pith. sign in

arxiv: 2605.23273 · v1 · pith:L3D4CITUnew · submitted 2026-05-22 · 💻 cs.MA

Self-Refining Topology Optimization via an LLM-Based Multi-Agent Framework

Hyunjee Park , Hayoung Chung This is my paper

Pith reviewed 2026-05-25 03:04 UTC · model grok-4.3

classification 💻 cs.MA
keywords topology optimizationmulti-agent systemslarge language modelsself-refinementdesign automationLLM collaborationnumerical optimization
0
0 comments X

The pith

A system of six LLM agents automates topology optimization by iteratively refining setups, code, and designs through collaboration.

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

The paper presents TopOptAgents as a multi-agent framework that lets large language models handle the entire topology optimization workflow, including the expert choices about parameters and feasibility checks that normally require human input. Agents cycle through formulation, validation, code writing, execution, and assessment, correcting errors and building better results over iterations. This matters because many topology problems involve decisions that break full automation, and the approach targets cases where models have seen little prior training data. If the claim holds, it widens the set of design tasks that can run end-to-end without manual intervention.

Core claim

TopOptAgents consists of six LLM-based agents collaborating through iterative self-refinement cycles spanning problem formulation, validation, code generation and execution, and quality assessment of the optimized structure. This process enables error correction and progressive improvement of both the optimization setup and resulting design. The framework is demonstrated on optimization problems selected to cover a range of settings that differ in their literature coverage and numerical characteristics. The benefits of iterative self-refinement are found to be particularly pronounced for problem classes where the pretrained language model has limited prior exposure, such as formulations with

What carries the argument

TopOptAgents, the six-agent LLM system that runs repeated self-refinement cycles across formulation, validation, coding, execution, and design evaluation.

If this is right

  • Expert decisions about parameters and feasibility become part of the automated loop instead of external interruptions.
  • Self-refinement yields the largest gains on problems whose formulations and code examples are rare in training data.
  • The range of topology optimization tasks that LLM automation can complete reliably expands beyond what single models achieve.
  • Converged designs emerge from the process even when initial agent outputs contain inconsistencies or invalid code.

Where Pith is reading between the lines

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

  • The same agent-division and refinement pattern could transfer to other iterative engineering tasks such as structural sizing or fluid-device layout.
  • If the collaboration pattern holds across models, it offers a route to reduce dependence on hand-tuned commercial solvers for standard problems.
  • Extending the agents to include explicit manufacturing or cost checks would test whether the loop can absorb constraints left out of the base optimization.
  • Measuring whether refinement steps actually decrease error metrics over cycles would give a direct check on whether the process improves rather than stabilizes.

Load-bearing premise

That the back-and-forth among the six agents produces genuine error correction and steady improvement rather than simply repeating or compounding the base model's mistakes.

What would settle it

A side-by-side test on the same set of low-literature topology problems measuring the fraction of runs that reach converged, physically feasible designs for the multi-agent system versus a single LLM.

Figures

Figures reproduced from arXiv: 2605.23273 by Hayoung Chung, Hyunjee Park.

Figure 1
Figure 1. Figure 1: Overall workflow of the decision-making and computational processes for topology optimization. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Effect of the filter radius rmin and the projection threshold η on the optimized structure after Song et al. [12]. 2.2. Self-refining multi-agent system In this work, we present TopOptAgents, a self-refining LLM-based multi-agent framework specially tailored to solve topology optimization problems by making a sequence of decisions required as described in Sec. 2.1. A schematic configuration of TopOptAgent … view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the TopOptAgents workflow and the outputs of each agent. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Here, the base LLM models are selected differently across agents based on the trade-off between [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: Example of the system prompt structure for an LLM agent represented by Scientist agent. The prompt includes [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Profiles of the agents in TopOptAgents. 11 [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Benchmark problem settings and initial user queries used for evaluating TopOptAgents. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Summary of expected self-refinement and retry counts for the benchmark problems. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Success rates over 10 trials for three benchmark topology optimization problems, comparing TopOptAgents and [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Overall execution flow of TopOptAgents for the cantilever beam compliance minimization problem. [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Example of Validator agent directly correcting an incorrectly specified external load location in Scientist’s initial [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Example of Validator agent requesting reformulation when the specified external load location overlaps with a [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: (a) Example of code-level error correction by Reviewer agent within the code generation team. (b) Example of [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: (a) Example of Critic agent identifying an incorrect choice on the constraint function. (b) Example of Critic agent [PITH_FULL_IMAGE:figures/full_fig_p019_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Example of Critic agent refining the filter radius [PITH_FULL_IMAGE:figures/full_fig_p020_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Example of Critic agent refining the beta continuation scheme for a more discrete design. [PITH_FULL_IMAGE:figures/full_fig_p021_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Example of the final report generated by TopOptAgents, including the problem formulation, generated Python [PITH_FULL_IMAGE:figures/full_fig_p023_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Example of human feedback reflected in a subsequent execution through adding a hole in the problem domain. [PITH_FULL_IMAGE:figures/full_fig_p024_17.png] view at source ↗
read the original abstract

