Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

This survey argues that LLM-for-optimization research is best organized as a single modeling-to-solving workflow, and claims to be the first to systematically cover all four roles LLMs can play.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-04 20:52 UTC pith:BNV7CI5Z

load-bearing objection A useful, well-organized survey with a coherent taxonomy, but the 'first systematic coverage' claim rests on an undocumented literature search and a generous reading of prior surveys. the 3 major comments →

arxiv 2509.08269 v6 pith:BNV7CI5Z submitted 2025-09-10 cs.NE cs.AI

A Systematic Survey on Large Language Models for Evolutionary Optimization: From Modeling to Solving

Yisong Zhang , Ran Cheng , Guoxing Yi , Kay Chen Tan This is my paper
classification cs.NE cs.AI
keywords large language modelsevolutionary computationoptimization modelingoptimization solvingtaxonomyalgorithm generationalgorithm selectionsurvey
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This survey claims that research on large language models for evolutionary optimization is currently fragmented, with each prior survey covering only a slice of the field. It proposes a workflow-oriented taxonomy that divides the literature into two stages—LLMs for optimization modeling and LLMs for optimization solving—and then splits solving into three paradigms: LLMs as stand-alone optimizers, low-level components embedded inside evolutionary algorithms, and high-level orchestrators that select or generate algorithms. Against earlier reviews, it claims to be the first to give systematic coverage to all four categories, and it uses that taxonomy to organize representative methods, benchmarks, applications, and open problems. A sympathetic reader would care because the taxonomy gives the field a common language and makes the missing links visible, especially the gap between modeling and solving and the absence of dynamic, self-evolving approaches.

Core claim

The central claim is taxonomic: the many ways LLMs touch optimization are not independent tricks but stations on a single pipeline. First, LLMs can translate natural-language problem descriptions into mathematical models—via prompt-based routes (two-stage, multi-agent, interactive) or learning-based routes (synthetic data plus fine-tuning). Second, at the solving stage, an LLM can act as a stand-alone optimizer that iterates over candidate solutions in dialogue; a low-level component that performs initialization, operator, configuration, or surrogate-evaluation duties inside an evolutionary algorithm; or a high-level manager that selects among algorithms or generates new heuristic code. The

What carries the argument

The load-bearing object is the workflow-oriented taxonomy itself, defined by two axes: the stage (modeling vs solving) and, within solving, the level of involvement (stand-alone optimizer, low-level embedded component, high-level orchestrator). The solving level is anchored to evolutionary algorithms: low-level LLMs sit inside the EA loop (initialization, operators, configuration, evaluation), while high-level LLMs act on the algorithm as a whole (selection, generation). The taxonomy does the work of sorting roughly a hundred cited methods into four boxes and exposing which boxes are mature and which are nearly empty, which in turn drives the survey's claims about research gaps.

Load-bearing premise

The survey assumes that the papers it selected are representative and complete and that its reading of each cited work is accurate, but it does not describe its literature search or inclusion criteria; if notable work was missed or misclassified, the taxonomy and the claim to first systematic coverage weaken.

What would settle it

Re-run the literature survey with a documented search query, code every retrieved paper into the four categories independently, and check for a substantial cluster that fits none of them—for example, LLMs used to synthesize fitness functions or to generate training data for learned optimizers. Any sizable uncodable cluster, or low agreement between independent coders, would refute the claim that the taxonomy has systematic coverage. A simpler falsifier would be uncovering an earlier survey that already contains all four categories.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • The four-category taxonomy gives researchers a common vocabulary; a new method can be located as modeling, stand-alone solving, low-level, or high-level, making comparisons and transfers across papers easier.
  • Treating modeling and solving as one workflow reframes progress: work that only improves modeling but leaves solving to external solvers is incomplete, and end-to-end LLM-driven pipelines become the natural goal.
  • Because the solving stage is split by level of involvement, hybrid designs (LLMs called only when population improvement stalls) are not outliers but a recognizable strategy within the low-level paradigm.
  • The claimed coverage makes the empty cells actionable: dynamic algorithm selection and generation, and self-evolving workflows, are the directions the survey explicitly points to.

Where Pith is reading between the lines

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

  • If the taxonomy wins acceptance, a likely practical consequence is that benchmark suites will be organized by the same four categories, making it easier to measure progress within each role rather than across a mixed bag.
  • The modeling/solving boundary is likely to blur: frameworks such as OptiMUS already combine modeling with solver invocation, and the survey's own end-to-end vision implies that a fifth category—LLMs as full-pipeline orchestrators—may eventually absorb the first four.
  • A testable extension would be to code every paper from a fixed corpus into the four categories and measure inter-annotator agreement; low agreement would indicate the categories need sharper definitions rather than that the field is unorganizable.
  • The 'first systematic coverage' claim is only as strong as the literature search behind it; since the paper does not describe its search protocol, a reader should treat completeness as an assertion to verify rather than a settled fact.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This survey proposes a workflow-oriented taxonomy of LLM use in optimization. It separates the literature into LLMs for optimization modeling and LLMs for optimization solving, with the latter subdivided into LLMs as stand-alone optimizers, low-level LLMs embedded in evolutionary algorithm components, and high-level LLMs for algorithm selection and generation. It further reviews representative methods, benchmarks, applications, and future directions, and maintains a public GitHub repository of related literature. The paper's central novelty claim is that it offers the first systematic coverage of all four categories in a single framework, as stated in Table I and Section II.

