pith. machine review for the scientific record. sign in

arxiv: 2604.06788 · v2 · submitted 2026-04-08 · 💻 cs.CE · cs.CL· cs.MA

Recognition: no theorem link

From Perception to Autonomous Computational Modeling: A Multi-Agent Approach

Daniel N. Wilke
Authors on Pith no claims yet

Pith reviewed 2026-05-10 17:51 UTC · model grok-4.3

classification 💻 cs.CE cs.CLcs.MA
keywords multi-agent systemslarge language modelscomputational mechanicsfinite element analysisautonomous modelinguncertainty quantificationperceptual dataengineering assessment
0
0 comments X

The pith

Coordinated LLM agents can autonomously execute the full computational mechanics workflow from a photograph of an engineering part to a code-compliant analysis report.

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

The paper sets out a framework in which multiple large language model agents collaborate to carry out every stage of structural analysis starting from raw perceptual input such as a photograph. Agents move sequentially through geometry extraction, material property inference, mesh generation, solver runs, uncertainty handling, and limit-state checks against engineering codes, finally producing a report with redesign recommendations. A single demonstration on an L-bracket image shows the pipeline completing without manual edits in its first pass. If the approach holds, routine finite-element work could shift from manual setup to automated execution, leaving engineers primarily in a review role.

Core claim

The authors formalise LLM agents as conditioned operators that act on a shared context space and incorporate quality gates to trigger conditional iteration across pipeline layers. They supply a mathematical treatment for deriving engineering quantities from perceptual data under uncertainty by combining interval bounds, probability densities, and fuzzy membership functions, together with a rule for task-dependent conservatism that resolves conflicting parameter directions across different limit states. The framework is applied end-to-end to a photograph of a steel L-bracket, automatically generating a 171504-node tetrahedral mesh, executing seven separate analyses under three boundary-case假设

What carries the argument

Coordinated LLM agents operating as conditioned operators on a shared context space with quality gates that enforce conditional iteration, supported by an uncertainty model using interval bounds, probability densities, and fuzzy functions plus task-dependent conservatism.

If this is right

  • The entire analysis chain from image to engineering report can complete in one autonomous iteration without manual correction.
  • Uncertainty arising from perceptual data can be represented and propagated using intervals, densities, and fuzzy sets.
  • Opposing trends in material or geometric parameters across different limit states can be handled consistently through task-dependent conservatism.
  • Multiple boundary-condition hypotheses can be explored automatically within the same workflow.
  • The final output includes quantified redesign guidance that satisfies code compliance checks.

Where Pith is reading between the lines

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

  • If the method generalises beyond the single L-bracket example, field photographs could feed directly into on-site structural assessments.
  • The same agent-coordination pattern might be adapted to other simulation domains such as thermal or fluid problems.
  • Repeated application to the same component could allow the system to refine its uncertainty estimates from accumulating results.
  • As underlying language models improve at technical reasoning, the framework's accuracy on geometry and material inference would be expected to increase.

Load-bearing premise

That the LLM agents can extract accurate geometry, infer correct material properties, generate valid meshes, and perform reliable limit-state assessments from a single photograph without introducing errors that quality gates fail to catch.

What would settle it

Running the same pipeline on additional photographs of the identical L-bracket or on a physically tested specimen and finding that the autonomously predicted failure load or mode deviates substantially from laboratory measurements would disprove reliable autonomous performance.

Figures

Figures reproduced from arXiv: 2604.06788 by Daniel N. Wilke.

Figure 1
Figure 1. Figure 1: Three-layer architecture with orchestrator and iterative feedback. The [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: From photograph to finite element mesh. (a) Input image with visual cues used for [PITH_FULL_IMAGE:figures/full_fig_p017_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Autonomous mesh feature audit. The discretisation reviewer agent scans the generated [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Von Mises stress across all three BC variants (LC1, 200 N UDL). Left column: full [PITH_FULL_IMAGE:figures/full_fig_p021_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Deformed shapes (front view, warp ×0.5, common colour scale) for all three BC variants under LC1 (200 N UDL). Nominal: δmax = 10.77 mm; flexible: δmax = 11.15 mm (stiffness-conservative); stiff: δmax = 6.59 mm. The flexible variant produces the largest deflec￾tion, confirming it is conservative for stiffness under load-controlled conditions (Section 5). 0 1000 2000 3000 4000 Peak von Mises stress (MPa) 165… view at source ↗
Figure 6
Figure 6. Figure 6: FEA simulation matrix results across all seven runs. Top: peak von Mises stress with [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Autonomous redesign outcome: baseline (t = 2.5 mm, top) versus redesign RD2 (t = 5.0 mm, bottom) for the nominal BC variant. Left: bracket geometry; centre: inner￾bend mesh detail; right: mesh and result metrics including scaling ratios versus beam-theory predictions. The redesigned bracket achieves conditional pass for LC1 (distributed loading) but fails LC2 (concentrated tip load), confirming that the br… view at source ↗
Figure 8
Figure 8. Figure 8: Two-tier self-improving feedback loop. Prompt-level refinement (olive dashed): engi￾neer corrections are distilled into abstract patterns updating agent definitions and shared mem￾ory, lightweight, interpretable, and reversible. Model-level refinement (red dashed): when the same correction recurs across multiple analyses, the feedback operator triggers supervised retrain￾ing (fine-tuning or preference opti… view at source ↗
read the original abstract

