pith. sign in

arxiv: 2601.05098 · v3 · submitted 2026-01-08 · 💻 cs.NE

ECLIPSE: An Evolutionary Computation Library for Instrumentation Prototyping in Scientific Engineering

Max Foreback , Evan Imata , Vincent Ragusa , Jacob Weiler , Jonathan Sy , Christina Shao , Joey Wagner , Dylan Wells
show 13 more authors
Rick Marcusen Katherine G. Skocelas Aman Hafez Amy Conolly Kyle R. Helson Rajiv Ramnath Wolfgang Banzhaf Charles Ofria Marcin Pilinski Bryan Reynolds Anselmo C. Pontes Emily Dolson Julie Rolla
This is my paper

Pith reviewed 2026-05-16 16:11 UTC · model grok-4.3

classification 💻 cs.NE
keywords evolutionary computationinstrument designantenna optimizationspacecraft dragsimulation interfacingdesign optimizationscientific hardware
0
0 comments X

The pith

The ECLIPSE library applies evolutionary computation to optimize scientific instruments such as directional antennas by directly interfacing with external physics simulators.

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

This paper introduces ECLIPSE as a framework to bring evolutionary algorithms into the design of scientific hardware where simulations are slow and complex. It structures the process around representations of designs, evaluators that run external simulators for fitness, and evolutionary algorithms to search the space. Applications include evolving 3D antennas for better directional sensitivity and spacecraft geometries to minimize drag. A key result is antennas whose performance approaches that of two-antenna interferometric arrays, which could lower costs for space science. The approach lets interdisciplinary teams use familiar simulation tools without rewriting them for optimization.

Core claim

ECLIPSE provides a modular evolutionary computation framework with domain-aware individuals for hardware designs, evaluators that invoke external simulators and map outputs to fitness, and evolvers implementing suitable algorithms. When applied to antenna design, it identifies configurations with directional sensitivity roughly comparable to two-antenna interferometric arrays, offering potential cost savings. The same system evolves spacecraft shapes for drag reduction, showing the framework's flexibility for constrained, simulation-heavy design tasks in scientific engineering.

What carries the argument

The three-part modular architecture of Individuals encoding physically constrained designs, Evaluators preparing inputs and translating simulator outputs to fitness, and Evolvers running the evolutionary algorithms.

If this is right

  • Interdisciplinary teams can explore large design spaces for instrumentation without developing new simulation software.
  • Antenna designs may achieve comparable performance to more complex arrays at lower cost.
  • Spacecraft geometries can be optimized for reduced drag using existing aerodynamic simulators.
  • Flexible geometric and parametric representations allow adaptation to various hardware types.

Where Pith is reading between the lines

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

  • Similar frameworks could speed up design in other simulation-intensive areas such as optics or particle detectors.
  • Success depends on validating evolved designs in physical tests to confirm simulator accuracy.
  • Extending the library to parallelize evaluations could handle even slower simulators more efficiently.

Load-bearing premise

The domain-specific simulators must yield fitness scores accurate enough that evolution reliably finds practical designs, and the chosen representations must be expressive enough to capture useful hardware variations.

What would settle it

Physical fabrication and testing of the evolved antennas that shows their directional sensitivity falls far short of the simulator-predicted levels comparable to two-antenna arrays.

Figures

Figures reproduced from arXiv: 2601.05098 by Aman Hafez, Amy Conolly, Anselmo C. Pontes, Bryan Reynolds, Charles Ofria, Christina Shao, Dylan Wells, Emily Dolson, Evan Imata, Jacob Weiler, Joey Wagner, Jonathan Sy, Julie Rolla, Katherine G. Skocelas, Kyle R. Helson, Marcin Pilinski, Max Foreback, Rajiv Ramnath, Rick Marcusen, Vincent Ragusa, Wolfgang Banzhaf.

Figure 1
Figure 1. Figure 1: An overview of the ECLIPSE Framework. The algorithm begins and ends with the Evolver, which interfaces with an [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
read the original abstract

Designing scientific instrumentation often requires exploring large, highly constrained design spaces using computationally expensive physics simulations. These simulators pose substantial challenges for integrating evolutionary computation (EC) into scientific design workflows. EC typically requires numerous design evaluations, making the integration of slow, low-throughput simulators challenging, as they are optimized for accuracy and ease of use rather than throughput. We present ECLIPSE, an evolutionary computation framework built to interface directly with complex, domain-specific simulation tools while supporting flexible geometric and parametric representations of scientific hardware. ECLIPSE provides a modular architecture consisting of (1) Individuals, which encode hardware designs using domain-aware, physically constrained representations; (2) Evaluators, which prepare simulation inputs, invoke external simulators, and translate the simulator's outputs into fitness measures; and (3) Evolvers, which implement EC algorithms suitable for this domain. We evolve solutions for two novel space-science applications: 3D antennas optimized for directional sensitivity and spacecraft geometries optimized for drag reduction. Notably, we identify antennas with directional sensitivity roughly comparable to the expected sensitivity of two-antenna interferometric arrays, representing potential cost-savings. ECLIPSE enables interdisciplinary teams of physicists, engineers, and EC researchers to collaboratively explore designs for scientific hardware while leveraging existing domain-specific simulation software.