Significance. The proposed taxonomy is coherent and likely to be useful to researchers entering this fast-moving area. The paper brings together a large and recent corpus, including many 2024–2025 preprints, and the GitHub repository is a practical contribution. If the coverage claim could be substantiated, the survey would serve as a valuable reference. However, the claim of 'first systematic coverage' is currently not verifiable because the literature selection methodology is absent, the scope is ambiguous with respect to evolutionary vs. general optimization, and several load-bearing conclusions rest on the authors' own prior work without independent corroboration. These issues are fixable in a revision, but they are central to the paper's stated contribution.

major comments (3)
  1. [Section II, Table I] The central 'first systematic coverage' claim is not operationally supported. The manuscript does not describe the search databases, query terms, inclusion/exclusion criteria, screening steps, time window, or duplicate handling used to assemble the literature. Tables II and III label works as 'representative' but provide no selection protocol, and the GitHub repository is not a substitute for a documented methodology. As a result, a reader cannot distinguish a genuine gap in the literature from a gap in the search. Add a methodology subsection (or appendix) and either substantiate or soften the 'first' claim.
  2. [Footnote 1; Sections V-A, V-C] The scope statement is inconsistent with the material covered. Footnote 1 says the survey 'primarily focuses on evolutionary optimization', yet Section V-A surveys general-purpose LLM optimizers such as OPRO [28] and POM [114] without any evolutionary component, and Section V-C covers algorithm selection and generation that are not inherently evolutionary. This ambiguity weakens the title and the claimed 'unified modeling-to-solving' framework. The authors should either broaden the scope statement to cover LLMs for optimization generally, with EAs as one instantiation, or narrow the inclusion criteria and explain why non-evolutionary works are retained.
  3. [Section V-D; Table III] Several load-bearing limitations are supported primarily by the authors' own prior work. For example, the claim that LLMs as optimizers 'often struggle to outperform classical algorithms' rests on [29], the model-dependency conclusion on [117], the operator-selection behavior on [145], and the constrained multi-objective findings on [134]. These are preprints or recent workshop papers from the same group, and the survey does not discuss contradictory evidence or how representative these evaluations are. Please add a critical assessment of the evidence base, flag where conclusions rest on a single study, and clearly separate established findings from initial observations.
minor comments (5)
  1. [Table III] Header typo: 'Optmization Algorithms' should be 'Optimization Algorithms'.
  2. [Section IV-B] 'regular match correction function' is likely a typo for 'regular expression matching correction'; please clarify.
  3. [Section I] The text after 'enhance performance; .' contains a stray period and semicolon; please clean up the punctuation.
  4. [Table I] The binary checkmarks do not convey the degree of coverage; for example, [34] is marked as not covering modeling even though the text admits modeling was 'briefly introduced'. A short footnote explaining what a checkmark means would improve precision.
  5. [References/Table III venue entries] Several venue labels appear inconsistent with the reference list: [111] is listed as KDD 2023 but the reference is an arXiv preprint; [112] is listed as GECCO 2024 but the reference is an arXiv preprint. Please verify all venue fields.

Circularity Check

0 steps flagged

No significant circularity: the survey's taxonomy is a classification of external literature, not a derived result.

full rationale

The paper is a systematic literature survey. Its central claims—the LM/LO/LL/HL taxonomy and the assertion of first systematic coverage—are organizational statements about published work, not quantities derived from fitted parameters or from the paper's own definitions. There is no equation whose output is an input in disguise, no fitted parameter renamed as a prediction, and no uniqueness theorem imported from the authors' prior work. The closest self-referential element is Section II/Table I, where the gap is defined by contrasting with prior surveys, some of which (Wu et al. [30]; Huang et al. [29]) share authors with this paper. However, those citations are to published, externally checkable surveys, and the comparison consists of descriptive scope claims rather than an unverified premise that the present paper must assume to make its argument. The taxonomy is defined independently in Section II and then applied to the cited literature; the conclusion does not reduce to the definitions. A genuine weakness is that the literature selection methodology (databases, inclusion criteria, screening) is undocumented, which undermines reproducibility and completeness of the 'first systematic coverage' claim, but this is an external-validity/correctness concern, not circularity under the specified patterns. Therefore the circularity score is minimal (2).