Topology optimization is a widely used design method that produces optimized material distributions for prescribed objectives and constraints through well-established numerical algorithms. Throughout the workflow, engineers make a series of decisions ranging from setting and adjusting numerical parameters to assessing whether the converged design meets considerations beyond those explicitly included in the optimization problem, such as physical feasibility. These decisions, which draw on domain expertise, interfere with the autonomous design process. To address this difficulty, this study presents TopOptAgents, a multi-agent system for automating not only the design process but also decision-making during the key stages of the topology optimization process. TopOptAgents consists of six LLM-based agents collaborating through iterative self-refinement cycles spanning problem formulation, validation, code generation and execution, and quality assessment of the optimized structure. This process enables error correction and progressive improvement of both the optimization setup and resulting design. The framework is demonstrated on optimization problems selected to cover a range of settings that differ in their literature coverage and numerical characteristics The benefits of iterative self-refinement are found to be particularly pronounced for problem classes where the pretrained language model has limited prior exposure, such as formulations whose literature and open-source implementations are comparatively sparse. In such cases, the proposed framework reliably produces converged designs where a single state-of-the-art LLM struggles, suggesting that self-refinement broadens the range of topology optimization problems that LLM-based automation can reliably address.

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 / 0 minor

Summary. The manuscript introduces TopOptAgents, a multi-agent framework of six LLM-based agents that collaborate via iterative self-refinement cycles to automate the full topology optimization workflow, encompassing problem formulation, validation, code generation/execution, and post-optimization quality assessment. It claims that this process enables error correction and progressive improvement, reliably yielding converged designs on selected problems (chosen for varying literature coverage and numerical characteristics) where a single state-of-the-art LLM fails, with benefits most pronounced on sparse-literature formulations.

Significance. If substantiated, the work would demonstrate a practical route for extending LLM reliability in engineering design automation beyond base-model limits on low-exposure problem classes, potentially reducing reliance on human expertise for parameter tuning and feasibility checks in topology optimization.

major comments (2)
  1. [Demonstration and results (abstract and implied experimental section)] The central claim that iterative collaboration among the six agents produces genuine error correction (rather than variance from repeated sampling) is load-bearing yet unsupported by any ablation that disables the refinement cycle while holding total token budget or LLM-call count fixed; the abstract's demonstration on sparse problems therefore cannot isolate the claimed mechanism.
  2. [Demonstration and results (abstract and implied experimental section)] No per-iteration error-type logs, success-rate statistics over repeated trials, or baseline comparisons with equivalent compute (e.g., single-LLM retries) are described, leaving open the possibility that observed convergence on low-literature-coverage cases arises from base-model stochasticity rather than cross-agent correction of invalid constraints or mesh errors.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive critique of the experimental support for the self-refinement mechanism. We agree that the current manuscript does not provide the controlled comparisons needed to isolate iterative cross-agent correction from base-model stochasticity, and we will revise the paper to include the requested analyses.

read point-by-point responses

  1. Referee: The central claim that iterative collaboration among the six agents produces genuine error correction (rather than variance from repeated sampling) is load-bearing yet unsupported by any ablation that disables the refinement cycle while holding total token budget or LLM-call count fixed; the abstract's demonstration on sparse problems therefore cannot isolate the claimed mechanism.