We present a solver-agnostic framework in which coordinated large language model (LLM) agents autonomously execute the complete computational mechanics workflow, from perceptual data of an engineering component through geometry extraction, material inference, discretisation, solver execution, uncertainty quantification, and code-compliant assessment, to an engineering report with actionable recommendations. Agents are formalised as conditioned operators on a shared context space with quality gates that introduce conditional iteration between pipeline layers. We introduce a mathematical framework for extracting engineering information from perceptual data under uncertainty using interval bounds, probability densities, and fuzzy membership functions, and introduce task-dependent conservatism to resolve the ambiguity of what `conservative' means when different limit states are governed by opposing parameter trends. The framework is demonstrated through a finite element analysis pipeline applied to a photograph of a steel L-bracket, producing a 171,504-node tetrahedral mesh, seven analyses across three boundary condition hypotheses, and a code-compliant assessment revealing structural failure with a quantified redesign. All results are presented as generated in the first autonomous iteration without manual correction, reinforcing that a professional engineer must review and sign off on any such analysis.

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 paper claims to present a solver-agnostic multi-agent LLM framework that autonomously executes the full computational mechanics workflow from perceptual input (a photograph of an engineering component) through geometry extraction, material inference, discretization, solver execution, uncertainty quantification, and code-compliant assessment to a final engineering report. Agents operate as conditioned operators on a shared context space equipped with quality gates that enable conditional iteration. A mathematical framework using interval bounds, probability densities, and fuzzy membership functions is introduced for handling uncertainty in perceptual data, along with task-dependent conservatism to address ambiguity in conservative choices across opposing limit-state trends. The framework is demonstrated on a single photograph of a steel L-bracket, yielding a 171,504-node tetrahedral mesh, seven FEA runs under three boundary-condition hypotheses, and a failure assessment with quantified redesign recommendations, all generated autonomously in the first iteration without manual correction.

Significance. If the reliability of the autonomous pipeline can be established, the work would represent a notable step toward integrating perception, modeling, and regulatory assessment in computational engineering. The combination of multi-agent coordination, quality gates, and a mixed uncertainty formalism (intervals, probabilities, fuzzy sets) directly targets practical barriers in applying LLMs to technical workflows. The single-case demonstration, while illustrative, currently limits broader significance; strengthening it with quantitative validation would position the contribution as a foundation for reproducible autonomous analysis tools.

major comments (2)
  1. [Demonstration] Demonstration section: The central claim that coordinated LLM agents autonomously produce a usable 171,504-node tetrahedral mesh, seven FEA runs, and a code-compliant failure assessment directly from a single photograph rests on a single first-iteration run. No quantitative checks are reported—such as comparison of extracted geometry dimensions to the physical L-bracket, mesh-quality metrics (aspect ratio, skewness, or Jacobian determinants) against a reference discretization, material-property inference error relative to known steel values, or sensitivity of the final safety factor to the three boundary-condition hypotheses. Without these, it is impossible to confirm that quality gates prevented propagation of errors or that the reported failure conclusion is based on accurate intermediates rather than plausible artifacts.
  2. [Mathematical framework] Mathematical framework for uncertainty and task-dependent conservatism: The task-dependent conservatism parameters are introduced to resolve ambiguity when different limit states are governed by opposing parameter trends, yet they are listed among the free parameters and appear defined with reference to the specific limit states under evaluation. This risks circularity, as the conservatism choice depends on the very trends the framework is meant to assess objectively, potentially reducing generality beyond the L-bracket case.
minor comments (2)
  1. The description of the shared context space and quality gates would benefit from an explicit pseudocode listing or flowchart showing how iteration is triggered and how information is passed between agents; this would improve reproducibility without altering the technical content.
  2. Notation for the combined interval-probabilistic-fuzzy representation could be standardized more clearly (e.g., consistent symbols for bounds versus membership functions) to avoid ambiguity when reading the uncertainty propagation steps.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We address each major comment below and indicate where revisions will be made to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Demonstration] Demonstration section: The central claim that coordinated LLM agents autonomously produce a usable 171,504-node tetrahedral mesh, seven FEA runs, and a code-compliant failure assessment directly from a single photograph rests on a single first-iteration run. No quantitative checks are reported—such as comparison of extracted geometry dimensions to the physical L-bracket, mesh-quality metrics (aspect ratio, skewness, or Jacobian determinants) against a reference discretization, material-property inference error relative to known steel values, or sensitivity of the final safety factor to the three boundary-condition hypotheses. Without these, it is impossible to confirm that quality gates prevented propagation of errors or that the reported failure conclusion is based on accurate intermediates rather than plausible artifacts.