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

As a survey, the paper introduces no free parameters or new entities. Its central claim relies on the representativeness of the literature selection and the accuracy of the authors' categorization, neither of which is backed by a formal reproducibility mechanism.

axioms (3)
  • domain assumption The selected papers are representative of the entire field of LLMs for evolutionary optimization.
    The survey does not state a systematic search protocol, so representativeness is an implicit assumption. This underlies the completeness of the taxonomy.
  • domain assumption The proposed categories (modeling, solving; optimizers, low-level, high-level) are mutually exclusive and exhaustively cover the field.
    The classification is imposed by the authors without formal criteria for assignment, and a different reviewer might draw boundaries differently.
  • domain assumption The summaries of prior surveys in Table I are accurate representations of those works.
    The comparison and the novelty claim depend on the authors' reading of [30], [31], [32], [33], [34], and [35].

pith-pipeline@v1.3.0-alltime-deepseek · 36248 in / 6267 out tokens · 67712 ms · 2026-08-04T20:52:06.884526+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of A Systematic Survey on Large Language Models for Evolutionary Optimization: From Modeling to Solving." pith.science (2026). https://pith.science/paper/BNV7CI5Z

@misc{pith2026250908269,
  author       = {Pith},
  title        = {Pith review of: A Systematic Survey on Large Language Models for Evolutionary Optimization: From Modeling to Solving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BNV7CI5Z}},
  note         = {Machine review of arXiv:2509.08269}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Large language models (LLMs) are increasingly integrated with evolutionary computation to support optimization tasks. This survey primarily focuses on evolutionary optimization, i.e., optimization based on evolutionary computation. For brevity, we use the term optimization throughout to denote this scope. However, existing surveys typically examine isolated roles of LLMs and do not provide a unified view that connects optimization modeling with optimization solving. To address this gap, we systematically review recent developments through a workflow-oriented framework. First, we organize the literature into two primary stages: LLMs for optimization modeling and LLMs for optimization solving (in this survey, the terms optimization modeling and optimization solving are used as concise forms of optimization problem modeling and optimization problem solving, respectively). Second, we divide the solving stage into three paradigms according to the role of the LLM: stand-alone optimizers, low-level components embedded in optimization algorithms, and high-level managers for algorithm selection and generation. Third, we analyze representative methods, identify their technical limitations, and clarify their relationships with traditional optimization approaches. We further substantiate this taxonomy through benchmark systematization, baseline comparisons, and practitioner-oriented guidance, and we review interdisciplinary applications across the natural sciences, engineering, and machine learning. Based on the resulting analysis, we identify research directions toward dynamic, self-evolving, and agentic optimization ecosystems. An up-to-date collection of related literature is maintained at https://github.com/ishmael233/LLM4OPT.

Figures

Figures reproduced from arXiv: 2509.08269 by Guoxing Yi, Kay Chen Tan, Ran Cheng, Yisong Zhang.

Figure 1
Figure 1. Figure 1: Overall organization of this survey. standalone optimizers often fail to surpass classical algorithms in performance [29], and although fine-tuning with domain￾specific data can enhance task outcomes, constructing such datasets remains costly and expertise-intensive [27]. Against this backdrop, this survey provides a systematic analysis and guide for this emerging interdisciplinary field. The main contribu… view at source ↗
Figure 2
Figure 2. Figure 2: Technological dependencies of LLMs for optimization, including EA paradigm and workflow, LLM architecture, and related enabling technologies. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of LLMs for optimization modeling. Approaches can be broadly divided into two categories: (i) prompt-based methods, typically implemented [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of LLMs as optimizers. This paradigm primarily relies [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of low-level LLMs for optimization algorithms. LLMs can be applied at various stages within EAs, including initialization, evolutionary [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of high-level LLM-assisted optimization algorithms. Algorithm selection involves two key stages: feature extraction and selector construction. [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Search Hardness-Aware LLM-Based Problem Formulation for Expensive Simulation-Driven Design

    cs.NE 2026-07 conditional novelty 6.0

    SHA-PF uses initial simulation data to select a 'hard but promising' anchor satisfaction state and evolves LLM-generated formulations that prioritize it, reaching target designs with fewer expensive simulations on ant...

  2. Large Language Models for Operations Research: A Comprehensive Survey

    math.OC 2026-05 unverdicted novelty 2.0

    A survey compiling roles, applications, benchmarks, challenges, and future directions for large language models in operations research.

Reference graph

Works this paper leans on

242 extracted references · 21 canonical work pages · cited by 2 Pith papers · 17 internal anchors

  1. [1]

    Evolutionary complex engineering optimization: opportunities and challenges,

    T. Chai, Y . Jin, and S. Bernhard, “Evolutionary complex engineering optimization: opportunities and challenges,”IEEE Computational In- telligence Magazine, vol. 8, no. 3, pp. 12–15, 2013

    2013

  2. [2]

    Economic dispatch using particle swarm optimization: A review,

    A. Mahor, V . Prasad, and S. Rangnekar, “Economic dispatch using particle swarm optimization: A review,”Renewable and Sustainable Energy Reviews, vol. 13, no. 8, pp. 2134–2141, 2009

    2009

  3. [3]

    Black-box optimization for automated discovery,

    K. Terayama, M. Sumita, R. Tamura, and K. Tsuda, “Black-box optimization for automated discovery,”Accounts of Chemical Research, vol. 54, no. 6, pp. 1334–1346, 2021

    2021

  4. [4]

    A brief introduction to exact, approximation, and heuristic algorithms for solving hard combinatorial optimization problems,

    P. Festa, “A brief introduction to exact, approximation, and heuristic algorithms for solving hard combinatorial optimization problems,” in Proceedings of the International Conference on Transparent Optical Networks, 2014, pp. 1–20

    2014

  5. [5]

    No free lunch theorems for optimization,

    D. H. Wolpert and W. G. Macready, “No free lunch theorems for optimization,”IEEE Transactions on Evolutionary Computation, vol. 1, no. 1, pp. 67–82, 2002

