor graph
    Choice of neighborhood size for constructing the decoding canvas graph; value not stated in abstract.
  • initial persistence threshold and decay schedule
    Adaptive threshold for basin merging; its initialization and update rule constitute tunable elements.
axioms (1)
  • domain assumption Fitness landscapes possess a recoverable peak-basin organization that can be approximated from sampled points via graph connectivity and valley depth.
    Invoked by the persistence-guided basin growing step that decides merges only across deep valleys.

pith-pipeline@v0.9.0 · 5769 in / 1315 out tokens · 39542 ms · 2026-05-19T23:31:12.277624+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

48 extracted references · 48 canonical work pages

  1. [1]

    Bilal Alatas. 2010. Chaotic harmony search algorithms.Applied mathematics and computation216, 9 (2010), 2687–2699

    work page 2010

  2. [2]

    Riccardo Caponetto, Luigi Fortuna, Stefano Fazzino, and Maria Gabriella Xibilia

    work page

  3. [3]

    IEEE transactions on evolutionary computation7, 3 (2003), 289–304

    Chaotic sequences to improve the performance of evolutionary algorithms. IEEE transactions on evolutionary computation7, 3 (2003), 289–304

    work page 2003

  4. [4]

    Frédéric Chazal and Bertrand Michel. 2017. An Introduction to Topological Data Analysis: Fundamental and Practical Aspects for Data Scientists.arXiv preprint arXiv:1710.04019(2017)

    work page arXiv 2017

  5. [5]

    Weicheng Chen, Wenlong Zhang, and Weixuan Tang. 2019. A Novel Differen- tial Evolution for Multimodal Optimization.IEEE Transactions on Evolutionary Computation(2019)

    work page 2019

  6. [6]

    Ran Cheng, Yaochu Jin, Markus Olhofer, and Bernhard Sendhoff. 2016. A reference vector guided evolutionary algorithm for many-objective optimization.IEEE Transactions on Evolutionary Computation20, 5 (2016), 773–791

    work page 2016

  7. [7]

    Tinkle Chugh, Yaochu Jin, Kaisa Miettinen, Jussi Hakanen, and Karthik Sindhya

    work page

  8. [8]

    A surrogate-assisted reference vector guided evolutionary algorithm for computationally expensive many-objective optimization.IEEE Transactions on Evolutionary Computation22, 1 (2016), 129–142

    work page 2016

  9. [9]

    Coello Coello

    Carlos A. Coello Coello. 2007. Evolutionary Multi-objective Optimization: A Historical View of the Field.IEEE Computational Intelligence Magazine1, 1 (2007), 28–36

    work page 2007

  10. [10]

    Meyarivan

    Kalyanmoy Deb, Amrit Pratap, Sameer Agarwal, and T. Meyarivan. 2002. A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II.IEEE Transactions on Evolutionary Computation6, 2 (2002), 182–197

    work page 2002

  11. [11]

    Kalyanmoy Deb and Amit Saha. 2010. Multimodal Optimization Using a Bi- objective Evolutionary Algorithm.Evolutionary Computation18, 3 (2010), 373– 395

    work page 2010

  12. [12]

    2010.Computational Topology: An Intro- duction

    Herbert Edelsbrunner and John Harer. 2010.Computational Topology: An Intro- duction. American Mathematical Society

    work page 2010

  13. [13]

    Jonathan E Fieldsend. 2014. Running up those hills: Multi-modal search with the niching migratory multi-swarm optimiser. In2014 IEEE Congress on Evolutionary Computation (CEC). IEEE, 2593–2600

    work page 2014

  14. [14]

    Goldberg and Jon Richardson

    David E. Goldberg and Jon Richardson. 1987. Genetic Algorithms with Sharing for Multimodal Function Optimization. InProceedings of the 2nd International Conference on Genetic Algorithms. 41–49

    work page 1987

  15. [15]

    Wenhua Li, Xingyi Yao, Kaiwen Li, Rui Wang, Tao Zhang, and Ling Wang. 2023. Coevolutionary Framework for Generalized Multimodal Multi-Objective Opti- mization.IEEE/CAA Journal of Automatica Sinica10, 7 (2023), 1544–1556

    work page 2023

  16. [16]

    Wenhua Li, Xingyi Yao, Tao Zhang, Rui Wang, and Ling Wang. 2022. Hierarchy Ranking Method for Multimodal Multiobjective Optimization with Local Pareto Fronts.IEEE Transactions on Evolutionary Computation26, 1 (2022), 98–110

    work page 2022

  17. [17]