    Authors: We agree that an ablation holding total token budget and LLM-call count fixed is required to separate the contribution of the multi-agent refinement loop from repeated independent sampling. The manuscript currently compares the framework against single-LLM attempts but does not enforce compute parity in this way. In revision we will add this ablation on the sparse-literature problems, reporting success rates for the full TopOptAgents loop versus an equivalent-budget regime of independent single-LLM generations (with the same total calls and tokens). revision: yes

  2. Referee: No per-iteration error-type logs, success-rate statistics over repeated trials, or baseline comparisons with equivalent compute (e.g., single-LLM retries) are described, leaving open the possibility that observed convergence on low-literature-coverage cases arises from base-model stochasticity rather than cross-agent correction of invalid constraints or mesh errors.

    Authors: We acknowledge that the manuscript lacks per-iteration error logs, multi-trial success statistics, and matched-compute single-LLM baselines. These omissions leave the mechanism under-supported. In the revised version we will include (i) logs classifying error types corrected at each iteration, (ii) success-rate tables over at least five independent trials per problem, and (iii) direct comparisons against single-LLM retry baselines that consume the same total token or call budget. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical system demonstration with no derivations or self-referential loops

full rationale

The paper describes a multi-agent LLM framework (TopOptAgents) for topology optimization and reports empirical results on selected problems. No equations, parameters, or mathematical derivations are present in the provided text. Claims rest on observed performance differences between single-LLM and multi-agent setups rather than any prediction or result that reduces to its own inputs by construction. Self-citations, if present, are not load-bearing for any derivation chain. This is a standard non-circular empirical presentation.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only; no free parameters, axioms, or invented entities can be extracted beyond the high-level description of LLM agents and iterative cycles.

pith-pipeline@v0.9.0 · 5772 in / 994 out tokens · 18373 ms · 2026-05-25T03:04:15.137465+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 3 internal anchors

  1. [1]

    J. D. Deaton, R. V. Grandhi, A survey of structural and multidisciplinary continuum topology optimization: post 2000, Structural and multidisciplinary optimization 49 (1) (2014) 1–38

    work page 2000

  2. [2]

    Sigmund, K

    O. Sigmund, K. Maute, Topology optimization approaches: A comparative review, Structural and multidisciplinary optimization 48 (6) (2013) 1031–1055

    work page 2013

  3. [3]

    Zhu, W.-H

    J.-H. Zhu, W.-H. Zhang, L. Xia, Topology optimization in aircraft and aerospace structures design, Archives of computational methods in engineering 23 (4) (2016) 595–622. 25

    work page 2016

  4. [4]

    R. Yang, A. Chahande, Automotive applications of topology optimization, Structural optimization 9 (3) (1995) 245–249

    work page 1995

  5. [5]

    N. M. Patel, B.-S. Kang, J. E. Renaud, A. Tovar, Crashworthiness design using topology optimiza- tion (2009)

    work page 2009

  6. [6]

    Lundgaard, J

    C. Lundgaard, J. Alexandersen, M. Zhou, C. S. Andreasen, O. Sigmund, Revisiting density-based topology optimization for fluid-structure-interaction problems, Structural and Multidisciplinary Op- timization 58 (3) (2018) 969–995

    work page 2018

  7. [7]

    Pietropaoli, F

    M. Pietropaoli, F. Montomoli, A. Gaymann, Three-dimensional fluid topology optimization for heat transfer, Structural and Multidisciplinary Optimization 59 (3) (2019) 801–812

    work page 2019

  8. [8]

    M. E. Lynch, S. Sarkar, K. Maute, Machine learning to aid tuning of numerical parameters in topology optimization, Journal of Mechanical Design 141 (11) (2019) 114502

    work page 2019

  9. [9]

    Zhang, J

    S. Zhang, J. A. Norato, Finding better local optima in topology optimization via tunneling, in: Inter- national Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Vol. 51760, American Society of Mechanical Engineers, 2018, p. V02BT03A014

    work page 2018

  10. [10]

    D. Ha, J. Carstensen, Automatic hyperparameter tuning of topology optimization algorithms using surrogate optimization: D. ha, j. carstensen, Structural and Multidisciplinary Optimization 67 (9) (2024) 157

    work page 2024

  11. [11]

    M. Zhou, Y. Shyy, H. Thomas, Checkerboard and minimum member size control in topology opti- mization, Structural and Multidisciplinary Optimization 21 (2) (2001) 152–158

    work page 2001

  12. [12]