    2002

  6. [6]

    Paramils: An automatic algorithm configuration framework,

    F. Hutter, H. H. Hoos, K. Leyton-Brown, and T. St ¨utzle, “Paramils: An automatic algorithm configuration framework,”Journal of Artificial Intelligence Research, vol. 36, pp. 267–306, 2009

    2009

  7. [7]

    Automated algorithm selection: Survey and perspectives,

    P. Kerschke, H. H. Hoos, F. Neumann, and H. Trautmann, “Automated algorithm selection: Survey and perspectives,”Evolutionary Computa- tion, vol. 27, no. 1, pp. 3–45, 2019

    2019

  8. [8]

    Heuristics and their design: A survey,

    H. M ¨uller-Merbach, “Heuristics and their design: A survey,”European Journal of Operational Research, vol. 8, no. 1, pp. 1–23, 1981

    1981

  9. [9]

    Reinforcement learning: A survey,

    L. P. Kaelbling, M. L. Littman, and A. W. Moore, “Reinforcement learning: A survey,”Journal of Artificial Intelligence Research, vol. 4, pp. 237–285, 1996

    1996

  10. [10]

    Deep reinforcement learning: An overview,

    Y . Li, “Deep reinforcement learning: An overview,”arXiv preprint arXiv:1701.07274, 2017

    Pith/arXiv arXiv 2017

  11. [11]

    Reinforcement learning for combinatorial optimization: A survey,

    N. Mazyavkina, S. Sviridov, S. Ivanov, and E. Burnaev, “Reinforcement learning for combinatorial optimization: A survey,”Computers & Operations Research, vol. 134, p. 105400, 2021

    2021

  12. [12]

    Symbolic regression via neural-guided genetic programming population seeding,

    T. N. Mundhenk, M. Landajuela, R. Glatt, C. P. Santiago, D. M. Faissol, and B. K. Petersen, “Symbolic regression via neural-guided genetic programming population seeding,”arXiv preprint arXiv:2111.00053, 2021

    Pith/arXiv arXiv 2021

  13. [13]

    A surrogate-assisted controller for expensive evolutionary reinforcement learning,

    Y . Wang, T. Zhang, Y . Chang, X. Wang, B. Liang, and B. Yuan, “A surrogate-assisted controller for expensive evolutionary reinforcement learning,”Information Sciences, vol. 616, pp. 539–557, 2022

    2022

  14. [14]

    Controlling genetic algorithms with reinforcement learning,

    J. E. Pettinger and R. M. Everson, “Controlling genetic algorithms with reinforcement learning,” inProceedings of the Genetic and Evolutionary Computation Conference, 2002, pp. 692–692

    2002

  15. [15]

    Rein- forcement learning for online control of evolutionary algorithms,

    A. Eiben, M. Horvath, W. Kowalczyk, and M. C. Schut, “Rein- forcement learning for online control of evolutionary algorithms,” inProceedings of the International Workshop on Engineering Self- Organising Applications, 2006, pp. 151–160. 17

    2006

  16. [16]

    Algorithm selection using reinforcement learning,

    M. G. Lagoudakis and M. L. Littman, “Algorithm selection using reinforcement learning,” inProceedings of the International Conference on Machine Learning, 2000, pp. 511–518

    2000

  17. [17]

    Symbol: Generating flexible black-box optimizers through symbolic equation learning,

    J. Chen, Z. Ma, H. Guo, Y . Ma, J. Zhang, and y. Gong, “Symbol: Generating flexible black-box optimizers through symbolic equation learning,”arXiv preprint arXiv:2402.02355, 2024

    Pith/arXiv arXiv 2024

  18. [18]

    A survey of large language models,

    W. X. Zhao, K. Zhou, J. Li, T. Tang, X. Wang, Y . Hou, Y . Min, B. Zhang, J. Zhang, Z. Dong, Y . Du, C. Yang, Y . Chen, Z. Chen, J. Jiang, R. Ren, Y . Li, X. Tang, Z. Liu, P. Liu, J. Nie, and J. Wen, “A survey of large language models,”arXiv preprint arXiv:2303.18223, 2023

    Pith/arXiv arXiv 2023

  19. [19]

    A review of current trends, techniques, and challenges in large language models (llms),

    R. Patil and V . Gudivada, “A review of current trends, techniques, and challenges in large language models (llms),”Applied Sciences, vol. 14, no. 5, p. 2074, 2024

    2074

  20. [20]

    Chain-of-thought prompting elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans, M. Bosma, I. Brian, F. Xia, E. Chi, Q. V . Le, and D. Zhou, “Chain-of-thought prompting elicits reasoning in large language models,”Advances in Neural Information Processing Systems, vol. 35, pp. 24 824–24 837, 2022

    2022

  21. [21]

    Don’t hallucinate, abstain: Identifying llm knowledge gaps via multi- llm collaboration,

    S. Feng, W. Shi, Y . Wang, W. Ding, V . Balachandran, and Y . Tsvetkov, “Don’t hallucinate, abstain: Identifying llm knowledge gaps via multi- llm collaboration,”arXiv preprint arXiv:2402.00367, 2024

    Pith/arXiv arXiv 2024

  22. [22]