    Wenhua Li, Tao Zhang, Rui Wang, and Hisao Ishibuchi. 2021. Weighted Indicator- Based Evolutionary Algorithm for Multimodal Multiobjective Optimization.IEEE Transactions on Evolutionary Computation25, 6 (2021), 1064–1078

    work page 2021

  18. [18]

    Xiaodong Li. 2002. A Speciation-Based Genetic Algorithm for Multimodal Func- tion Optimization. InProceedings of the 2002 IEEE Congress on Evolutionary Computation (CEC). 1457–1462

    work page 2002

  19. [19]

    Xiaodong Li, Anthony Engelbrecht, and Mengjie Zhang. 2011. Evolutionary Multimodal Optimization: A Short Survey.IEEE Congress on Evolutionary Com- putation Tutorial Notes(2011)

    work page 2011

  20. [20]

    Jing Liang, Qianqian Guo, Caitong Yue, Boyang Qu, and Kunjie Yu. 2018. A self- organizing multi-objective particle swarm optimization algorithm for multimodal multi-objective problems. InAdvances in Swarm Intelligence: 9th International Conference, ICSI 2018, Shanghai, China, June 17-22, 2018, Proceedings, Part I 9. Springer, 550–560

    work page 2018

  21. [21]

    Jing J Liang, CT Yue, and Bo-Yang Qu. 2016. Multimodal multi-objective opti- mization: A preliminary study. In2016 IEEE congress on evolutionary computation (CEC). Ieee, 2454–2461. doi:10.1109/CEC.2016.7744093

    work page doi:10.1109/cec.2016.7744093 2016

  22. [22]

    Qiuzhen Lin, Xunfeng Wu, Lijia Ma, Jianqiang Li, Maoguo Gong, and Carlos A Coello Coello. 2021. An ensemble surrogate-based framework for expensive multiobjective evolutionary optimization.IEEE Transactions on Evolutionary Computation26, 4 (2021), 631–645

    work page 2021

  23. [23]

    1995.Niching methods for genetic algorithms

    Samir W Mahfoud. 1995.Niching methods for genetic algorithms. University of Illinois at Urbana-Champaign

    work page 1995

  24. [24]

    Fei Ming, Wenyin Gong, and Yaochu Jin. 2024. Growing neural gas network-based surrogate-assisted pareto set learning for multimodal multi-objective optimiza- tion.Swarm and Evolutionary Computation87 (2024), 101541

    work page 2024

  25. [25]

    Fei Ming, Wenyin Gong, Ling Wang, and Liang Gao. 2023. Balancing Convergence and Diversity in Objective and Decision Spaces for Multimodal Multi-Objective Optimization.IEEE Transactions on Emerging Topics in Computational Intelligence 7, 2 (2023), 474–486

    work page 2023

  26. [26]

    Gabriela Ochoa and Nadarajen Veerapen. 2018. Mapping the global structure of TSP fitness landscapes.Journal of Heuristics24, 3 (2018), 265–294. doi:10.1007/ s10732-017-9334-0

    work page 2018

  27. [27]

    Linqiang Pan, Cheng He, Ye Tian, Handing Wang, Xingyi Zhang, and Yaochu Jin

    work page

  28. [28]

    A classification-based surrogate-assisted evolutionary algorithm for expen- sive many-objective optimization.IEEE Transactions on Evolutionary Computation 23, 1 (2018), 74–88

    work page 2018

  29. [29]

    Yan Pei. 2014. Chaotic evolution: fusion of chaotic ergodicity and evolutionary iteration for optimization.Natural Computing13 (2014), 79–96

    work page 2014

  30. [30]

    Bo Peng, Bo Liu, Fu-Yi Zhang, and Ling Wang. 2009. Differential evolution algorithm-based parameter estimation for chaotic systems.Chaos, Solitons & Fractals39, 5 (2009), 2110–2118

    work page 2009

  31. [31]

    André Petrowski. 1996. A Clearing Procedure as a Niching Method for Ge- netic Algorithms. InProceedings of IEEE International Conference on Evolutionary Computation. 798–803

    work page 1996

  32. [32]

    Markus Pitzer and Martin Affenzeller. 2012. A Comprehensive Survey on Fitness Landscape Analysis.Swarm and Evolutionary Computation(2012)

    work page 2012

  33. [33]

