pith. sign in

arxiv: 2605.28164 · v1 · pith:2BBD5OFBnew · submitted 2026-05-27 · 💻 cs.NE · cs.AI

Performance and Explainability Requirements of Evolutionary Algorithms in Real-World Physics-Informed Optimization

Helena Stegherr , Michael Heider , Nils Meyer , Tobias Thummerer , Thomas Wendler , Pierre Aublin , Ennio Idrobo-\`Avila , Lars Mikelsons
show 2 more authors
Sebastian Zaunseder J\"org H\"ahner
This is my paper

Pith reviewed 2026-06-29 09:33 UTC · model grok-4.3

classification 💻 cs.NE cs.AI
keywords evolutionary algorithmsphysics-informed optimizationexplainabilityperformance requirementsreal-world applicationsdomain expert inputtrust and usability
0
0 comments X

The pith

Domain experts in five physics optimization problems all require evolutionary algorithms to converge quickly and supply some explanations of results, with other demands varying by problem.

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

The paper presents five real-world physics-based optimization problems along with the performance and explainability requirements stated by their domain experts. It finds that fast convergence to a good solution and some insight into how results were reached are expected in every case, while needs such as constraint handling or multi-objective support differ across applications. This matters because without addressing these points, evolutionary methods are often overlooked by practitioners who work with physics models and need trustworthy tools. The authors also note existing techniques that could meet the requirements but have not been tested in these complex settings, pointing to an unclosed gap between evolutionary computation research and physics practice.

Core claim

Through the five problems and expert input, the central finding is that universal requirements focus on rapid convergence and partial explainability to build trust, while additional requirements are problem-dependent, and that leveraging known but unused methods from evolutionary computation could close the gap and increase usability in physics-informed optimization.

What carries the argument

The set of five physics problems and the requirements collected directly from their domain experts, used to separate common performance and explainability needs from problem-specific ones.

If this is right

  • Meeting the common requirements would make evolutionary algorithms more likely to be adopted in physics modeling contexts.
  • Problem-specific tailoring beyond the shared needs remains necessary for full usability.
  • Applying existing convergence and explanation techniques that have not yet been tried in these settings could reduce the identified gap.

Where Pith is reading between the lines

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

  • Similar requirement-gathering exercises in other optimization domains might reveal whether fast convergence and partial explainability are broadly shared demands.
  • Testing the unused techniques on the five problems could produce concrete evidence of improved expert trust.
  • The variation in other requirements suggests that one-size-fits-all evolutionary algorithms will continue to face adoption barriers without customization.

Load-bearing premise

The five chosen physics problems and the requirements voiced by their experts stand in for the performance and explainability needs that would raise trust and adoption of evolutionary algorithms across physics-informed work.

What would settle it

Finding a set of additional physics optimization problems whose domain experts do not list fast convergence or any form of explanation among their top requirements would undermine the claimed common needs.

Figures

Figures reproduced from arXiv: 2605.28164 by Ennio Idrobo-\`Avila, Helena Stegherr, J\"org H\"ahner, Lars Mikelsons, Michael Heider, Nils Meyer, Pierre Aublin, Sebastian Zaunseder, Thomas Wendler, Tobias Thummerer.

Figure 1
Figure 1. Figure 1: Illustration of the SCARA, consisting of two actuated rotational axes (red), and a pen (blue) at the end-effector to draw on a sheet of paper. The friction force created by the contact between the pen and the paper must be overcome by the two motors that drive the two axes. The farther away the pen is from the mounting point of the robot, the more torque the motors have to apply to overcome static friction… view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of a planar fiber patch placement problem: A structural component is assembled from [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of the shape optimization problem: A quarter-symmetrical plate is loaded at two edges [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: A DeepONet should learn an operator that maps an arbitrary input function [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of the acquisition setup for Electrical Impedance Tomography ( [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: 3D dynamic PET frames (dPET) depict the distribution of a radioactive tracer over space and time, and [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
read the original abstract

Evolutionary computation offers a variety of tools to solve complex real-world optimization problems. However, research often focuses on smaller, simplified problems and optimization algorithms that sometimes miss expectations in real-world scenarios. Additionally, trust in the applied algorithm and the solutions it provides is often essential in such settings, but requires an understanding of the search process itself. This leads to evolutionary computation often not being seriously considered by practitioners in many application contexts, among them physics-based modeling. In this article, techniques from evolutionary computation are detailed that can alleviate these problems. First, five real-world physics-based optimization problems are introduced and described by domain experts. For each of these, the requirements for the evolutionary algorithm regarding performance and explainability to increase trust and usability are presented. We found that all domain experts expect fast convergence to a good solution and want some explanations for how the results were formed, while other requirements strongly depend on the respective problem. Finally, we present existing approaches that can be leveraged to improve those aspects of evolutionary algorithms but have to our knowledge never been employed in complex real-world scenarios. This implies a gap between both domains that needs to be closed to exploit the full potential of evolutionary computation.

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

1 major / 0 minor

Summary. The manuscript introduces five real-world physics-based optimization problems described by domain experts and presents the performance and explainability requirements that evolutionary algorithms must meet to increase trust and usability in these settings. It reports that every expert requires fast convergence to a good solution together with some explanation of how results were obtained, while additional requirements are problem-dependent. The paper then surveys existing EC techniques that could address these needs but have not, to the authors' knowledge, been deployed in complex real-world physics scenarios, thereby identifying a gap between the two domains.