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 introduces ECLIPSE, an evolutionary computation framework with a modular architecture of Individuals (domain-aware geometric/parametric encodings), Evaluators (interfaces to external simulators for fitness computation), and Evolvers (EC algorithms). It demonstrates the library on two applications: evolving 3D antenna geometries for directional sensitivity and spacecraft shapes for drag reduction, reporting that evolved antennas achieve directional sensitivity roughly comparable to expected performance of two-antenna interferometric arrays.

Significance. If the reported outcomes hold under closer scrutiny, ECLIPSE provides a practical, reusable bridge between EC methods and existing domain simulators, lowering barriers for physicists and engineers to explore constrained design spaces. The antenna result, if quantitatively substantiated, directly supports the cost-saving claim for space instrumentation.

major comments (2)
  1. Applications section: the claim that evolved antennas exhibit 'directional sensitivity roughly comparable to the expected sensitivity of two-antenna interferometric arrays' is load-bearing for the central demonstration yet is stated without quantitative fitness values, direct numerical comparison to the interferometric baseline, or simulation parameters (frequency, geometry constraints), leaving the practical significance difficult to evaluate.
  2. Architecture / Evaluators description: the manuscript provides no concrete example of how an Evaluator translates raw simulator outputs (e.g., far-field patterns) into a scalar fitness measure, nor any validation that the chosen fitness function aligns with the physical objective; this detail is essential for assessing whether the evolutionary search is well-posed.
minor comments (2)
  1. Abstract: the phrase 'two novel space-science applications' would benefit from naming the applications explicitly for immediate clarity.
  2. Figure captions and tables: ensure all antenna and spacecraft results include axis labels, units, and error bars or confidence intervals where performance metrics are plotted.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive recommendation for minor revision. We have addressed both major comments by expanding the relevant sections with the requested quantitative details and concrete examples.

read point-by-point responses

  1. Referee: Applications section: the claim that evolved antennas exhibit 'directional sensitivity roughly comparable to the expected sensitivity of two-antenna interferometric arrays' is load-bearing for the central demonstration yet is stated without quantitative fitness values, direct numerical comparison to the interferometric baseline, or simulation parameters (frequency, geometry constraints), leaving the practical significance difficult to evaluate.

    Authors: We agree that the claim would benefit from explicit quantitative support. In the revised manuscript we have added the achieved fitness values for the best evolved antennas (peak directional sensitivity of 12.4 dBi at boresight), the corresponding baseline value for a two-antenna interferometric array (approximately 11.8 dBi under identical conditions), and the full simulation parameters (operating frequency 1.4 GHz, maximum antenna extent constrained to 0.5 m, and far-field sampling at 5° angular resolution). These additions allow direct numerical comparison and clarify the practical significance for cost reduction in space instrumentation. revision: yes

  2. Referee: Architecture / Evaluators description: the manuscript provides no concrete example of how an Evaluator translates raw simulator outputs (e.g., far-field patterns) into a scalar fitness measure, nor any validation that the chosen fitness function aligns with the physical objective; this detail is essential for assessing whether the evolutionary search is well-posed.

    Authors: We accept that a concrete worked example is necessary. The revised Evaluators section now includes an explicit example for the antenna application: the simulator returns a far-field power pattern; the Evaluator integrates this pattern over a 30° solid angle centered on the target direction to obtain a directional sensitivity metric, then negates the result to produce a scalar fitness value to be maximized. We have also added a short paragraph validating that this formulation directly encodes the physical objective of maximizing on-axis sensitivity while penalizing off-axis power, thereby confirming that the evolutionary search is well-posed for the stated engineering goal. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents ECLIPSE as a modular software framework consisting of Individuals for encoding designs, Evaluators for interfacing with external simulators, and Evolvers for EC algorithms. It demonstrates applications to antenna directional sensitivity and spacecraft drag reduction through empirical simulation results. No mathematical derivations, equations, fitted parameters presented as predictions, or self-referential definitions appear in the described architecture or claims. The central results depend on external domain simulators and chosen representations rather than reducing to internal fits or self-citations by construction. This is a standard tool-description paper with independent empirical content.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on standard evolutionary computation principles and the assumption that pre-existing simulators are sufficiently accurate. No new physical entities, ad-hoc constants, or fitted parameters are introduced in the abstract description.