    Recent advances in natural language processing via large pre-trained language models: A survey,

    B. Min, H. Ross, E. Sulem, A. P. B. Veyseh, T. H. Nguyen, O. Sainz, E. Agirre, I. Heintz, and D. Roth, “Recent advances in natural language processing via large pre-trained language models: A survey,”ACM Computing Surveys, vol. 56, no. 2, pp. 1–40, 2023

    2023

  23. [23]

    Large language models in medicine,

    A. J. Thirunavukarasu, D. S. J. Ting, K. Elangovan, L. Gutierrez, T. F. Tan, and D. S. W. Ting, “Large language models in medicine,”Nature Medicine, vol. 29, no. 8, pp. 1930–1940, 2023

    1930

  24. [24]

    Large language models in finance: a survey,

    Y . Li, S. Wang, H. Ding, and H. Chen, “Large language models in finance: a survey,” inProceedings of the ACM International Conference on AI in Finance, 2023, pp. 374–382

    2023

  25. [25]

    Pangu-bot: efficient generative dia- logue pre-training from pre-trained language model,

    F. Mi, Y . Li, Y . Zeng, J. Zhou, Y . Wang, C. Xu, L. Shang, X. Jiang, S. Zhao, and Q. Liu, “Pangu-bot: efficient generative dia- logue pre-training from pre-trained language model,”arXiv preprint arXiv:2203.17090, 2022

    Pith/arXiv arXiv 2022

  26. [26]

    Bloomberggpt: A large language model for finance,

    S. Wu, O. Irsoy, S. Lu, V . Dabravolski, M. Dredze, S. Gehrmann, P. Kambadur, D. Rosenberg, and G. Mann, “Bloomberggpt: A large language model for finance,”arXiv preprint arXiv:2303.17564, 2023

    Pith/arXiv arXiv 2023

  27. [27]

    Orlm: A customizable framework in training large models for automated optimization modeling,

    C. Huang, Z. Tang, S. Hu, R. Jiang, X. Zheng, D. Ge, B. Wang, and Z. Wang, “Orlm: A customizable framework in training large models for automated optimization modeling,”Operations Research, 2025

    2025

  28. [28]

    Large language models as optimizers,

    C. Yang, X. Wang, Y . Lu, H. Liu, Q. V . Le, D. Zhou, and X. Chen, “Large language models as optimizers,” inProceedings of the Interna- tional Conference on Learning Representations, 2023, pp. 1–10

    2023

  29. [29]

    Exploring the true potential: Evaluating the black-box optimization capability of large language models,

    B. Huang, X. Wu, Y . Zhou, J. Wu, L. Feng, R. Cheng, and K. C. Tan, “Exploring the true potential: Evaluating the black-box optimization capability of large language models,”arXiv preprint arXiv:2404.06290, 2024

    Pith/arXiv arXiv 2024

  30. [30]

    Evolutionary computation in the era of large language model: survey and roadmap,

    X. Wu, S. Wu, J. Wu, L. Feng, and K. C. Tan, “Evolutionary computation in the era of large language model: survey and roadmap,” IEEE Transactions on Evolutionary Computation, vol. 29, no. 2, pp. 534–554, 2025

    2025

  31. [31]

    When large language model meets optimization,

    S. Huang, K. Yang, S. Qi, and R. Wang, “When large language model meets optimization,”Swarm and Evolutionary Computation, vol. 90, p. 101663, 2024

    2024

  32. [32]

    When large language models meet evolutionary algorithms,

    W. Chao, J. Zhao, L. Jiao, L. Li, F. Liu, and S. Yang, “When large language models meet evolutionary algorithms,”arXiv preprint arXiv:2401.10510, 2024

    Pith/arXiv arXiv 2024

  33. [33]

    Deep insights into automated optimization with large language models and evolutionary algorithms,

    H. Yu and J. Liu, “Deep insights into automated optimization with large language models and evolutionary algorithms,”arXiv preprint arXiv:2410.20848, 2024

    Pith/arXiv arXiv 2024

  34. [34]

    A systematic survey on large language models for algorithm design,

    F. Liu, Y . Yao, P. Guo, Z. Yang, Z. Zhao, X. Lin, X. Tong, M. Yuan, Z. Lu, Z. Wang, and Q. Zhang, “A systematic survey on large language models for algorithm design,”arXiv preprint arXiv:2410.14716, 2024

    arXiv 2024

  35. [35]

    Toward automated algorithm design: A survey and practical guide to meta-black-box- optimization,

    Z. Ma, H. Guo, Y . Gong, J. Zhang, and K. C. Tan, “Toward automated algorithm design: A survey and practical guide to meta-black-box- optimization,”IEEE Transactions on Evolutionary Computation, 2025

    2025

  36. [36]

    Evolu- tionary algorithms,

    T. Bartz-Beielstein, J. Branke, J. Mehnen, and O. Mersmann, “Evolu- tionary algorithms,”Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, vol. 4, no. 3, pp. 178–195, 2014

    2014

  37. [37]

    An overview of evolutionary algorithms in multiobjective optimization,

    C. M. Fonseca and P. J. Fleming, “An overview of evolutionary algorithms in multiobjective optimization,”Evolutionary Computation, vol. 3, no. 1, pp. 1–16, 1995