Significance. If the elicited requirements prove representative, the work could usefully direct evolutionary-computation research toward algorithms that satisfy the convergence speed and minimal explainability demands common across physics-informed problems. The explicit mapping of unmet needs to existing but unused techniques supplies a concrete research agenda.

major comments (1)
  1. [Introduction and problem-description sections] Sections describing the five problems and their requirements (Introduction and the problem-specific sections): the manuscript states that requirements were obtained from domain experts but supplies no information on the number of experts consulted per problem, the elicitation protocol (structured interview, survey, etc.), how conflicts among experts were resolved, validation steps, or the criteria used to select both the problems and the experts. Without these details the central empirical claim—that all experts require fast convergence plus some explanations—cannot be assessed for selection bias or generalizability.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for highlighting the need for greater transparency in how domain-expert requirements were elicited. We address this point directly below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: Sections describing the five problems and their requirements (Introduction and the problem-specific sections): the manuscript states that requirements were obtained from domain experts but supplies no information on the number of experts consulted per problem, the elicitation protocol (structured interview, survey, etc.), how conflicts among experts were resolved, validation steps, or the criteria used to select both the problems and the experts. Without these details the central empirical claim—that all experts require fast convergence plus some explanations—cannot be assessed for selection bias or generalizability.

    Authors: We agree that the manuscript currently omits these methodological details, which limits the ability to evaluate selection bias and generalizability. In the revised version we will insert a dedicated subsection (likely in the Introduction or as a new Methods paragraph) that reports: the number of experts consulted for each problem, the elicitation protocol employed, the procedure for resolving any conflicting views, the validation steps taken, and the criteria used to select both the problems and the participating experts. This addition will directly support the central claim and allow readers to assess its robustness. revision: yes

Circularity Check

0 steps flagged

No circularity; claims are direct reports of expert opinions with no derivation or self-referential reduction

full rationale

The paper introduces five real-world physics-based problems and their domain-expert requirements, then summarizes that all experts expect fast convergence plus some explanations while others are problem-dependent. This is a direct aggregation of external inputs rather than any mathematical derivation, fitted parameter renamed as prediction, or self-citation chain. No equations, ansatzes, or uniqueness theorems appear. The representativeness of the sample is a validity question outside the circularity criteria.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The central claim rests on the representativeness of the five problems and the accuracy of the domain experts' stated needs; the abstract introduces no free parameters, mathematical axioms, or new postulated entities.

pith-pipeline@v0.9.1-grok · 5780 in / 1021 out tokens · 29575 ms · 2026-06-29T09:33:09.012653+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

97 extracted references · 86 canonical work pages · 3 internal anchors

  1. [1]

    In Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19)

    Local optima networks for continuous fitness landscapes. In Proceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19). ACM, New York, NY, USA, Performance and Explainability Requirements of EAs in Real-World Physics-Informed Optimization 23 1407–1414. doi:10.1145/3319619.3326852 Andy Adler, John H Arnold, Richard Bayford, An...

    work page doi:10.1145/3319619.3326852

  2. [2]

    doi:10.1088/0967-3334/30/6/S03 Oleg Agafonov, Aleksandar Babic, Sonia Sousa, and Sharmini Alagaratnam

    GREIT: a unified approach to 2D linear EIT reconstruction of lung images.Physiological Measurement30, 6 (2009), S35. doi:10.1088/0967-3334/30/6/S03 Oleg Agafonov, Aleksandar Babic, Sonia Sousa, and Sharmini Alagaratnam

    work page doi:10.1088/0967-3334/30/6/s03 2009

  3. [3]

    Frontiers in Digital Health6 (2024), 1427233

    Editorial: Trustworthy AI for healthcare. Frontiers in Digital Health6 (2024), 1427233. doi:10.3389/fdgth.2024.1427233 Noor A. Aziz, Awais Manzoor, Muhammad Deedahwar Mazhar Qureshi, M. Atif Qureshi, and Wael Rashwan

    work page doi:10.3389/fdgth.2024.1427233 2024

  4. [4]

    Unveiling Explainable AI in Healthcare: Current Trends, Challenges, and Future Directions. (2024). doi:10.1101/2024.08.10.24311735 Jaume Bacardit, Alexander E. I. Brownlee, Stefano Cagnoni, Giovanni Iacca, John McCall, and David Walker

    work page doi:10.1101/2024.08.10.24311735 2024

  5. [5]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion(Boston, Massachusetts)(GECCO ’22)

    The intersection of evolutionary computation and explainable AI. InProceedings of the Genetic and Evolutionary Computation Conference Companion(Boston, Massachusetts)(GECCO ’22). ACM, New York, NY, USA, 1757–1762. doi:10.1145/3520304. 3533974 Nathan Baker, Frank Alexander, Timo Bremer, Aric Hagberg, Yannis Kevrekidis, Habib Najm, Manish Parashar, Abani Pa...

    work page doi:10.1145/3520304

  6. [6]

    Workshop Report on Basic Research Needs for Scientific Machine Learning: Core Technologies for Artificial Intelligence. (2019). doi:10.2172/1478744 Vicente-José Bevia, Carlos Andreu-Vilarroig, Juan-Carlos Cortés, and Rafael-Jacinto Villanueva

    work page doi:10.2172/1478744 2019

  7. [7]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22)