    Authors: We agree that the demonstration would benefit from additional context on validation. The manuscript presents a proof-of-concept for autonomous end-to-end execution from perceptual input, with all outputs generated in a single first-iteration run without manual correction. Because the sole input is an unlabeled photograph, no reference measurements or ground-truth model exist for direct quantitative comparison of geometry, material properties, or mesh quality. The quality gates enforce internal consistency within the pipeline layers rather than external accuracy. In revision we will expand the demonstration section to explicitly discuss this limitation, report any additional mesh statistics derivable from the generated output (e.g., element type distribution and node count already stated), and reiterate that the framework requires professional engineering review before use. This preserves the paper's focus on autonomy while addressing the concern. revision: partial

  2. Referee: [Mathematical framework] Mathematical framework for uncertainty and task-dependent conservatism: The task-dependent conservatism parameters are introduced to resolve ambiguity when different limit states are governed by opposing parameter trends, yet they are listed among the free parameters and appear defined with reference to the specific limit states under evaluation. This risks circularity, as the conservatism choice depends on the very trends the framework is meant to assess objectively, potentially reducing generality beyond the L-bracket case.

    Authors: The task-dependent conservatism parameters are user-specified inputs chosen a priori on the basis of known engineering relationships between parameters and opposing limit states; they are not computed from or conditioned on the numerical results of the current analysis. Their role is to encode application-specific judgment when the direction of conservatism is ambiguous, while the subsequent uncertainty propagation and assessment remain objective. We will revise the mathematical-framework section to state this distinction clearly, emphasize that the parameters are selected independently of analysis outcomes, and include guidance on their general selection to improve applicability beyond the demonstrated case. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper presents a multi-agent LLM framework for end-to-end computational mechanics from perceptual data, introducing interval/fuzzy uncertainty handling and task-dependent conservatism as new formalisms. No equations or steps reduce by construction to inputs (no self-definitional loops, no fitted parameters relabeled as predictions, no load-bearing self-citations or imported uniqueness theorems). The single L-bracket demonstration is offered as an autonomous run rather than a tautological output, and the central claims rest on external LLM capabilities plus the introduced formalism rather than circular reduction. The derivation remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 1 invented entities

The framework rests on the assumption that LLMs can act as reliable conditioned operators and that the introduced conservatism resolves ambiguity without introducing new fitting parameters; limited details available from abstract.

free parameters (1)
  • task-dependent conservatism parameters
    Introduced to resolve opposing trends in different limit states; values not specified but appear chosen per task.
axioms (2)
  • domain assumption LLM agents can be formalised as conditioned operators on a shared context space that execute engineering tasks reliably when quality gates are applied.
    Invoked to justify autonomous execution without manual correction.
  • standard math Perceptual data can be converted to engineering quantities using interval bounds, probability densities, and fuzzy membership functions.
    Standard uncertainty representations assumed to suffice for geometry and material inference.
invented entities (1)
  • shared context space with quality gates no independent evidence
    purpose: Enables conditional iteration between pipeline layers for autonomous correction.
    New coordination mechanism introduced for the agent workflow.

pith-pipeline@v0.9.0 · 5490 in / 1519 out tokens · 33549 ms · 2026-05-10T17:51:28.792339+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

71 extracted references · 58 canonical work pages · 14 internal anchors

  1. [1]

    M. Chen, J. Tworek, H. Jun, Q. Yuan, H. P. de Oliveira Pinto, J. Kaplan, H. Ed- wards, Y. Burda, N. Joseph, G. Brockman, A. Ray, R. Puri, G. Krueger, M. Petrov, H. Khlaaf, G. Sastry, P. Mishkin, B. Chan, S. Gray, N. Ryder, M. Pavlov, A. Power, L. Kaiser, M. Bavarian, C. Winter, P. Tillet, F. P. Such, D. Cummings, M. Plappert, F. Chantzis, E. Barnes, A. He...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2107.03374 2021

  2. [2]

    R. Li, L. B. Allal, Y. Zi, N. Muennighoff, D. Kocetkov, C. Mou, M. Marone, C. Akiki, J. Li, J. Chim, Q. Liu, E. Zheltonozhskii, T. Y. Zhuo, T. Wang, O. Dehaene, M. Davaadorj, J. Lamy-Poirier, J. Monteiro, O. Shliazhko, N. Gontier, N. Meade, A. Zebaze, M. H. Yee, L. K. Umapathi, J. Zhu, B. Lipkin, M. Oblokulov, Z. Wang, R. Murthy, J. Stillerman, S. S. Pate...