    1995

  38. [38]

    A competitive swarm optimizer for large scale optimization,

    R. Cheng and Y . Jin, “A competitive swarm optimizer for large scale optimization,”IEEE Transactions on Cybernetics, vol. 45, no. 2, pp. 191–204, 2014

    2014

  39. [39]

    Platemo: A matlab platform for evolutionary multi-objective optimization [educational forum],

    Y . Tian, R. Cheng, X. Zhang, and Y . Jin, “Platemo: A matlab platform for evolutionary multi-objective optimization [educational forum],” IEEE Computational Intelligence Magazine, vol. 12, no. 4, pp. 73– 87, 2017

    2017

  40. [40]

    Genetic algorithms,

    J. H. Holland, “Genetic algorithms,”Scientific American, vol. 267, no. 1, pp. 66–73, 1992

    1992

  41. [41]

    Genetic programming as a means for programming computers by natural selection,

    J. R. Koza, “Genetic programming as a means for programming computers by natural selection,”Statistics and Computing, vol. 4, no. 2, pp. 87–112, 1994

    1994

  42. [42]

    Differential evolution: A survey of the state-of-the-art,

    S. Das and P. N. Suganthan, “Differential evolution: A survey of the state-of-the-art,”IEEE Transactions on Evolutionary Computation, vol. 15, no. 1, pp. 4–31, 2010

    2010

  43. [43]

    Particle swarm optimization,

    J. Kennedy and R. Eberhart, “Particle swarm optimization,” inPro- ceedings of the International Conference on Neural Networks, vol. 4, 1995, pp. 1942–1948

    1995

  44. [44]

    Multiob- jective evolutionary algorithms in aeronautical and aerospace engineer- ing,

    A. Arias-Montano, C. A. C. Coello, and E. Mezura-Montes, “Multiob- jective evolutionary algorithms in aeronautical and aerospace engineer- ing,”IEEE Transactions on Evolutionary Computation, vol. 16, no. 5, pp. 662–694, 2012

    2012

  45. [45]

    Neural architecture search: A survey,

    T. Elsken, J. H. Metzen, and F. Hutter, “Neural architecture search: A survey,”Journal of Machine Learning Research, vol. 20, no. 55, pp. 1–21, 2019

    2019

  46. [46]

    Machine learning into metaheuristics: A survey and taxonomy,

    E.-G. Talbi, “Machine learning into metaheuristics: A survey and taxonomy,”ACM Computing Surveys, vol. 54, no. 6, pp. 1–32, 2021

    2021

  47. [47]

    Deep rein- forcement learning based adaptive operator selection for evolutionary multi-objective optimization,

    Y . Tian, X. Li, H. Ma, X. Zhang, K. C. Tan, and Y . Jin, “Deep rein- forcement learning based adaptive operator selection for evolutionary multi-objective optimization,”IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 7, no. 4, pp. 1051–1064, 2022

    2022

  48. [48]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in Neural Information Processing Systems, vol. 30, 2017

    2017

  49. [49]

    Compare encoder-decoder, encoder-only, and decoder-only architectures for text generation on low-resource datasets,

    P. Cai, Y . Fan, and F. Leu, “Compare encoder-decoder, encoder-only, and decoder-only architectures for text generation on low-resource datasets,” inProceedings of the International Conference on Broadband and Wireless Computing, Communication and Applications, 2021, pp. 216–225

    2021

  50. [50]

    Bart: Denoising sequence-to- sequence pre-training for natural language generation, translation, and comprehension,

    M. Lewis, Y . Liu, N. Goyal, M. Ghazvininejad, A. Mohamed, O. Levy, V . Stoyanov, and L. Zettlemoyer, “Bart: Denoising sequence-to- sequence pre-training for natural language generation, translation, and comprehension,”arXiv preprint arXiv:1910.13461, 2019

    Pith/arXiv arXiv 1910

  51. [51]

    Exploring the limits of transfer learning with a unified text-to-text transformer,

    C. Raffel, N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y . Zhou, W. Li, and P. J. Liu, “Exploring the limits of transfer learning with a unified text-to-text transformer,”Journal of Machine Learning Research, vol. 21, no. 1, pp. 5485–5551, 2020

    2020

  52. [52]

    Bert: pre-training of deep bidirectional transformers for language understanding,

    J. Devlin, M. Chang, K. Lee, and K. Toutanova, “Bert: pre-training of deep bidirectional transformers for language understanding,” in Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Tech- nologies, 2019, pp. 4171–4186

    2019

  53. [53]

    Gpt-3: its nature, scope, limits, and consequences,

    L. Floridi and M. Chiriatti, “Gpt-3: its nature, scope, limits, and consequences,”Minds and Machines, vol. 30, no. 4, pp. 681–694, 2020

    2020

  54. [54]

    Deepseek-coder: when the large language model meets programming–the rise of code intelligence,

    D. Guo, Q. Zhu, D. Yang, Z. Xie, K. Dong, W. Zhang, G. Chen, X. Bi, Y . Wu, Y . K. Li, F. Luo, Y . Xiong, and W. Liang, “Deepseek-coder: when the large language model meets programming–the rise of code intelligence,”arXiv preprint arXiv:2401.14196, 2024