    Probability density function computation in evolutionary model calibration with uncertainty. InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22). ACM, New York, NY, USA, 1902–1908. doi:10.1145/3520304.3534017 Marco Evangelos Biancolini, Andrea Chiappa, Ubaldo Cella, Emiliano Costa, Corrado Groth, and Stefano Porziani

    work page doi:10.1145/3520304.3534017 1902

  8. [8]

    InComputational Science – ICCS 2020, Valeria V

    Radial Basis Functions Mesh Morphing. InComputational Science – ICCS 2020, Valeria V. Krzhizhanovskaya, Gábor Závodszky, Michael H. Lees, Jack J. Dongarra, Peter M. A. Sloot, Sérgio Brissos, and João Teixeira (Eds.). Springer International Publishing, Cham, 294–308. G. S. Jr. Bjorkman and R. Jr. Richards

    2020

  9. [9]

    doi:10.1115/1.3423882 Rania Bouzid, Jyotindra Narayan, and Hassène Gritli

    Harmonic Holes—An Inverse Problem in Elasticity.Journal of Applied Mechanics 43, 3 (1976), 414–418. doi:10.1115/1.3423882 Rania Bouzid, Jyotindra Narayan, and Hassène Gritli

    work page doi:10.1115/1.3423882 1976

  10. [10]

    In2024 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies (3ICT)

    Hybrid Metaheuristic and Artificial Neural Network Approach for Solving Inverse Kinematics of a SCARA Manipulator Robot. In2024 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies (3ICT). 385–392. doi:10.1109/3ict64318.2024.10824671 Borja Calvo, Ofer M. Shir, Josu Ceberio, Carola Doerr, Hao Wang, Thomas Bäc...

    work page doi:10.1109/3ict64318.2024.10824671 2024

  11. [11]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19)

    Bayesian perfor- mance analysis for black-box optimization benchmarking. InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19). ACM, New York, NY, USA, 1789–1797. doi:10.1145/3319619.3326888 Hongqing Cao, Lishan Kang, Yuping Chen, and Jingxian Yu

    work page doi:10.1145/3319619.3326888

  12. [12]

    Iacopo Cappellini, Lorenzo Campagnola, and Guglielmo Consales

    Evolutionary modeling of systems of ordinary differential equations with genetic programming.Genetic Programming and Evolvable Machines1, 4 (2000), 309–337. Iacopo Cappellini, Lorenzo Campagnola, and Guglielmo Consales

    2000

  13. [13]

    Electrical Impedance Tomography, Artificial Intelligence, and Variable Ventilation: Transforming Respiratory Monitoring and Treatment in Critical Care.Journal of Personalized Medicine14, 7 (2024),

    2024

  14. [14]

    doi:10.3390/jpm14070677 Richard E. Carson

    work page doi:10.3390/jpm14070677

  15. [15]

    InPositron Emission Tomography: Basic Sciences, Dale L

    Tracer Kinetic Modeling in PET. InPositron Emission Tomography: Basic Sciences, Dale L. Bailey, David W. Townsend, Peter E. Valk, and Michael N. Maisey (Eds.). Springer London, London, 127–159. doi:10.1007/1- 84628-007-9_6 Josu Ceberio, Juan-Carlos Cortés, Francisco Fernández de Vega, Oscar Garnica, J. Ignacio Hidalgo, J. Manuel Velasco, and Rafael-Jacint...

    work page doi:10.1007/1-

  16. [16]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22)

    Approaching epistemic and aleatoric uncertainty with evolutionary optimization: examples and challenges. InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22). ACM, New York, NY, USA, 1909–1915. doi:10.1145/3520304.3533978 Cheng Chen, Haokai Hong, Wanyu Lin, and Kay Chen Tan

    work page doi:10.1145/3520304.3533978 1909

  17. [17]

    In2025 IEEE Congress on Evolutionary Computation (CEC)

    A Physics-Informed Evolutionary Transfer Optimization Framework for Material Design. In2025 IEEE Congress on Evolutionary Computation (CEC). 1–9. doi:10.1109/CEC65147. 2025.11042916 Min-you Chen, Gang Hu, Wei He, Yan-li Yang, and Jin-qian Zhai. 2010.A Reconstruction Method for Electrical Impedance Tomography Using Particle Swarm Optimization. Lecture Note...

    work page doi:10.1109/cec65147 2025

  18. [18]

    doi:10.1007/978-3-642-15597-0_38 Margaret Cheney, David Isaacson, and Jonathan C

    Springer Berlin Heidelberg, Berlin, Heidelberg, 342–350. doi:10.1007/978-3-642-15597-0_38 Margaret Cheney, David Isaacson, and Jonathan C. Newell

    work page doi:10.1007/978-3-642-15597-0_38

  19. [19]

    doi:10.1137/S0036144598333613 Manuel Chica, Angel A

    Electrical Impedance Tomography.SIAM Rev.41, 1 (1999), 85–101. doi:10.1137/S0036144598333613 Manuel Chica, Angel A. Juan, Christopher Bayliss, Oscar Cordón, and W. David Kelton

    work page doi:10.1137/s0036144598333613 1999

  20. [20]

    doi:10.2436/20.8080.02.104 Francesca De Benetti, Walter Simson, Magdalini Paschali, Hasan Sari, Axel Rominger, Kuangyu Shi, Nassir Navab, and Thomas Wendler