    Mike Preuss. 2012. Improved topological niching for real-valued global optimiza- tion. InEuropean Conference on the Applications of Evolutionary Computation. Springer, 386–395

    work page 2012

  34. [34]

    Boyang Qu and P. N. Suganthan. 2012. Differential Evolution with Neighbor- hood Mutation for Multimodal Optimization.IEEE Transactions on Evolutionary Computation16, 5 (2012), 601–614

    work page 2012

  35. [35]

    Takumi Sonoda and Masaya Nakata. 2022. Multiple classifiers-assisted evolu- tionary algorithm based on decomposition for high-dimensional multiobjective problems.IEEE Transactions on Evolutionary Computation26, 6 (2022), 1581–1595

    work page 2022

  36. [36]

    Catalina Stoean, Remus Stoean, and Mihai Preuss. 2010. Topological Species Conservation for Global Optimization.Soft Computing14, 6 (2010), 599–618

    work page 2010

  37. [37]

    Zhiping Tan, Kangshun Li, and Yi Wang. 2021. Differential evolution with adaptive mutation strategy based on fitness landscape analysis.Information Sciences549 (2021), 142–163

    work page 2021

  38. [38]

    Rene Thomsen. 2004. Multimodal optimization using crowding-based differential evolution. InProceedings of the 2004 Congress on Evolutionary Computation (CEC). 1382–1389

    work page 2004

  39. [39]

    Ye Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin. 2017. PlatEMO: A MATLAB platform for evolutionary multi-objective optimization.IEEE Computational Intelligence Magazine12, 4 (2017), 73–87

    work page 2017

  40. [40]

    Xi Wang, Mengmeng Sheng, Kangfei Ye, Jian Lin, Jiafa Mao, Shengyong Chen, and Weiguo Sheng. 2019. A multilevel sampling strategy based memetic differential evolution for multimodal optimization.Neurocomputing334 (2019), 79–88

    work page 2019

  41. [41]

    Zi-Jia Wang, Zhi-Hui Zhan, Ying Lin, Wei-Jie Yu, Hua Wang, Sam Kwong, and Jun Zhang. 2019. Automatic niching differential evolution with contour pre- diction approach for multimodal optimization problems.IEEE Transactions on Evolutionary Computation24, 1 (2019), 114–128

    work page 2019

  42. [42]

    Zi-Jia Wang, Zhi-Hui Zhan, Jun Zhang, and Qingfu Zhang. 2018. Dual-Strategy Differential Evolution with Affinity Propagation Clustering for Multimodal Opti- mization Problems.IEEE Transactions on Evolutionary Computation22, 6 (2018), 894–908

    work page 2018

  43. [43]

    Caitong Yue, Boyang Qu, and Jing Liang. 2018. A Multiobjective Particle Swarm Optimizer Using Ring Topology for Solving Multimodal Multiobjective Problems. IEEE Transactions on Evolutionary Computation22, 5 (2018), 805–817

    work page 2018

  44. [44]

    Ivan Zelinka, Sergej Celikovsk `y, Hendrik Richter, and Guanrong Chen. 2010. Evolutionary algorithms and chaotic systems. Vol. 267. Springer

    work page 2010

  45. [45]

    Jinyuan Zhang, Aimin Zhou, and Guixu Zhang. 2015. A classification and Pareto domination based multiobjective evolutionary algorithm. In2015 IEEE congress on evolutionary computation (CEC). IEEE, 2883–2890

    work page 2015

  46. [46]

    Qingfu Zhang and Hui Li. 2007. MOEA/D: A Multiobjective Evolutionary Algo- rithm Based on Decomposition.IEEE Transactions on Evolutionary Computation 11, 6 (2007), 712–731

    work page 2007

  47. [47]

    Eckart Zitzler, Marco Laumanns, and Lothar Thiele. 2001. SPEA2: Improving the strength Pareto evolutionary algorithm.TIK report103 (2001)

    work page 2001

  48. [48]

    Feng Zou, Debao Chen, Hui Liu, Siyu Cao, Xuying Ji, and Yan Zhang. 2022. A survey of fitness landscape analysis for optimization.Neurocomputing503 (2022), 129–139. Conference’17, July 2017, Washington, DC, USA Meng Xiang and Pei Yan Supplementary materials 6 Single-Objective Multimodal Optimization 6.1 Benchmark Definitions Single-objective multimodal ben...

    work page 2022