    W. S. Song, H. Park, J. Park, S. Min, Adaptive beta update scheme in heaviside projection method of topology optimization, Computer Methods in Applied Mechanics and Engineering 453 (2026) 118805

    work page 2026

  13. [13]

    Dunning, F

    P. Dunning, F. Wein, Automatic projection parameter increase for three-field density-based topology optimization, Structural and multidisciplinary optimization 68 (2) (2025) 33

    work page 2025

  14. [14]

    W. X. Zhao, K. Zhou, J. Li, T. Tang, X. Wang, Y. Hou, Y. Min, B. Zhang, J. Zhang, Z. Dong, et al., A survey of large language models, arXiv preprint arXiv:2303.18223 1 (2) (2023) 1–124

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  15. [15]

    J. Gu, X. Jiang, Z. Shi, H. Tan, X. Zhai, C. Xu, W. Li, Y. Shen, S. Ma, H. Liu, et al., A survey on llm-as-a-judge, The Innovation (2024)

    work page 2024

  16. [16]

    A. Zhao, D. Huang, Q. Xu, M. Lin, Y.-J. Liu, G. Huang, Expel: Llm agents are experiential learners, in: Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, 2024, pp. 19632–19642

    work page 2024

  17. [17]

    L. Wang, C. Ma, X. Feng, Z. Zhang, H. Yang, J. Zhang, Z. Chen, J. Tang, X. Chen, Y. Lin, et al., A survey on large language model based autonomous agents, Frontiers of Computer Science 18 (6) (2024) 186345

    work page 2024

  18. [18]

    M. J. Buehler, Melm, a generative pretrained language modeling framework that solves forward and inverse mechanics problems, Journal of the Mechanics and Physics of Solids 181 (2023) 105454

    work page 2023

  19. [19]

    M. J. Buehler, Mechgpt, a language-based strategy for mechanics and materials modeling that connects knowledge across scales, disciplines, and modalities, Applied Mechanics Reviews 76 (2) (2024) 021001

    work page 2024

  20. [20]

    Pandey, R

    S. Pandey, R. Xu, W. Wang, X. Chu, Openfoamgpt: A retrieval-augmented large language model (llm) agent for openfoam-based computational fluid dynamics, Physics of Fluids 37 (3) (2025)

    work page 2025

  21. [21]

    T. Guo, X. Chen, Y. Wang, R. Chang, S. Pei, N. V. Chawla, O. Wiest, X. Zhang, Large language model based multi-agents: A survey of progress and challenges, arXiv preprint arXiv:2402.01680 (2024)

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  22. [22]

    J. Chen, Y. Lu, X. Wang, H. Zeng, J. Huang, J. Gesi, Y. Xu, B. Yao, D. Wang, Multi-agent-as- judge: Aligning llm-agent-based automated evaluation with multi-dimensional human evaluation, arXiv preprint arXiv:2507.21028 (2025). 26

    work page arXiv 2025

  23. [23]

    J. Li, Q. Zhang, Y. Yu, Q. Fu, D. Ye, More agents is all you need, arXiv preprint arXiv:2402.05120 (2024)

    work page arXiv 2024

  24. [24]

    B. Ni, M. J. Buehler, Mechagents: Large language model multi-agent collaborations can solve mechanics problems, generate new data, and integrate knowledge, Extreme Mechanics Letters 67 (2024) 102131

    work page 2024

  25. [25]

    D. Park, H. Moon, S. Ryu, A self-correcting multi-agent llm framework for language-based physics simulation and explanation, npj Artificial Intelligence 2 (1) (2026) 10

    work page 2026

  26. [26]

    Deotale, A

    R. Deotale, A. Srinivasan, M. Golestanian, Y. Tian, T. Zhang, P. Vlachos, H. Gomez, All-fem: Agentic large language models fine-tuned for finite element methods, Computer Methods in Applied Mechanics and Engineering 457 (2026) 118985

    work page 2026

  27. [27]

    P. Chen, Y. Cai, Z. Zhou, J. Yao, J. Li, W. You, L. Sun, An llm-based multi-agent system to assist early-stage product design and evaluation, Journal of Engineering Design (2026) 1–36

    work page 2026

  28. [28]