    Why simheuristics? Benefits, limitations, and best practices when combining metaheuristics with simulation.SORT-Statistics and Operations Research Transactions44, 2 (2020), 311–334. doi:10.2436/20.8080.02.104 Francesca De Benetti, Walter Simson, Magdalini Paschali, Hasan Sari, Axel Rominger, Kuangyu Shi, Nassir Navab, and Thomas Wendler

    work page doi:10.2436/20.8080.02.104 2020

  21. [21]

    InMedical Image Computing and Computer Assisted Intervention – MICCAI 2023 (Lecture Notes in Computer Science), 24 Stegherr et al

    Self-supervised Learning for Physiologically-Based Pharmacokinetic Modeling in Dynamic PET. InMedical Image Computing and Computer Assisted Intervention – MICCAI 2023 (Lecture Notes in Computer Science), 24 Stegherr et al. Hayit Greenspan, Anant Madabhushi, Parvin Mousavi, Septimiu Salcudean, James Duncan, Tanveer Syeda-Mahmood, and Russell Taylor (Eds.)....

    work page doi:10.1007/978-3-031-43907-0_28 2023

  22. [22]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19)

    An analysis of dimensionality reduction techniques for visualizing evolution. InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’19). ACM, New York, NY, USA, 1864–1872. doi:10.1145/3319619.3326868 Jacob de Nobel, Diederick Vermetten, Anna V. Kononova, Ofer M. Shir, and Thomas Bäck

    work page doi:10.1145/3319619.3326868

  23. [23]

    In vivo electrical conductivity of hepatic tumours.Physiological Measurement24, 2 (2003),

    2003

  24. [24]

    Sotiriadis

    doi:10.1088/0967-3334/24/2/302 Christos Dimas, Vassilis Alimisis, Nikolaos Uzunoglu, and Paul P. Sotiriadis

    work page doi:10.1088/0967-3334/24/2/302

  25. [25]

    doi:10.1109/ACCESS.2024.3382939 Wellington Dos Santos, Ricardo de Souza, Reiga Ribeiro, Allan Feitosa, Valter Barbosa, Victor Silva, David Ribeiro, and Rafaela Covello de Freitas

    Advances in Electrical Impedance Tomography Inverse Problem Solution Methods: From Traditional Regularization to Deep Learning.IEEE Access12 (2024), 47797–47829. doi:10.1109/ACCESS.2024.3382939 Wellington Dos Santos, Ricardo de Souza, Reiga Ribeiro, Allan Feitosa, Valter Barbosa, Victor Silva, David Ribeiro, and Rafaela Covello de Freitas. 2018.Electrical...

    work page doi:10.1109/access.2024.3382939 2024

  26. [26]

    doi:10.1109/TEVC.2021.3092343 Erika Antonette T

    Searching for Robustness Intervals in Evolutionary Robust Optimization.IEEE Transactions on Evolutionary Computation26, 1 (2022), 58–72. doi:10.1109/TEVC.2021.3092343 Erika Antonette T. Enriquez, Renier G. Mendoza, and Arrianne Crystal T. Velasco

    work page doi:10.1109/tevc.2021.3092343 2022

  27. [27]

    doi:10.1109/ACCESS.2022.3158357 Benjamin Fischer, Bernhard Horn, Christian Bartelt, and Yannick Blößl

    Philippine Eagle Optimization Algorithm.IEEE Access10 (2022), 29089–29120. doi:10.1109/ACCESS.2022.3158357 Benjamin Fischer, Bernhard Horn, Christian Bartelt, and Yannick Blößl

    work page doi:10.1109/access.2022.3158357 2022

  28. [28]

    doi:10.5923/j

    Method for an Automated Optimization of Fiber Patch Placement Layup Designs.International Journal of Composite Materials5, 2 (2015), 37–46. doi:10.5923/j. cmaterials.20150502.03 Martin Fyvie. 2024.Explainability of non-deterministic solvers: explanatory feature generation from the data mining of the search trajectories of population-based metaheuristics. ...

    work page doi:10.5923/j 2015

  29. [29]

    InProceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO ’23 Companion)

    Explaining a Staff Rostering Genetic Algorithm using Sensitivity Analysis and Trajectory Analysis. InProceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO ’23 Companion). ACM, New York, NY, USA, 1648–1656. doi:10.1145/3583133.3596353 Tobias Glasmachers

    work page doi:10.1145/3583133.3596353

  30. [30]

    arXiv:2505.10987 [math.OC] https://arxiv.org/ abs/2505.10987 Xavier Glorot and Yoshua Bengio

    A Superlinearly Convergent Evolution Strategy. arXiv:2505.10987 [math.OC] https://arxiv.org/ abs/2505.10987 Xavier Glorot and Yoshua Bengio

    work page arXiv

  31. [31]

    doi:10.1080/00207549508904789 Z

    A genetic algorithm approach to curve fitting.International Journal of Production Research33, 7 (1995), 1911–1923. doi:10.1080/00207549508904789 Z. Hashin

    work page doi:10.1080/00207549508904789 1995

  32. [32]

    doi:10.1115/1.3153664 Michael Heider, Marcus Albrecht, Johannes Schilp, and Jörg Hähner

    Failure Criteria for Unidirectional Fiber Composites.Journal of Applied Mechanics47, 2 (1980), 329–334. doi:10.1115/1.3153664 Michael Heider, Marcus Albrecht, Johannes Schilp, and Jörg Hähner

    work page doi:10.1115/1.3153664 1980

  33. [33]