    work page internal anchor Pith review doi:10.48550/arxiv.2305.06161 2023

  3. [3]

    Galactica: A Large Language Model for Science

    R. Taylor, M. Kardas, G. Cucurull, T. Scialom, A. Hartshorn, E. Saravia, A. Poulton, V. Kerkez, R. Stojnic, Galactica: A large language model for science, arXiv preprint arXiv:2211.09085, 2022. doi:10.48550/arXiv.2211.09085

    work page internal anchor Pith review doi:10.48550/arxiv.2211.09085 2022

  4. [4]

    Solving Quantitative Reasoning Problems with Language Models

    A. Lewkowycz, A. Andreassen, D. Dohan, E. Dyer, H. Michalewski, V. Ramasesh, A. Slone, C. Anil, I. Schlag, T. Gutman-Solo, Y. Wu, B. Neyshabur, G. Gur-Ari, V. Misra, Solving quantitative reasoning problems with language models, in: Advances in Neural Information Processing Systems 35, 2022. doi:10.48550/arXiv.2206.14858

    work page internal anchor Pith review doi:10.48550/arxiv.2206.14858 2022

  5. [5]

    GPT-4 Technical Report

    OpenAI, GPT-4 technical report, arXiv preprint arXiv:2303.08774, 2023. doi:10.48550/arXiv.2303.08774

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2303.08774 2023

  6. [6]

    Computer Methods in Applied Mechanics and Engineering 403, 115731

    K. Linka, E. Kuhl, A new family of Constitutive Artificial Neural Networks towards au- tomated model discovery, Computer Methods in Applied Mechanics and Engineering 403, Part A, (2023) 115731. doi:10.1016/j.cma.2022.115731

    work page doi:10.1016/j.cma.2022.115731 2023

  7. [7]

    M.J. Buehler, MechGPT, a language-based strategy for mechanics and materials modeling that connects knowledge across scales, disciplines and modalities, Applied Mechanics Re- views 76 (2) (2024) 021001. doi:10.1115/1.4063843

    work page doi:10.1115/1.4063843 2024

  8. [8]

    J. Guo, C. Park, D. Qian, T.J.R. Hughes, W.K. Liu, Large language model-empowered next-generation computer-aided engineering, Computer Methods in Applied Mechanics and Engineering 450 (2026) 118591. doi:10.1016/j.cma.2025.118591

    work page doi:10.1016/j.cma.2025.118591 2026

  9. [9]

    Y. Qi, R. Xu, X. Chu, FeaGPT: an end-to-end agentic-AI for finite element analysis, arXiv preprint arXiv:2510.21993, 2025. doi:10.48550/arXiv.2510.21993

    work page doi:10.48550/arxiv.2510.21993 2025

  10. [10]

    ALL-FEM: Agentic Large Language models Fine-tuned for Finite Element Methods

    R. Deotale, A. Srinivasan, Y. Tian, T. Zhang, P. P. Vlachos, and H. Gomez. ALL-FEM: Agentic Large Language models Fine-tuned for Finite Element Methods. arXiv preprint arXiv:2603.21011, 2026. doi:10.48550/arXiv.2603.21011 29

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2603.21011 2026

  11. [11]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

    M. Raissi, P. Perdikaris, G.E. Karniadakis, Physics-informed neural networks: A deep learn- ing framework for solving forward and inverse problems involving nonlinear partial differ- ential equations, Journal of Computational Physics 378 (2019) 686–707. doi:10.1016/j.jcp.2018.10.045

    work page doi:10.1016/j.jcp.2018.10.045 2019

  12. [12]

    Kevrekidis, Lu Lu, Paris Perdikaris, Sifan Wang, and Liu Yang

    G.E. Karniadakis, I.G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, L. Yang, Physics-informed machine learning, Nature Reviews Physics 3 (2021) 422–440. doi:10.1038/s42254-021-00314-5

    work page doi:10.1038/s42254-021-00314-5 2021

  13. [13]

    L. Lu, P. Jin, G. Pang, Z. Zhang, G.E. Karniadakis, Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators, Nature Machine Intelligence 3 (2021) 218–229. doi:10.1038/s42256-021-00302-5

    work page doi:10.1038/s42256-021-00302-5 2021

  14. [14]

    Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, A. Anandku- mar, Fourier neural operator for parametric partial differential equations, in: International Conference on Learning Representations, 2021. doi:10.48550/arXiv.2010.08895

    work page internal anchor Pith review doi:10.48550/arxiv.2010.08895 2021

  15. [15]

    Keyak, I.Y

    J.H. Keyak, I.Y. Lee, H.B. Skinner, Correlations between orthogonal mechanical properties and density of trabecular bone: use of different densitometric measures, Journal of Biomed- ical Materials Research 28 (11) (1994) 1329–1336. doi:10.1002/jbm.820281111

    work page doi:10.1002/jbm.820281111 1994

  16. [16]