    Pith/arXiv arXiv 2024

  55. [55]

    A prompt pattern catalog to enhance prompt engineering with chatgpt,

    J. White, Q. Fu, S. Hays, M. Sandborn, C. Olea, H. Gilbert, A. El- nashar, J. Spencer-Smith, and D. C. Schmidt, “A prompt pattern catalog to enhance prompt engineering with chatgpt,”arXiv preprint arXiv:2302.11382, 2023

    Pith/arXiv arXiv 2023

  56. [56]

    Parameter-efficient fine-tuning for large models: A comprehensive survey,

    Z. Han, C. Gao, J. Liu, J. Zhang, and S. Q. Zhang, “Parameter-efficient fine-tuning for large models: A comprehensive survey,”arXiv preprint arXiv:2403.14608, 2024

    Pith/arXiv arXiv 2024

  57. [57]

    Large lan- guage models are zero-shot reasoners,

    T. Kojima, S. S. Gu, M. Reid, Y . Matsuo, and Y . Iwasawa, “Large lan- guage models are zero-shot reasoners,”Advances in Neural Information Processing Systems, vol. 35, pp. 22 199–22 213, 2022

    2022

  58. [58]

    Tree of thoughts: Deliberate problem solving with large language models,

    S. Yao, D. Yu, J. Zhao, I. Shafran, T. Griffiths, Y . Cao, and K. Narasimhan, “Tree of thoughts: Deliberate problem solving with large language models,”Advances in Neural Information Processing Systems, vol. 36, pp. 11 809–11 822, 2023

    2023

  59. [59]

    Graph of thoughts: solving elaborate problems with large language models,

    M. Besta, N. Blach, A. Kubicek, R. Gerstenberger, M. Podstawski, L. Gianinazzi, J. Gajda, T. Lehmann, H. Niewiadomski, P. Nyczyk, and T. Hoefler, “Graph of thoughts: solving elaborate problems with large language models,” inProceedings of the AAAI Conference on Artificial Intelligence, 2024, pp. 17 682–17 690

    2024

  60. [60]

    Chain of code: Reasoning 18 with a language model-augmented code emulator,

    C. Li, J. Liang, A. Zeng, X. Chen, K. Hausman, D. Sadigh, S. Levine, L. Fei-Fei, F. Xia, and B. Ichter, “Chain of code: Reasoning 18 with a language model-augmented code emulator,”arXiv preprint arXiv:2312.04474, 2023

    Pith/arXiv arXiv 2023

  61. [61]

    Chain-of-knowledge: Grounding large language models via dynamic knowledge adapting over heterogeneous sources,

    X. Li, R. Zhao, Y . K. Chia, B. Ding, S. Joty, S. Poria, and L. Bing, “Chain-of-knowledge: Grounding large language models via dynamic knowledge adapting over heterogeneous sources,”arXiv preprint arXiv:2305.13269, 2023

    Pith/arXiv arXiv 2023

  62. [62]

    Visual instruction tuning,

    H. Liu, C. Li, Q. Wu, and Y . J. Lee, “Visual instruction tuning,” Advances in Neural Information Processing Systems, vol. 36, pp. 34 892–34 916, 2023

    2023

  63. [63]

    Scaling instruction- finetuned language models,

    H. W. Chung, L. Hou, S. Longpre, B. Zoph, Y . Tay, W. Fedus, Y . Li, X. Wang, M. Dehghani, S. Brahma, A. Webson, S. S. Gu, Z. Dai, M. Suzgun, X. Chen, A. Chowdhery, A. Castro-Ros, M. Pellat, K. Robinson, D. Valter, S. Narang, G. Mishra, A. Yu, V . Zhao, Y . Huang, A. Dai, H. Yu, S. Petrov, E. H. Chi, J. Dean, J. Devlin, A. Roberts, D. Zhou, Q. V . Le, and...

    2024

  64. [64]

    Training language models to follow instructions with human feedback,

    L. Ouyang, J. Wu, X. Jiang, D. Almeida, C. Wainwright, P. Mishkin, C. Zhang, S. Agarwal, K. Slama, A. Ray, J. Schulman, J. Hilton, F. Kelton, L. Miller, M. Simens, A. Askell, P. Welinder, P. F. Christiano, J. Leike, and R. Lowe, “Training language models to follow instructions with human feedback,”Advances in Neural Information Processing Systems, vol. 35...

    2022

  65. [65]

    Direct preference optimization: Your language model is secretly a reward model,

    R. Rafailov, A. Sharma, E. Mitchell, C. D. Manning, S. Ermon, and C. Finn, “Direct preference optimization: Your language model is secretly a reward model,”Advances in Neural Information Processing Systems, vol. 36, pp. 53 728–53 741, 2023

    2023

  66. [66]

    Ner4opt: Named entity recognition for optimization modelling from natural language,

    P. P. Dakle, S. Kadıo ˘glu, K. Uppuluri, R. Politi, P. Raghavan, S. Ralla- bandi, and R. Srinivasamurthy, “Ner4opt: Named entity recognition for optimization modelling from natural language,” inProceedings of the International Conference on Integration of Constraint Programming, Artificial Intelligence, and Operations Research, 2023, pp. 299–319