    Michael Heider, Helena Stegherr, Richard Nordsieck, and Jörg Hähner

    Assessing Stakeholder Perspectives on the Explainability of AI Solutions for Smart Production Planning with Just-In-Time Logistics.Expert Systems with Applications (2026). Michael Heider, Helena Stegherr, Richard Nordsieck, and Jörg Hähner

    2026

  34. [34]

    Assessing model requirements for explainable AI: a template and exemplary case study.Artificial Life29, 4 (2023), 468 –

    2023

  35. [35]

    doi:10.1162/artl_a_00414 Chih-Kang Huang, Weichung Wang, Kai-Yuan Tzen, Win-Li Lin, and Cheng-Ying Chou

    work page doi:10.1162/artl_a_00414

  36. [36]

    In2012 9th IEEE International Symposium on Biomedical Imaging (ISBI)

    FDOPA kinetics analysis in PET images for Parkinson’s disease diagnosis by use of particle swarm optimization. In2012 9th IEEE International Symposium on Biomedical Imaging (ISBI). 586–589. doi:10.1109/ISBI.2012.6235616 Marija Kacarska and Suzana Loskovska. 2002.Comparative analysis of time efficiency and spatial resolution between different EIT reconstru...

    work page doi:10.1109/isbi.2012.6235616 2012

  37. [37]

    doi:10.1109/TMI.2005.845317 Seung Kwan Kang, Seongho Seo, Chul-Hee Lee, Mi Jeong Kim, Su Jin Kim, and Jae Sung Lee

    Direct reconstruction of kinetic parameter images from dynamic PET data.IEEE Transactions on Medical Imaging24, 5 (2005), 636–650. doi:10.1109/TMI.2005.845317 Seung Kwan Kang, Seongho Seo, Chul-Hee Lee, Mi Jeong Kim, Su Jin Kim, and Jae Sung Lee

    work page doi:10.1109/tmi.2005.845317 2005

  38. [38]

    Physica Medica72 (2020), 60–72

    Robust nonlinear parameter estimation in tracer kinetic analysis using infinity norm regularization and particle swarm optimization. Physica Medica72 (2020), 60–72. doi:10.1016/j.ejmp.2020.03.013 Mustafa Karakaplan

    work page doi:10.1016/j.ejmp.2020.03.013 2020

  39. [39]

    doi:10.1016/j.aca.2007.01.058 Performance and Explainability Requirements of EAs in Real-World Physics-Informed Optimization 25 Ralph Kussmaul, Markus Zogg, and Paolo Ermanni

    Fitting Lorentzian peaks with evolutionary genetic algorithm based on stochastic search procedure.Analytica Chimica Acta587, 2 (2007), 235–239. doi:10.1016/j.aca.2007.01.058 Performance and Explainability Requirements of EAs in Real-World Physics-Informed Optimization 25 Ralph Kussmaul, Markus Zogg, and Paolo Ermanni

    work page doi:10.1016/j.aca.2007.01.058 2007

  40. [40]

    doi:10.1016/j.compstruct.2019.111165 Fernando Lezama, Jose Almeida, Joao Soares, and Zita Vale

    A failure mechanics and strength optimization study for patched laminates.Composite Structures226 (2019), 111165. doi:10.1016/j.compstruct.2019.111165 Fernando Lezama, Jose Almeida, Joao Soares, and Zita Vale

    work page doi:10.1016/j.compstruct.2019.111165 2019

  41. [41]

    In2023 IEEE Symposium Series on Computational Intelligence (SSCI)

    Explainergy: Towards Explainability of Metaheuristic Performance in the Energy Field. In2023 IEEE Symposium Series on Computational Intelligence (SSCI). 783–788. doi:10. 1109/SSCI52147.2023.10371844 Yinlin Li and Bijoy K Kundu

    work page arXiv 2023

  42. [42]

    doi:10.1088/1361-6560/aaac02 Jiao Liu, Abhishek Gupta, Chinchun Ooi, and Yew-Soon Ong

    An improved optimization algorithm of the three-compartment model with spillover and partial volume corrections for dynamic FDG PET images of small animal hearts in vivo.Physics in Medicine & Biology63, 5 (2018), 055003. doi:10.1088/1361-6560/aaac02 Jiao Liu, Abhishek Gupta, Chinchun Ooi, and Yew-Soon Ong

    work page doi:10.1088/1361-6560/aaac02 2018

  43. [43]

    doi:10.1109/TEVC.2023.3349313 Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karniadakis

    ExTrEMO: Transfer Evolutionary Multiobjective Optimization With Proof of Faster Convergence.IEEE Transactions on Evolutionary Computation29, 1 (2025), 102–116. doi:10.1109/TEVC.2023.3349313 Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karniadakis

    work page doi:10.1109/tevc.2023.3349313 2025

  44. [44]

    doi:10.1038/s42256-021-00302-5 Lu Lu, Raphaël Pestourie, Steven G

    Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators.Nature Machine Intelligence3, 3 (2021), 218–229. doi:10.1038/s42256-021-00302-5 Lu Lu, Raphaël Pestourie, Steven G. Johnson, and Giuseppe Romano

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

  45. [45]

    Multifidelity deep neural operators for efficient learning of partial differential equations with application to fast inverse design of nanoscale heat transport.Phys. Rev. Res.4 (2022), 023210. Issue