    Hollister, N

    S.J. Hollister, N. Kikuchi, Homogenization theory and digital imaging: A basis for studying the mechanics and design principles of bone tissue, Biotechnology and Bioengineering 43 (7) (1994) 586–596. doi:10.1002/bit.260430708

    work page doi:10.1002/bit.260430708 1994

  17. [17]

    Wooldridge, An Introduction to MultiAgent Systems, 2nd Edition, John Wiley & Sons,

    M. Wooldridge, An Introduction to MultiAgent Systems, 2nd Edition, John Wiley & Sons,

  18. [18]

    ISBN: 978-0-470-51946-2

  19. [19]

    Dorri, S.S

    A. Dorri, S.S. Kanhere, R. Jurdak, Multi-agent systems: A survey, IEEE Access 6 (2018) 28573–28593. doi:10.1109/ACCESS.2018.2831228

    work page doi:10.1109/access.2018.2831228 2018

  20. [20]

    S. Hong, M. Zhuge, J. Chen, X. Zheng, Y. Cheng, C. Zhang, J. Wang, Z. Wang, S. K. S. Yau, Z. Lin, L. Zhou, C. Ran, L. Xiao, C. Wu, and J. Schmidhuber, MetaGPT: Meta program- ming for a multi-agent collaborative framework, in: International Conference on Learning Representations, 2024. doi:10.48550/arXiv.2308.00352

    work page internal anchor Pith review doi:10.48550/arxiv.2308.00352 2024

  21. [21]

    Q. Wu, G. Bansal, J. Zhang, Y. Wu, B. Li, E. Zhu, L. Jiang, X. Zhang, S. Zhang, J. Liu, A. H. Awadallah, R. W. White, D. Burger, and C. Wang, AutoGen: Enabling next-gen LLM applications via multi-agent conversation, First Conference on Language Modeling, 1328, 2024. COLM: 1328

    2024

  22. [22]

    S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, Y. Cao, ReAct: Synergizing reasoning and acting in language models, in: International Conference on Learning Repre- sentations, 2023. doi:10.48550/arXiv.2210.03629

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2210.03629 2023

  23. [23]

    CorpusID:258833055

    N.Shinn, F.Cassano, E.Berman, A.Gopinath, K.Narasimhan, S.Yao, Reflexion: Language agents with verbal reinforcement learning, in: Advances in Neural Information Processing 30 Systems 36, 2023. CorpusID:258833055

    2023

  24. [24]

    Autonomous chemical research with large language models

    D.A. Boiko, R. MacKnight, B. Kline, G. Gomes, Autonomous chemical research with large language models, Nature 624 (2023) 570–578. doi:10.1038/s41586-023-06792-0

    work page doi:10.1038/s41586-023-06792-0 2023

  25. [25]

    A.M. Bran, S. Cox, O. Schilter, C. Baldassari, A.D. White, P. Schwaller, Augmenting large- language models with chemistry tools, Nature Machine Intelligence 6 (5) (2024) 525–535. doi:10.1038/s42256-024-00832-8

    work page doi:10.1038/s42256-024-00832-8 2024

  26. [26]

    URLhttps://doi.org/10

    C. Geuzaine, J.-F. Remacle, Gmsh: A 3-D finite element mesh generator with built-in pre- and post-processing facilities, International Journal for Numerical Methods in Engineering 79 (11) (2009) 1309–1331. doi:10.1002/nme.2579

    work page doi:10.1002/nme.2579 2009

  27. [27]

    Dhondt, The Finite Element Method for Three-Dimensional Thermomechanical Appli- cations, Wiley, 2004

    G. Dhondt, The Finite Element Method for Three-Dimensional Thermomechanical Appli- cations, Wiley, 2004. doi:10.1002/0470021217

    work page doi:10.1002/0470021217 2004

  28. [28]

    Blacker, W.J

    T.D. Blacker, W.J. Bohnhoff, T.L. Edwards, J.R. Hipp, R.R. Lober, S.A. Mitchell, G.D. Sjaardema, T.J. Tautges, T.J. Wilson, W.R. Oakes, CUBIT mesh generation envi- ronment, Volume 1: Users manual, Technical Report SAND94-1100, Sandia National Lab- oratories, 1994. doi:10.2172/10176386

    work page doi:10.2172/10176386 1994

  29. [29]

    Eurocode 3: Design of steel structures

    CEN, EN 1993-1-1:2005 – Eurocode 3: Design of Steel Structures, Brussels, 2005. Eurocode 3: Design of steel structures

    1993

  30. [30]

    CEN, EN 1990:2002

    CEN, EN 1990:2002 – Eurocode: Basis of Structural Design, Brussels, 2002. CEN, EN 1990:2002