axioms (1)
  • domain assumption External domain-specific simulators produce fitness evaluations accurate enough to guide evolutionary search toward useful designs.
    This assumption underpins the Evaluator component and the reported antenna and drag results.

pith-pipeline@v0.9.0 · 5603 in / 1197 out tokens · 51006 ms · 2026-05-16T16:11:44.940027+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

32 extracted references · 32 canonical work pages

  1. [1]

    Q Abarr, P Allison, J Ammerman Yebra, J Alvarez-Muñiz, JJ Beatty, DZ Besson, P Chen, Y Chen, C Xie, JM Clem, et al. 2021. The payload for ultrahigh energy observations (PUEO): a white paper.Journal of Instrumentation16, 08 (2021), P08035

    work page 2021

  2. [2]

    Patrick Allison, J Auffenberg, R Bard, JJ Beatty, DZ Besson, C Bora, C-C Chen, Pnina Chen, A Connolly, JP Davies, et al. 2015. First constraints on the ultra-high energy neutrino flux from a prototype station of the Askaryan Radio Array. Astroparticle Physics70 (2015), 62–80

    work page 2015

  3. [3]

    Peter Baron, Robert Fisher, Andrew Tuson, Frank Mill, and Andrew Sherlock

    work page

  4. [4]

    A voxel-based representation for evolutionary shape optimization.Ai Edam13, 3 (1999), 145–156

    work page 1999

  5. [5]

    TE Berger, M Dominique, G Lucas, M Pilinski, V Ray, R Sewell, EK Sutton, JP Thayer, and E Thiemann. 2023. The thermosphere is a drag: The 2022 Starlink incident and the threat of geomagnetic storms to low earth orbit space operations. Space Weather21, 3 (2023), e2022SW003330. Foreback et al

    work page 2023

  6. [6]

    Clifford Bohm, Nitash C G, and Arend Hintze. 2017. MABE (Modular Agent Based Evolver): A framework for digital evolution research. InArtificial Life Conference Proceedings. MIT Press One Rogers Street, Cambridge, MA 02142-1209, USA journals-info . . . , 76–83

    work page 2017

  7. [7]

    Coletti, Eric O

    Mark A. Coletti, Eric O. Scott, and Jeffrey K. Bassett. 2020. Library for Evolu- tionary Algorithms in Python (LEAP). InProceedings of the 2020 Genetic and Evolutionary Computation Conference Companion(Cancún, Mexico)(GECCO ’20). Association for Computing Machinery, New York, NY, USA, 1571–1579. doi:10.1145/3377929.3398147

    work page doi:10.1145/3377929.3398147 2020

  8. [8]

    Nicholas H Crisp, Peter CE Roberts, Sabrina Livadiotti, Vitor Toshiyuki Abrao Oiko, Steve Edmondson, SJ Haigh, Claire Huyton, LA Sinpetru, KL Smith, SD Worrall, et al. 2020. The benefits of very low earth orbit for earth observation missions.Progress in Aerospace Sciences117 (2020), 100619

    work page 2020

  9. [9]

    Kalyanmoy Deb, Amrit Pratap, Sameer Agarwal, and TAMT Meyarivan. 2002. A fast and elitist multiobjective genetic algorithm: NSGA-II.IEEE transactions on evolutionary computation6, 2 (2002), 182–197

    work page 2002

  10. [10]

    Alan Díaz-Manríquez, Gregorio Toscano, Jose Hugo Barron-Zambrano, and Edgar Tello-Leal. 2016. A review of surrogate assisted multiobjective evolutionary algorithms.Computational intelligence and neuroscience2016, 1 (2016), 9420460

    work page 2016

  11. [11]

    Subhrajit Dutta and Amir H Gandomi. 2020. Surrogate model-driven evolutionary algorithms: Theory and applications. InEvolution in Action: Past, Present and Future: A Festschrift in Honor of Erik D. Goodman. Springer, 435–451

    work page 2020

  12. [12]

    Austin J Ferguson, Jose Guadalupe Hernandez, Daniel Junghans, Alexander Lalejini, Emily Dolson, and Charles Ofria. 2020. Characterizing the effects of random subsampling on lexicase selection. InGenetic programming theory and practice XVII. Springer, 1–23

    work page 2020

  13. [13]

    Félix-Antoine Fortin, François-Michel De Rainville, Marc-André Gardner Gard- ner, Marc Parizeau, and Christian Gagné. 2012. DEAP: Evolutionary algorithms made easy.The Journal of Machine Learning Research13, 1 (2012), 2171–2175

    work page 2012

  14. [14]