    2022

  46. [46]

    doi:10.1103/PhysRevResearch.4.023210 Yingbo Ma, Vaibhav Dixit, Michael J Innes, Xingjian Guo, and Chris Rackauckas

    work page doi:10.1103/physrevresearch.4.023210

  47. [47]

    doi:10.3390/a14020040 Sofiene Mansouri, Yousef Alharbi, Fatma Haddad, Souhir Chabcoub, Anwar Alshrouf, and Amr A

    A Survey of Advances in Landscape Analysis for Optimisation.Algorithms14, 2 (2021). doi:10.3390/a14020040 Sofiene Mansouri, Yousef Alharbi, Fatma Haddad, Souhir Chabcoub, Anwar Alshrouf, and Amr A. Abd-Elghany

    work page doi:10.3390/a14020040 2021

  48. [48]

    doi:10.2478/joeb-2021-0007 Sebastien Martin and Charles T

    Electrical Impedance tomography – recent applications and developments.Journal of Electrical Bioimpedance12, 1 (2021), 50–62. doi:10.2478/joeb-2021-0007 Sebastien Martin and Charles T. M. Choi

    work page doi:10.2478/joeb-2021-0007 2021

  49. [49]

    doi:10.1109/TMAG

    Nonlinear Electrical Impedance Tomography Reconstruction Using Artificial Neural Networks and Particle Swarm Optimization.IEEE Transactions on Magnetics52, 3 (2016), 1–4. doi:10.1109/TMAG. 2015.2488901 Emmanuel Martínez-Guerrero, Pedro Lagos-Eulogio, Pedro Miranda-Romagnoli, Roberto Noriega-Papaqui, and Juan Carlos Seck-Tuoh-Mora

    work page doi:10.1109/tmag 2016

  50. [50]

    Historical Elite Differential Evolution Based on Particle Swarm Optimization Algorithm for Texture Optimization with Application in Particle Physics.Applied Sciences14, 19 (2024),

    2024

  51. [51]

    Mattheck

    doi:10.3390/app14199110 C. Mattheck

    work page doi:10.3390/app14199110

  52. [52]

    doi:10.1111/j.1460- 2695.1990.tb00623.x David Issa Mattos, Jan Bosch, and Helena Holmström Olsson

    DESIGN AND GROWTH RULES FOR BIOLOGICAL STRUCTURES AND THEIR APPLICATION TO ENGINEERING.Fatigue & Fracture of Engineering Materials & Structures13, 5 (1990), 535–550. doi:10.1111/j.1460- 2695.1990.tb00623.x David Issa Mattos, Jan Bosch, and Helena Holmström Olsson

    work page doi:10.1111/j.1460- 1990

  53. [53]

    Statistical Models for the Analysis of Optimization Algorithms With Benchmark Functions.IEEE Transactions on Evolutionary Computation25, 6 (2021), 1163–1177. doi:10. 1109/TEVC.2021.3081167 Michael Merry, Pat Riddle, and Jim Warren

    work page arXiv 2021

  54. [54]

    BMC Medical Informatics and Decision Making21, 1 (2021)

    A mental models approach for defining explainable artificial intelligence. BMC Medical Informatics and Decision Making21, 1 (2021). doi:10.1186/s12911-021-01703-7 Efrén Mezura-Montes and Carlos A. Coello Coello

    work page doi:10.1186/s12911-021-01703-7 2021

  55. [55]

    doi:10.1016/j.swevo.2011.10.001 Isabelle Miederer, Kuangyu Shi, and Thomas Wendler

    Constraint-handling in nature-inspired numerical optimization: Past, present and future.Swarm and Evolutionary Computation1, 4 (2011), 173–194. doi:10.1016/j.swevo.2011.10.001 Isabelle Miederer, Kuangyu Shi, and Thomas Wendler

    work page doi:10.1016/j.swevo.2011.10.001 2011

  56. [56]

    Nuklearmedizin - NuclearMedicine62, 6 (2023), 370–378

    Machine learning methods for tracer kinetic modelling. Nuklearmedizin - NuclearMedicine62, 6 (2023), 370–378. doi:10.1055/a-2179-5818 Publisher: Georg Thieme Verlag KG. A. Murari, Riccardo Rossi, Luca Spolladore, Ivan Wyss, and Michela Gelfusa

    work page doi:10.1055/a-2179-5818 2023

  57. [57]

    doi:10.1007/s10462-025- 11282-y Ana Nikolikj, Mario Andrés Muñoz, and Tome Eftimov

    Informed machine learning to reconcile interpretability with fidelity in scientific applications.Artificial Intelligence Review58 (2025). doi:10.1007/s10462-025- 11282-y Ana Nikolikj, Mario Andrés Muñoz, and Tome Eftimov. 2025a. Benchmarking Footprints of Continuous Black-Box Optimization Algorithms: Explainable Insights into Algorithm Success and Failure...

    work page doi:10.1007/s10462-025- 2025

  58. [58]

    doi:10.1016/j.asoc.2021.107492 R

    Search trajectory networks: A tool for analysing and visualising the behaviour of metaheuristics.Applied Soft Computing109 (2021), 107492. doi:10.1016/j.asoc.2021.107492 R. Olmi, M. Bini, and S. Priori

    work page doi:10.1016/j.asoc.2021.107492 2021

  59. [59]

    doi:10.1109/4235.843497 Eneko Osaba, Esther Villar-Rodriguez, Javier Del Ser, Antonio J