    2023

  67. [67]

    Towards an automatic optimisation model genera- tor assisted with generative pre-trained transformer,

    B. Almonacid, “Towards an automatic optimisation model genera- tor assisted with generative pre-trained transformer,”arXiv preprint arXiv:2305.05811, 2023

    Pith/arXiv arXiv 2023

  68. [68]

    Holy grail 2.0: From natural language to constraint models,

    D. Tsouros, H. Verhaeghe, S. Kadıo ˘glu, and T. Guns, “Holy grail 2.0: From natural language to constraint models,”arXiv preprint arXiv:2308.01589, 2023

    Pith/arXiv arXiv 2023

  69. [69]

    Synthesizing mixed-integer lin- ear programming models from natural language descriptions,

    Q. Li, L. Zhang, and V . Mak-Hau, “Synthesizing mixed-integer lin- ear programming models from natural language descriptions,”arXiv preprint arXiv:2311.15271, 2023

    Pith/arXiv arXiv 2023

  70. [70]

    Chain-of-experts: When llms meet complex operations research problems,

    Z. Xiao, D. Zhang, Y . Wu, L. Xu, Y . J. Wang, X. Han, X. Fu, T. Zhong, J. Zeng, and M. Song, “Chain-of-experts: When llms meet complex operations research problems,” inProceedings of the International Conference on Learning Representations, 2023

    2023

  71. [71]

    Optimus: Scalable opti- mization modeling with (mi) lp solvers and large language models,

    A. AhmadiTeshnizi, W. Gao, and M. Udell, “Optimus: Scalable opti- mization modeling with (mi) lp solvers and large language models,” arXiv preprint arXiv:2402.10172, 2024

    Pith/arXiv arXiv 2024

  72. [72]

    Optimization modeling and verification from problem specifications using a multi-agent multi-stage llm framework,

    M. Mostajabdaveh, T. T. Yu, R. Ramamonjison, G. Carenini, Z. Zhou, and Y . Zhang, “Optimization modeling and verification from problem specifications using a multi-agent multi-stage llm framework,”Infor: Information Systems and Operational Research, vol. 62, no. 4, pp. 599–617, 2024

    2024

  73. [73]

    Democratizing energy man- agement with llm-assisted optimization autoformalism,

    M. Jin, B. Sel, F. Hardeep, and W. Yin, “Democratizing energy man- agement with llm-assisted optimization autoformalism,” inProceedings of the International Conference on Communications, Control, and Computing Technologies for Smart Grids, 2024, pp. 258–263

    2024

  74. [74]

    Cafa: Coding as auto- formulation can boost large language models in solving linear pro- gramming problem,

    H. Deng, B. Zheng, Y . Jiang, and T. H. Tran, “Cafa: Coding as auto- formulation can boost large language models in solving linear pro- gramming problem,” inProceedings of the Workshop on Mathematical Reasoning and Artificial Intelligence at the Conference on Neural Information Processing Systems, 2024

    2024

  75. [75]

    Llms for math- ematical modeling: Towards bridging the gap between natural and mathematical languages,

    X. Huang, Q. Shen, Y . Hu, A. Gao, and B. Wang, “Llms for math- ematical modeling: Towards bridging the gap between natural and mathematical languages,”arXiv preprint arXiv:2405.13144, 2024

    Pith/arXiv arXiv 2024

  76. [76]

    Abstract operations research modeling using natural language inputs,

    J. Li, R. Wickman, S. Bhatnagar, R. K. Maity, and A. Mukherjee, “Abstract operations research modeling using natural language inputs,” Information, vol. 16, no. 2, p. 128, 2025

    2025

  77. [77]

    Trip-pal: Travel planning with guarantees by combin- ing large language models and automated planners,

    T. de la Rosa, S. Gopalakrishnan, A. Pozanco, Z. Zeng, and D. Borrajo, “Trip-pal: Travel planning with guarantees by combin- ing large language models and automated planners,”arXiv preprint arXiv:2406.10196, 2024

    Pith/arXiv arXiv 2024

  78. [78]

    Solving general natural-language-description optimization problems with large language models,

    J. Zhang, W. Wang, S. Guo, L. Wang, F. Lin, C. Yang, and W. Yin, “Solving general natural-language-description optimization problems with large language models,”arXiv preprint arXiv:2407.07924, 2024

    Pith/arXiv arXiv 2024

  79. [79]

    Optimus-0.3: Using large language models to model and solve optimization problems at scale,

    A. AhmadiTeshnizi, W. Gao, H. Brunborg, S. Talaei, C. Law- less, and M. Udell, “Optimus-0.3: Using large language models to model and solve optimization problems at scale,”arXiv preprint arXiv:2407.19633, 2024

    Pith/arXiv arXiv 2024

  80. [80]

    I want it that way: Enabling interactive decision support using large language models and constraint programming,

    C. Lawless, J. Schoeffer, L. Le, K. Rowan, S. Sen, C. St. Hill, J. Suh, and B. Sarrafzadeh, “I want it that way: Enabling interactive decision support using large language models and constraint programming,” ACM Transactions on Interactive Intelligent Systems, vol. 14, no. 3, pp. 1–33, 2024

    2024

Showing first 80 references.