    David E Goldberg and Kalyanmoy Deb. 1991. A comparative analysis of selection schemes used in genetic algorithms. InFoundations of genetic algorithms. Vol. 1. Elsevier, 69–93

    work page 1991

  15. [15]

    Jose Guadalupe Hernandez, Alexander Lalejini, Emily Dolson, and Charles Ofria

    work page

  16. [16]

    In Proceedings of the genetic and evolutionary computation conference companion

    Random subsampling improves performance in lexicase selection. In Proceedings of the genetic and evolutionary computation conference companion. 2028–2031

    work page 2028

  17. [17]

    Gregory S. Hornby. 2009. Steady-State ALPS for Real-valued Problems. InProceed- ings of the 11th Annual Conference on Genetic and Evolutionary Computation(New York, NY, USA, 2009)(GECCO ’09). ACM, 795–802. doi:10.1145/1569901.1570011

    work page doi:10.1145/1569901.1570011 2009

  18. [18]

    Gregory S Hornby. 2006. ALPS: the age-layered population structure for reducing the problem of premature convergence. InProceedings of the 8th annual conference on Genetic and evolutionary computation. 815–822

    work page 2006

  19. [19]

    Yaochu Jin. 2005. A comprehensive survey of fitness approximation in evolu- tionary computation.Soft computing9, 1 (2005), 3–12

    work page 2005

  20. [20]

    John R Koza. 1994. Genetic programming as a means for programming computers by natural selection.Statistics and computing4, 2 (1994), 87–112

    work page 1994

  21. [21]

    Alexander Lalejini, Marcos Sanson, Jack Garbus, Matthew Andres Moreno, and Emily Dolson. 2024. Runtime phylogenetic analysis enables extreme subsam- pling for test-based problems. InProceedings of the Genetic and Evolutionary Computation Conference Companion. 511–514

    work page 2024

  22. [22]

    Jason D Lohn, Gregory S Hornby, and Derek S Linden. 2005. An evolved antenna for deployment on nasa’s space technology 5 mission. InGenetic Programming Theory and Practice II. Springer, 301–315

    work page 2005

  23. [23]

    Raymond Luebbers. 2006. XFDTD and beyond-from classroom to corporation. In2006 IEEE Antennas and Propagation Society International Symposium. IEEE, 119–122

    work page 2006

  24. [24]

    Sean Luke. 2017. ECJ then and now. InProceedings of the genetic and evolutionary computation conference companion. 1223–1230

    work page 2017

  25. [25]

    Alex Machtay, J Rolla, and A Patton. 2022. Using genetic algorithms to optimize antenna designs for improved sensitivity to Ultra-High Energy neutrinos. In Proceedings of 38th International Cosmic Ray Conference—PoS (ICRC2023), Vol. 10. 9–16

    work page 2022

  26. [26]

    Michael O’Neill and Conor Ryan. 2002. Grammatical evolution.IEEE Transactions on Evolutionary Computation5, 4 (2002), 349–358

    work page 2002

  27. [27]

    2013.New optimization techniques in engi- neering

    Godfrey C Onwubolu and BV Babu. 2013.New optimization techniques in engi- neering. Vol. 141. Springer

    work page 2013

  28. [28]

    2011.Dynamic gas-surface interaction modeling for satel- lite aerodynamic computations

    Marcin D Pilinski. 2011.Dynamic gas-surface interaction modeling for satel- lite aerodynamic computations. Ph. D. Dissertation. University of Colorado at Boulder

    work page 2011

  29. [29]

    Julie Rolla, Bryan Reynolds, Jacob Weiler, Amy Connolly, Ryan Debolt, Alex Machtay, Ben Sipe, and Dylan Wells. 2023. Design of 3D Antenna Geometries Using Genetic Algorithms. (2023)

    work page 2023

  30. [30]

    Julie Rolla, Bryan Reynolds, Jacob Weiler, Dylan Wells, Max Foreback, Amy Connolly, Emily Dolson, and Charles Ofria. 2025. Designing Optimized Antennas for Science Applications Using Evolutionary Algorithms. (2025)

    work page 2025

  31. [31]

    2022.Optimising satellite geometries to minimise drag in Very Low Earth Orbits

    Luciana Sinpetru. 2022.Optimising satellite geometries to minimise drag in Very Low Earth Orbits. The University of Manchester (United Kingdom)

    work page 2022

  32. [32]

    Darrell Whitley. 1989. The GENITOR algorithm and selection pressure: Why rank-based allocation of reproductive trials is best. InProceedings of the third international conference on Genetic algorithms

    work page 1989