    A genetic algorithm approach to image reconstruction in electrical impedance tomography.IEEE Transactions on Evolutionary Computation4, 1 (2000), 83–88. doi:10.1109/4235.843497 Eneko Osaba, Esther Villar-Rodriguez, Javier Del Ser, Antonio J. Nebro, Daniel Molina, Antonio LaTorre, Ponnuthurai N. Suganthan, Carlos A. Coello Coello, and Francisco Herrera

    work page doi:10.1109/4235.843497 2000

  60. [60]

    application of metaheuristic algorithms to real-world optimization problems.Swarm and Evolutionary Computation64 (2021), 100888

    A tutorial on the design, experimentation and 26 Stegherr et al. application of metaheuristic algorithms to real-world optimization problems.Swarm and Evolutionary Computation64 (2021), 100888. doi:10.1016/j.swevo.2021.100888 Austin R. Pantel, Varsha Viswanath, Mark Muzi, Robert K. Doot, and David A. Mankoff

    work page doi:10.1016/j.swevo.2021.100888 2021

  61. [61]

    arXiv:https://jnm.snmjournals.org/content/63/3/342.full.pdf doi:10.2967/jnumed.121.263518 Alexandros I

    Principles of Tracer Kinetic Analysis in Oncology, Part I: Principles and Overview of Methodology.Journal of Nuclear Medicine63, 3 (2022), 342–352. arXiv:https://jnm.snmjournals.org/content/63/3/342.full.pdf doi:10.2967/jnumed.121.263518 Alexandros I. Papadopoulos, Savvas I. Raptis, Antonios Lalas, Konstantinos Votis, Dimitrios Tyrovolas, Sotiris A. Tegos...

    work page doi:10.2967/jnumed.121.263518 2022

  62. [62]

    doi:10.1109/ MCOM.001.2300582 Tobias Pfaff, Meire Fortunato, Alvaro Sanchez-Gonzalez, and Peter W

    Physics-Informed Metaheuristics for Fast RIS Codebook Compilation.IEEE Communications Magazine62, 11 (2024), 152–158. doi:10.1109/ MCOM.001.2300582 Tobias Pfaff, Meire Fortunato, Alvaro Sanchez-Gonzalez, and Peter W. Battaglia

    2024

  63. [63]

    Battaglia

    Learning Mesh-Based Simulation with Graph Networks.CoRRabs/2010.03409 (2020). arXiv:2010.03409 https://arxiv.org/abs/2010.03409 Dimitris C. Psichogios and Lyle H. Ungar

    work page arXiv 2010

  64. [64]

    AIChE Journal38, 10 (1992), 1499–1511

    A hybrid neural network-first principles approach to process modeling. AIChE Journal38, 10 (1992), 1499–1511. doi:10.1002/aic.690381003 Christopher Rackauckas, Yingbo Ma, Julius Martensen, Collin Warner, Kirill Zubov, Rohit Supekar, Dominic Skin- ner, Ali Ramadhan, and Alan Edelman

    work page doi:10.1002/aic.690381003 1992

  65. [65]

    Universal Differential Equations for Scientific Machine Learning

    Universal Differential Equations for Scientific Machine Learning. arXiv:2001.04385 [cs.LG] https://arxiv.org/abs/2001.04385 Khalid Raza and Rafat Parveen

    work page internal anchor Pith review Pith/arXiv arXiv 2001

  66. [66]

    Evolutionary algorithms in genetic regulatory networks model

    Evolutionary algorithms in genetic regulatory networks model.arXiv preprint arXiv:1205.1986(2012). Quentin Renau, Johann Dreo, Carola Doerr, and Benjamin Doerr

    work page internal anchor Pith review Pith/arXiv arXiv 1986

  67. [67]

    Composite Structures116 (2014), 48–54

    Tailored patch placement on a base load carrying laminate: A computational structural optimisation with experimental validation. Composite Structures116 (2014), 48–54. doi:10.1016/j.compstruct.2014.04.028 Tobias Rodemann and Christiane Attig

    work page doi:10.1016/j.compstruct.2014.04.028 2014

  68. [68]

    On the definition and importance of interpretability in scientific machine learning

    On the definition and importance of interpretability in scientific machine learning. arXiv:2505.13510 [cs.LG] https://arxiv.org/abs/2505.13510 Tim Sabsch, Christian Braune, Alexander Dockhorn, and Rudolf Kruse

    work page internal anchor Pith review Pith/arXiv arXiv

  69. [69]

    In2017 IEEE Symposium Series on Computational Intelligence (SSCI)

    Using a multiobjective genetic algorithm for curve approximation. In2017 IEEE Symposium Series on Computational Intelligence (SSCI). 1–6. doi:10.1109/SSCI.2017. 8285179 Sophie Sadler, Alma Rahat, David Walker, and Daniel Archambault

    work page doi:10.1109/ssci.2017 2017

  70. [70]

    InProceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO ’23 Companion)

    Extrema Graphs: Fitness Landscape Analysis to the Extreme!. InProceedings of the Companion Conference on Genetic and Evolutionary Computation (GECCO ’23 Companion). ACM, New York, NY, USA, 2081–2089. doi:10.1145/3583133.3596343 Saurav Z. K. Sajib and Rosalind Sadleir

    work page doi:10.1145/3583133.3596343 2081

  71. [71]

    InElectrical Properties of Tissues: Quantitative Magnetic Resonance Mapping, Rosalind Sadleir and Atul Singh Minhas (Eds.)