    1990

  31. [31]

    https://www.aisc.org/aisc/publications/current-standards/aisc-360/

    AISC, ANSI/AISC 360-22, Chicago, 2022. https://www.aisc.org/aisc/publications/current-standards/aisc-360/

    2022

  32. [32]

    ISBN: 978-1-76072-947-9

    Standards Australia, AS 4100:2020 – Steel Structures, Sydney, 2020. ISBN: 978-1-76072-947-9

    2020

  33. [33]

    9781760720612

    StandardsAustralia/StandardsNewZealand, AS/NZS4600:2018–Cold-formedSteelStruc- tures, Sydney, 2018. 9781760720612

    2018

  34. [34]

    Oberkampf, C.J

    W.L. Oberkampf, C.J. Roy, Verification and Validation in Scientific Computing, Cambridge University Press, 2010. doi:10.1017/CBO9780511760396

    work page doi:10.1017/cbo9780511760396 2010

  35. [35]

    Roache, Verification and Validation in Computational Science and Engineering, Her- mosa, 1998

    P.J. Roache, Verification and Validation in Computational Science and Engineering, Her- mosa, 1998. ISBN: 978-0-913478-08-0. 978-0913478080

    1998

  36. [36]

    Pilkey, D.F

    W.D. Pilkey, D.F. Pilkey, Peterson’s Stress Concentration Factors, 3rd Ed., Wiley, 2008. doi:10.1002/9780470211106

    work page doi:10.1002/9780470211106 2008

  37. [37]

    Moore, Interval Analysis, Prentice-Hall, 1966

    R.E. Moore, Interval Analysis, Prentice-Hall, 1966. ISSN 2577-9435

    1966

  38. [38]

    Prince- ton University Press, Princeton, NJ, 2009

    A. Ben-Tal, L. El Ghaoui, A. Nemirovski, Robust Optimization, Princeton University Press, 2009. doi:10.1515/9781400831050 31

    work page doi:10.1515/9781400831050 2009

  39. [39]

    F. Tao, H. Zhang, A. Liu, A.Y.C. Nee, Digital twin in industry: State-of-the-art, IEEE Transactions on Industrial Informatics 15 (4) (2019) 2405–2415. doi:10.1109/TII.2018.2873186

    work page doi:10.1109/tii.2018.2873186 2019

  40. [40]

    https://docs.anthropic.com/en/docs/about-claude/models(accessed 8 April 2026)

    Anthropic, Claude Opus 4.6 Model Card, February 2026. https://docs.anthropic.com/en/docs/about-claude/models(accessed 8 April 2026)

    2026

  41. [41]

    Attention Is All You Need

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A.N. Gomez, Ł. Kaiser, I. Polo- sukhin, Attention is all you need, in: Advances in Neural Information Processing Systems 30, 2017. doi:10.48550/arXiv.1706.03762

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1706.03762 2017

  42. [42]

    Aparticlemethodforhistory-dependentmaterials

    D. Sulsky, Z. Chen, H.L. Schreyer, A particle method for history-dependent materials, Com- puter Methods in Applied Mechanics and Engineering 118 (1–2) (1994) 179–196. doi:10.1016/0045-7825(94)90112-0

    work page doi:10.1016/0045-7825(94)90112-0 1994

  43. [43]

    Sulsky, S.-J

    D. Sulsky, S.-J. Zhou, H.L. Schreyer, Application of a particle-in-cell method to solid me- chanics, Computer Physics Communications 87 (1–2) (1995) 236–252. doi:10.1016/0010-4655(94)00170-7

    work page doi:10.1016/0010-4655(94)00170-7 1995

  44. [44]

    Reformulation of elasticity theory for discontinuities and long-range forces

    S.A. Silling, Reformulation of elasticity theory for discontinuities and long-range forces, Journal of the Mechanics and Physics of Solids 48 (1) (2000) 175–209. doi:10.1016/S0022-5096(99)00029-0

    work page doi:10.1016/s0022-5096(99)00029-0 2000

  45. [45]

    Peridynamic States and Constitutive Modeling

    S.A. Silling, M. Epton, O. Weckner, J. Xu, E. Askari, Peridynamic states and constitutive modeling, Journal of Elasticity 88 (2) (2007) 151–184. doi:10.1007/s10659-007-9125-1

    work page doi:10.1007/s10659-007-9125-1 2007

  46. [46]

    Monaghan, Smoothed particle hydrodynamics, Reports on Progress in Physics 68 (8) (2005) 1703

    J.J. Monaghan, Smoothed particle hydrodynamics, Reports on Progress in Physics 68 (8) (2005) 1703. doi:10.1088/0034-4885/68/8/R01

    work page doi:10.1088/0034-4885/68/8/r01 2005

  47. [47]

