pith. sign in

arxiv: 2511.12642 · v2 · submitted 2025-11-16 · 🌀 gr-qc · astro-ph.IM· cs.LG

Auto-encoder model for faster generation of effective one-body gravitational waveform approximations

Suyog Garg , Feng-Li Lin , Kipp Cannon This is my paper

Pith reviewed 2026-05-17 22:02 UTC · model grok-4.3

classification 🌀 gr-qc astro-ph.IMcs.LG
keywords auto-encodergravitational waveformsSEOBNRv4waveform approximationmachine learningparameter estimationaligned spinsblack hole mergers
0
0 comments X p. Extension

The pith

An auto-encoder generates aligned-spin SEOBNRv4 gravitational waveforms at 50 microseconds each, four orders of magnitude faster than direct computation.

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

The paper trains an auto-encoder neural network to approximate inspiral-merger-ringdown waveforms from the SEOBNRv4 model for black hole binaries with aligned spins. The input parameters are the two component masses, limited to a mass ratio of 10, and the two dimensionless spins, each ranging up to 0.99 in magnitude. Once trained, the network produces a new waveform in roughly 50 microseconds on a GPU, so that a batch of one thousand waveforms finishes in about 0.1 seconds. This speed is intended to support the high-volume waveform calls required for parameter estimation when future detectors register many more events per year. The generated waveforms match the original model to a median mismatch of about 0.01, with noticeably better fidelity inside a restricted interval of effective spin.

Core claim

An auto-encoder model, built on the architecture of Liao & Lin, can reproduce aligned-spin SEOBNRv4 waveforms over the mass range 5-75 solar masses and spin range plus or minus 0.99 with a median mismatch of 10 to the minus 2, while delivering an average generation time of 50 microseconds per waveform on a GPU.

What carries the argument

The auto-encoder neural network that compresses each target waveform into a low-dimensional latent vector and reconstructs an approximation from samples drawn in that latent space.

If this is right

  • Likelihood evaluations in parameter estimation can be performed with far higher throughput for third-generation detector data.
  • Rapid sky localization becomes practical even when thousands of candidate waveforms must be evaluated for each event.
  • The model supplies a built-in uncertainty estimate through the spread of latent-space samples, reported as a 4 times 10 to the minus 3 standard deviation in mismatch.
  • Performance remains usable across the full sampled mass and spin range, although it improves inside the effective-spin window from minus 0.8 to 0.8.

Where Pith is reading between the lines

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

  • The same latent-space representation could be reused to interpolate waveforms at points not present in the original training grid.
  • Extending the training set to include precessing spins would test whether the speed gain survives when the waveform family grows more complex.
  • Embedding the auto-encoder directly inside existing analysis codes would quantify the net reduction in wall-clock time for full Bayesian inference runs.

Load-bearing premise

A median mismatch near 0.01 remains acceptable for the high-volume approximate tasks such as rapid sky localization, and accuracy does not drop sharply outside the tested effective-spin interval.

What would settle it

A side-by-side parameter-estimation run on the same set of simulated signals that shows the sky-localization regions or other recovered parameters differ by more than the statistical uncertainty when the auto-encoder waveforms replace the full SEOBNRv4 waveforms.

Figures

Figures reproduced from arXiv: 2511.12642 by Feng-Li Lin, Kipp Cannon, Suyog Garg.

Figure 1
Figure 1. Figure 1: FIG. 1. Representative plot of a gravitational waveform time [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Architecture of the 2 conditionals, 2 encoder, 1 [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Various loss function component values plotted [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Overplot of the reconstructed waveform approximation with the original polarization time series for a randomly selected [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Histogram of the calculated mismatch values for [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Contour plot of the mismatch values for [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10. Distribution of the mismatch values for 1000 [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
read the original abstract

Upgrades to current gravitational wave detectors for the next observation run and upcoming third-generation observatories, like the Einstein telescope, are expected to have enormous improvements in detection sensitivities and compact object merger event rates. Estimation of source parameters for a wider parameter space that these detectable signals will lie in, will be a computational challenge. Thus, it is imperative to have methods to speed-up the likelihood calculations with theoretical waveform predictions, which can ultimately make the parameter estimation faster and aid in rapid multi-messenger follow-ups. In this work we study auto-encoder models for gravitational waveform generation by adopting the best-performing architecture of Liao & Lin (2021) to approximate aligned-spin SEOBNRv4 inspiral-merger-ringdown waveforms. Our parameter space consists of four parameters, [$m_1$, $m_2$, $\chi_1(z)$, $\chi_2(z)$]. The masses are uniformly sampled in $[5,75]\,M_{\odot}$ with a mass ratio limit at $10\,M_{\odot}$, while the spins are uniform in $[-0.99,0.99]$. Our model is able to generate $10^3$ waveforms in $\sim 0.1$ second at an average speed of about 50 microsecond per waveform on a GPU. This is about 4 orders of magnitude faster than the native SEOBNRv4 implementation, and 2--3 orders of magnitude faster than existing non-machine-learning accelerated waveform variants. The median mismatch for the generated waveforms in the test dataset is $\sim10^{-2}$, with better performance in a restricted parameter space of $\chi_{\rm eff}\in[-0.80,0.80]$. The latent sampling error of our model can be quantified at a median mismatch standard deviation of $4\times10^{-3}$. Although the accuracy of our model does not enable full production-use yet, the model could be useful wherever high-volume of approximate theoretical waveforms are required, for instance, for rapid sky localization.