    Magnetic Resonance Electrical Impedance Tomography. InElectrical Properties of Tissues: Quantitative Magnetic Resonance Mapping, Rosalind Sadleir and Atul Singh Minhas (Eds.). Springer International Publishing, Cham, 157–183. doi:10.1007/978-3-031-03873-0_7 Joel Sansana, Mark N. Joswiak, Ivan Castillo, Zhenyu Wang, Ricardo Rendall, Leo H. Chiang, and Marc...

    work page doi:10.1007/978-3-031-03873-0_7

  72. [73]

    arXiv:2406.09699 [math.NA] https://arxiv.org/abs/2406.09699 Malek Sarhani, Stefan Voß, and Raka Jovanovic

    Differentiable Programming for Differential Equations: A Review. arXiv:2406.09699 [math.NA] https://arxiv.org/abs/2406.09699 Malek Sarhani, Stefan Voß, and Raka Jovanovic

    work page arXiv

  73. [74]

    Initialization of metaheuristics: comprehensive review, crit- ical analysis, and research directions.International Transactions in Operational Research30, 6 (2023), 3361–3397. arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1111/itor.13237 doi:10.1111/itor.13237 Hasan Sari, Clemens Mingels, Ian Alberts, Jicun Hu, Dorothee Buesser, Vijay Shah, Robin Schep...

    work page doi:10.1111/itor.13237 2023

  74. [75]

    doi:10.1007/s00259-021-05623-6 Jörg Schubert, Rimvydas Simutis, Michael Dors, Ivo Havlík, and Andreas Lübbert

    First results on kinetic modelling and parametric imaging of dynamic 18F-FDG datasets from a long axial FOV PET scanner in oncological patients.European Journal of Nuclear Medicine and Molecular Imaging49, 6 (2022), 1997–2009. doi:10.1007/s00259-021-05623-6 Jörg Schubert, Rimvydas Simutis, Michael Dors, Ivo Havlík, and Andreas Lübbert

    work page doi:10.1007/s00259-021-05623-6 2022

  75. [76]

    Hybrid modelling of yeast production processes – combination of a priori knowledge on different levels of sophistication.Chemical Engineering & Technology 17, 1 (1994), 10–20. arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/ceat.270170103 doi:10.1002/ceat.270170103 Khemraj Shukla, Vivek Oommen, Ahmad Peyvan, Michael Penwarden, Nicholas Plewacki, Lui...

    work page doi:10.1002/ceat.270170103 1994

  76. [77]

    Performance and Explainability Requirements of EAs in Real-World Physics-Informed Optimization 27 Engineering Applications of Artificial Intelligence129 (2024), 107615

    Deep neural operators as accurate surrogates for shape optimization. Performance and Explainability Requirements of EAs in Real-World Physics-Informed Optimization 27 Engineering Applications of Artificial Intelligence129 (2024), 107615. doi:10.1016/j.engappai.2023.107615 Manjinder Singh, Alexander E. I. Brownlee, and David Cairns

    work page doi:10.1016/j.engappai.2023.107615 2024

  77. [78]

    InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22)

    Towards explainable metaheuristic: mining surrogate fitness models for importance of variables. InProceedings of the Genetic and Evolutionary Computation Conference Companion (GECCO ’22). ACM, New York, NY, USA, 1785–1793. doi:10.1145/3520304.3533966 L. Sokoloff, M. Reivich, C. Kennedy, M. H. Des Rosiers, C. S. Patlak, K. D. Pettigrew, O. Sakurada, and M....

    work page doi:10.1145/3520304.3533966

  78. [79]

    doi:10.1111/j.1471-4159.1977.tb10649.x Helena Stegherr, Michael Heider, and Jörg Hähner

    THE [14C]DEOXYGLUCOSE METHOD FOR THE MEASUREMENT OF LOCAL CEREBRAL GLUCOSE UTILIZATION: THEORY, PROCEDURE, AND NORMAL VALUES IN THE CONSCIOUS AND ANESTHETIZED ALBINO RAT.Journal of Neurochemistry28, 5 (1977), 897–916. doi:10.1111/j.1471-4159.1977.tb10649.x Helena Stegherr, Michael Heider, and Jörg Hähner. 2025a. A Comparison of Clustering Approaches for M...

    work page doi:10.1111/j.1471-4159.1977.tb10649.x 1977

  79. [80]

    InProceedings of the Companion Conference on Genetic and Evolutionary Computation(Lisbon, Portugal)(GECCO ’23 Companion)

    Neuroevo- lution of Physics-Informed Neural Nets: Benchmark Problems and Comparative Results. InProceedings of the Companion Conference on Genetic and Evolutionary Computation(Lisbon, Portugal)(GECCO ’23 Companion). ACM, New York, NY, USA, 2144–2151. doi:10.1145/3583133.3596397 Carl-Magnus Svensson, Stephen Coombes, and Jonathan Westley Peirce

    work page doi:10.1145/3583133.3596397

  80. [81]

    doi:10.1007/s12021-012-9140-7 Kartikay Tehlan and Thomas Wendler

    Using Evolutionary Algorithms for Fitting High-Dimensional Models to Neuronal Data.Neuroinformatics10, 2 (2012), 199–218. doi:10.1007/s12021-012-9140-7 Kartikay Tehlan and Thomas Wendler

    work page doi:10.1007/s12021-012-9140-7 2012

Showing first 80 references.