    Liu, M.B

    G.R. Liu, M.B. Liu, Smoothed Particle Hydrodynamics, World Scientific, 2003. doi:10.1142/5340

    work page doi:10.1142/5340 2003

  48. [48]

    Zhang, Z

    X. Zhang, Z. Chen, Y. Liu, The Material Point Method: A Continuum-Based Particle Method for Extreme Loading Cases, Academic Press, 2017. https://www.sciencedirect.com/book/monograph/9780124077164/the-material-point- method

    work page arXiv 2017

  49. [49]

    Zienkiewicz, J.Z

    O.C. Zienkiewicz, J.Z. Zhu, A simple error estimator and adaptive procedure for practical engineering analysis, International Journal for Numerical Methods in Engineering 24 (2) (1987) 337–357. doi:10.1002/nme.1620240206

    work page doi:10.1002/nme.1620240206 1987

  50. [50]

    Logg, K.-A

    A. Logg, K.-A. Mardal, G. Wells (Eds.), Automated Solution of Differential Equations by the Finite Element Method: The FEniCS Book, Springer, 2012. doi:10.1007/978-3-642-23099-8

    work page doi:10.1007/978-3-642-23099-8 2012

  51. [51]

    Arndt, W

    D. Arndt, W. Bangerth, B. Blais, M. Fehling, R. Gassmöller, T. Heister, L. Heltai, U. Köcher, M. Kronbichler, M. Maier, P. Munch, J. P. Pelteret, S. Proell, K. Simon, B. Turcksin, D. Wells, and J. Zhang, The deal.II library, version 9.3, Journal of Numer- ical Mathematics 29 (3) (2021) 171–186. doi:10.1515/jnma-2021-0081 32

    work page doi:10.1515/jnma-2021-0081 2021

  52. [52]

    Chourdakis, K

    G. Chourdakis, K. Davis, B. Rodenberg, M. Schulte, F. Simonis, B. Uekermann, G. Abrams, H. J. Bungartz, L. C. Yau, I. Desai, K. Eder, R. Hertrich, F. Lindner, A. Rusch, D. Sashko, D. Schneider, A. Totounferoush, D. Volland, P. Vollmer, and O. Z. Koseomur, preCICE v2: A sustainable and user-friendly coupling library, Open Research Europe 2 (2022) 51. doi:1...

    work page doi:10.12688/openreseurope.14445.2 2022

  53. [53]

    Smilauer, E

    V. Smilauer, E. Catalano, B. Chareyre, S. Dorofeenko, J. Duriez, N. Dyck, J. Elias, B. Er, A. Eulitz, A. Gladky, N. Guo, C. Jakob, F. Kneib, J. Kozicki, D. Marzougui, R. Maurin, C. Modenese, L. Scholtes, L. Sibille, J. Stransky, T. Sweijen, K. Thoeni, C. Yuan, Yade Documentation, 2nd Edition, The Yade Project, 2015. doi:10.5281/zenodo.34073

    work page doi:10.5281/zenodo.34073 2015

  54. [54]

    Weinhart, L

    T. Weinhart, L. Orefice, M. Post, M.P. van Schrojenstein Lantman, I.F.C. Denissen, D.R. Tunuguntla, J.M.F. Tsang, H. Cheng, M.Y. Shaheen, H. Shi, P. Rapino, E. Grannonio, N. Losacco, J. Barbosa, L. Jing, J.E. Alvarez Naranjo, S. Roy, W.K. den Otter, A.R Thorn- ton, Fast, flexible particle simulations—an introduction to MercuryDPM, Computer Physics Communi...

    work page doi:10.1016/j.cpc.2019.107129 2020

  55. [55]

    Schönberger, J.-M

    J.L. Schönberger, J.-M. Frahm, Structure-from-motion revisited, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 4104– 4113. doi:10.1109/CVPR.2016.445

    work page doi:10.1109/cvpr.2016.445 2016

  56. [56]

    Domínguez, G

    J.M. Domínguez, G. Fourtakas, C. Altomare, R.B. Canelas, A. Tafuni, O. García-Feal, I. Martínez-Estévez, A. Mokos, R. Vacondio, A.J.C. Crespo, B.D. Rogers, P.K. Stansby, M. Gómez-Gesteira, DualSPHysics: from fluid dynamics to multiphysics problems, Compu- tational Particle Mechanics 9 (5) (2022) 867–895. doi:10.1007/s40571-021-00404-2

    work page doi:10.1007/s40571-021-00404-2 2022

  57. [57]

    and Aktulga, H

    A.P. Thompson, H.M. Aktulga, R. Berger, D.S. Bolintineanu, W.M. Brown, P.S. Crozier, P.J. in ’t Veld, A. Kohlmeyer, S.G. Moore, T.D. Nguyen, R. Shan, M.J. Stevens, J. Tranchida, C. Trott, S.J. Plimpton, LAMMPS—a flexible simulation tool for particle- based materials modeling at the atomic, meso, and continuum scales, Computer Physics Communications 271 (2...