    Zhang, X

    S. Zhang, X. Li, C. Yuan, W. Feng, Q. Jiang, Desagent: A multi-agent mechanical design method based on collaborative large and small models, Journal of Mechanical Design 148 (5) (2026) 051706

    work page 2026

  29. [29]

    L. Chen, H. Zuo, Z. Cai, Y. Yin, Y. Zhang, L. Sun, P. Childs, B. Wang, Toward controllable gener- ative design: A conceptual design generation approach leveraging the function–behavior–structure ontology and large language models, Journal of Mechanical Design 146 (12) (2024) 121401

    work page 2024

  30. [30]

    Jadhav, A

    Y. Jadhav, A. Barati Farimani, Large language model agent as a mechanical designer, Journal of Engineering Design (2026) 1–37

    work page 2026

  31. [31]

    F. Liu, X. Zeng, H. Liu, Towards multimodal data-driven scientific discovery powered by llm agents, in: The Fourteenth International Conference on Learning Representations, 2026

    work page 2026

  32. [32]

    AgentCoder: Multi-Agent-based Code Generation with Iterative Testing and Optimisation

    D. Huang, J. M. Zhang, M. Luck, Q. Bu, Y. Qing, H. Cui, Agentcoder: Multi-agent-based code generation with iterative testing and optimisation, arXiv preprint arXiv:2312.13010 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  33. [33]

    Sigmund, Topology optimization: a tool for the tailoring of structures and materials, Philosophi- cal Transactions of the Royal Society of London

    O. Sigmund, Topology optimization: a tool for the tailoring of structures and materials, Philosophi- cal Transactions of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences 358 (1765) (2000) 211–227

    work page 2000

  34. [34]

    Sigmund, A 99 line topology optimization code written in matlab, Structural and multidisci- plinary optimization 21 (2) (2001) 120–127

    O. Sigmund, A 99 line topology optimization code written in matlab, Structural and multidisci- plinary optimization 21 (2) (2001) 120–127

    work page 2001

  35. [35]

    Rojas-Labanda, O

    S. Rojas-Labanda, O. Sigmund, M. Stolpe, A short numerical study on the optimization methods influence on topology optimization, Structural and Multidisciplinary Optimization 56 (6) (2017) 1603–1612

    work page 2017

  36. [36]

    K. Liu, A. Tovar, An efficient 3d topology optimization code written in matlab, Structural and multidisciplinary optimization 50 (6) (2014) 1175–1196

    work page 2014

  37. [37]

    Rojas-Labanda, M

    S. Rojas-Labanda, M. Stolpe, Benchmarking optimization solvers for structural topology optimiza- tion, Structural and Multidisciplinary Optimization 52 (3) (2015) 527–547

    work page 2015

  38. [38]

    Sigmund, J

    O. Sigmund, J. Petersson, Numerical instabilities in topology optimization: a survey on procedures dealing with checkerboards, mesh-dependencies and local minima, Structural optimization 16 (1) (1998) 68–75

    work page 1998

  39. [39]

    B. S. Lazarov, O. Sigmund, Filters in topology optimization based on helmholtz-type differential equations, International journal for numerical methods in engineering 86 (6) (2011) 765–781

    work page 2011

  40. [40]

    F. Wang, B. S. Lazarov, O. Sigmund, On projection methods, convergence and robust formulations in topology optimization, Structural and multidisciplinary optimization 43 (6) (2011) 767–784

    work page 2011

  41. [41]

    J. K. Guest, A. Asadpoure, S.-H. Ha, Eliminating beta-continuation from heaviside projection and density filter algorithms, Structural and Multidisciplinary Optimization 44 (4) (2011) 443–453

    work page 2011

  42. [42]

    O’Reilly Media, Inc

    M. Oshin, N. Campos, Learning LangChain, " O’Reilly Media, Inc.", 2025. 27

    work page 2025

  43. [43]

    Andreassen, A

    E. Andreassen, A. Clausen, M. Schevenels, B. S. Lazarov, O. Sigmund, Efficient topology optimiza- tion in matlab using 88 lines of code, Structural and Multidisciplinary Optimization 43 (1) (2011) 1–16

    work page 2011

  44. [44]