    work page doi:10.1016/j.cpc.2021.108171 2022

  58. [58]

    Kloss, C

    C. Kloss, C. Goniva, A. Hager, S. Amberger, S. Pirker, Models, algorithms and validation for opensource DEM and CFD–DEM, Progress in Computational Fluid Dynamics 12 (2–3) (2012) 140–152. doi:10.1504/PCFD.2012.047457

    work page doi:10.1504/pcfd.2012.047457 2012

  59. [59]

    Govender, D.N

    N. Govender, D.N. Wilke, S. Kok, Blaze-DEMGPU: Modular high performance DEM frame- work for the GPU architecture, SoftwareX 5 (2016) 62–66. doi:10.1016/j.softx.2016.04.004

    work page doi:10.1016/j.softx.2016.04.004 2016

  60. [60]

    Weller, G

    H.G. Weller, G. Tabor, H. Jasak, C. Fureby, A tensorial approach to computational con- tinuum mechanics using object-oriented techniques, Computers in Physics 12 (6) (1998) 620–631. doi:10.1063/1.168744

    work page doi:10.1063/1.168744 1998

  61. [61]

    Economon, F

    T.D. Economon, F. Palacios, S.R. Copeland, T.W. Lukaczyk, J.J. Alonso, SU2: an open- source suite for multiphysics simulation and design, AIAA Journal 54 (3) (2016) 828–846. doi:10.2514/1.J053813 33

    work page doi:10.2514/1.j053813 2016

  62. [62]

    Generative Adversarial Networks

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Generative adversarial nets, in: Advances in Neural Information Processing Systems 27, 2014. doi:10.48550/arXiv.1406.2661

    work page internal anchor Pith review doi:10.48550/arxiv.1406.2661 2014

  63. [63]

    Training language models to follow instructions with human feedback

    L. Ouyang, J. Wu, X. Jiang, D. Almeida, C. L. Wainwright, P. Mishkin, C. Zhang, S. Agar- wal, K. Slama, A. Ray, J. Schulman, J. Hilton, F. Kelton, L. Miller, M. Simens, A. Askell, P. Welinder, P. Christiano, J. Leike, R. Lowe, Training language models to follow instruc- tions with human feedback, in: Advances in Neural Information Processing Systems 35, 2...

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2203.02155 2022

  64. [64]

    2017 , month = dec, journal =

    P. Christiano, J. Leike, T. B. Brown, M. Martic, S. Legg, D. Amodei, Deep reinforcement learning from human preferences, in: Advances in Neural Information Processing Systems 30, 2017. doi:10.48550/arXiv.1706.03741

    work page doi:10.48550/arxiv.1706.03741 2017

  65. [65]

    Direct Preference Optimization: Your Language Model is Secretly a Reward Model

    R. Rafailov, A. Sharma, E. Mitchell, S. Ermon, C. D. Manning, C. Finn, Direct preference optimization: your language model is secretly a reward model, in: Advances in Neural Information Processing Systems 36, 2023. doi:10.48550/arXiv.2305.18290

    work page internal anchor Pith review doi:10.48550/arxiv.2305.18290 2023

  66. [66]

    2023 , url =

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y. Lo, P. Dollár, R. Girshick, Segment Anything, in: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023, pp. 4015–4026. doi:10.1109/ICCV51070.2023.00371

    work page doi:10.1109/iccv51070.2023.00371 2023

  67. [67]

    Learning Transferable Visual Models From Natural Language Supervision

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, I. Sutskever, Learning transferable visual models from natural language supervision, in: Proceedings of the 38th International Conference on Ma- chine Learning, PMLR 139, 2021, pp. 8748–8763. doi:10.48550/arXiv.2103.00020

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2103.00020 2021

  68. [68]

    L. A. Zadeh, Fuzzy sets, Information and Control 8 (3) (1965) 338–353. doi:10.1016/S0019-9958(65)90241-X

    work page doi:10.1016/s0019-9958(65)90241-x 1965

  69. [69]

    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. doi:10.1016/j.eml.2024.102131

    work page doi:10.1016/j.eml.2024.102131 2024

  70. [70]

    C. J. Roy, W. L. Oberkampf, A comprehensive framework for verification, validation, and uncertainty quantification in scientific computing, Computer Methods in Applied Mechanics and Engineering 200 (25–28) (2011) 2131–2144. doi:10.1016/j.cma.2011.03.016

    work page doi:10.1016/j.cma.2011.03.016 2011

  71. [71]

    ISBN 9780791873168

    ASME, Standard for Verification and Validation in Computational Solid Mechanics, ASME V&V 10-2019, American Society of Mechanical Engineers, New York, 2019. ISBN 9780791873168. 34

    2019