    Y. Wang, X. Li, K. Long, P. Wei, Open-source codes of topology optimization: A summary for beginners to start their research, Computer Modeling in Engineering & Sciences 137 (1) (2023) 1–34

    work page 2023

  45. [45]

    Holmberg, B

    E. Holmberg, B. Torstenfelt, A. Klarbring, Stress constrained topology optimization, Structural and Multidisciplinary Optimization 48 (1) (2013) 33–47

    work page 2013

  46. [46]

    D. M. De Leon, J. Alexandersen, J. S. O. Fonseca, O. Sigmund, Stress-constrained topology op- timization for compliant mechanism design, Structural and Multidisciplinary Optimization 52 (5) (2015) 929–943

    work page 2015

  47. [47]

    C. Le, J. Norato, T. Bruns, C. Ha, D. Tortorelli, Stress-based topology optimization for continua, Structural and Multidisciplinary Optimization 41 (4) (2010) 605–620

    work page 2010

  48. [48]

    S. Park, B. Goh, H. Chung, Topology optimization with material point method: investigation into the design sensitivity and the effect of shape functions, Engineering with Computers 41 (5) (2025) 3099–3116

    work page 2025

  49. [49]

    R. Pan, H. Zhang, C. Liu, Codecor: An llm-based self-reflective multi-agent framework for code generation, arXiv preprint arXiv:2501.07811 (2025)

    work page arXiv 2025

  50. [50]

    Alnæs, J

    M. Alnæs, J. Blechta, J. Hake, A. Johansson, B. Kehlet, A. Logg, C. Richardson, J. Ring, M. E. Rognes, G. N. Wells, The fenics project version 1.5, Archive of numerical software 3 (100) (2015)

    work page 2015

  51. [51]

    E. Wu, G. Kenway, C. A. Mader, J. Jasa, J. R. ra Martins, pyoptsparse: A python framework for large-scale constrained nonlinear optimization of sparse systems, Journal of Open Source Software 5 (54) (2020) 2564

    work page 2020

  52. [52]

    C. Li, J. Flanigan, Task contamination: Language models may not be few-shot anymore, in: Pro- ceedings of the AAAI Conference on Artificial Intelligence, Vol. 38, 2024, pp. 18471–18480

    work page 2024

  53. [53]

    Riddell, A

    M. Riddell, A. Ni, A. Cohan, Quantifying contamination in evaluating code generation capabilities of language models, in: Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2024, pp. 14116–14137

    work page 2024

  54. [54]

    Sigmund, On benchmarking and good scientific practise in topology optimization, Structural and Multidisciplinary Optimization 65 (11) (2022) 315

    O. Sigmund, On benchmarking and good scientific practise in topology optimization, Structural and Multidisciplinary Optimization 65 (11) (2022) 315

    work page 2022

  55. [55]

    Duysinx, M

    P. Duysinx, M. P. Bendsøe, Topology optimization of continuum structures with local stress con- straints, International journal for numerical methods in engineering 43 (8) (1998) 1453–1478

    work page 1998

  56. [56]

    P. E. Gill, W. Murray, M. A. Saunders, Snopt: An sqp algorithm for large-scale constrained opti- mization, SIAM review 47 (1) (2005) 99–131

    work page 2005

  57. [57]

    Prabhakar, M

    V. Prabhakar, M. A. Islam, A. Atanas, Y.-T. Wang, J. Han, A. Jhunjhunwala, R. Apte, R. Clark, K. Xu, Z. Wang, et al., Omniscience: A domain-specialized llm for scientific reasoning and discovery, arXiv preprint arXiv:2503.17604 (2025)

    work page arXiv 2025

  58. [58]

    I. A. Stewart, T. P. Hage, Y.-C. Hsu, M. J. Buehler, Graphagents: Knowledge graph-guided agentic ai for cross-domain materials design, arXiv preprint arXiv:2602.07491 (2026)

    work page arXiv 2026

  59. [59]

    H. Luo, G. Chen, Y. Zheng, X. Wu, Y. Guo, Q. Lin, Y. Feng, Z. Kuang, M. Song, Y. Zhu, et al., Hy- pergraphrag: Retrieval-augmented generation via hypergraph-structured knowledge representation, arXiv preprint arXiv:2503.21322 (2025). 28

    work page arXiv 2025