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 manuscript presents an auto-encoder neural network, adopting the architecture of Liao & Lin (2021), to approximate aligned-spin SEOBNRv4 inspiral-merger-ringdown waveforms over a four-dimensional parameter space (m1, m2, χ1(z), χ2(z)). Masses are sampled uniformly in [5, 75] M⊙ with mass-ratio limit 10; spins are uniform in [-0.99, 0.99]. The model is reported to generate 10^3 waveforms in ∼0.1 s (∼50 μs per waveform on GPU), a claimed four-order-of-magnitude speedup over native SEOBNRv4 and two-to-three orders over existing non-ML accelerated variants. Median mismatch on the test set is ∼10^{-2}, with improved performance inside χ_eff ∈ [-0.80, 0.80]; latent-sampling mismatch standard deviation is 4×10^{-3}. The authors suggest the model may be useful for high-volume approximate applications such as rapid sky localization.

Significance. If the reported mismatch level proves acceptable for downstream tasks, the work would supply a concrete, GPU-accelerated surrogate that could materially reduce the computational cost of high-volume waveform generation in parameter estimation and multi-messenger follow-up for current and third-generation detectors. Credit is given for the explicit adoption of a previously validated architecture and for the provision of concrete numerical benchmarks (speed, median mismatch, latent error).

major comments (2)
  1. [Abstract] Abstract: the claim that the model “could be useful” for rapid sky localization is not supported by any quantitative test (injected-signal recovery, sky-area comparison, or bias in localization volume). The median mismatch ∼10^{-2} is presented without demonstrating whether it systematically broadens posteriors or shifts peaks beyond tolerances acceptable for the intended high-volume uses.
  2. [Abstract and §3] Abstract and §3 (training/validation description): concrete performance numbers (speed, median mismatch, latent sampling error) are stated without accompanying training details, validation protocol, or error-bar analysis, leaving the central empirical claim only moderately supported.
minor comments (2)
  1. [Abstract] Clarify notation: the symbols χ1(z), χ2(z) should be defined explicitly (z-component spins versus effective spin) at first use.
  2. Add a figure or table showing mismatch versus χ_eff (or other parameters) to illustrate the reported improvement inside the restricted interval.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and outline the revisions we will make to strengthen the presentation of our results.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that the model “could be useful” for rapid sky localization is not supported by any quantitative test (injected-signal recovery, sky-area comparison, or bias in localization volume). The median mismatch ∼10^{-2} is presented without demonstrating whether it systematically broadens posteriors or shifts peaks beyond tolerances acceptable for the intended high-volume uses.

    Authors: We agree that the abstract statement regarding potential utility for rapid sky localization is not supported by direct quantitative tests such as injected-signal recovery or localization-volume comparisons. The reported median mismatch of ∼10^{-2} is a waveform-level metric only, and we did not perform parameter-estimation studies to quantify any systematic effects on posteriors. In the revised manuscript we will remove the specific example of rapid sky localization from the abstract and replace it with a more general statement that the model may be suitable for high-volume approximate applications where the reported accuracy is acceptable. We will also add a short paragraph in the conclusions section acknowledging this limitation and noting that downstream validation would be required before production use in localization tasks. revision: yes

  2. Referee: [Abstract and §3] Abstract and §3 (training/validation description): concrete performance numbers (speed, median mismatch, latent sampling error) are stated without accompanying training details, validation protocol, or error-bar analysis, leaving the central empirical claim only moderately supported.

    Authors: We acknowledge that the current description of the training and validation protocol is brief and does not include sufficient detail to fully support the reported performance numbers. The manuscript follows the architecture validated by Liao & Lin (2021) and uses a standard train/validation/test split on the generated SEOBNRv4 dataset, with the test-set mismatch computed after training. To address the referee’s concern we will expand §3 with explicit information on the dataset split ratios, optimizer, loss function, training hyperparameters, number of epochs, early-stopping criteria, and any regularization techniques employed. We will also report the standard deviation of the mismatch across the test set (already computed as 4×10^{-3} for latent sampling) and clarify how the median mismatch was obtained, thereby providing the requested error analysis and validation protocol. revision: yes

Circularity Check

1 steps flagged

Minor self-citation for architecture choice; central results are independent empirical measurements

specific steps
  1. self citation load bearing [Abstract]
    "by adopting the best-performing architecture of Liao & Lin (2021) to approximate aligned-spin SEOBNRv4 inspiral-merger-ringdown waveforms"

    The architecture is selected based on prior work by overlapping authors (Feng-Li Lin), but this choice does not force the reported speedup or mismatch values, which are measured on new training and test data.

full rationale

The paper trains an auto-encoder directly on SEOBNRv4 waveform data and reports measured generation speeds and reconstruction mismatches as empirical outcomes. No derivation or first-principles claim reduces by construction to fitted inputs or self-citations. The sole self-citation (adopting an architecture from Liao & Lin 2021, with author overlap) is minor and non-load-bearing, as the training, testing, and performance metrics stand independently. This yields a low circularity score consistent with normal non-circular ML approximation papers.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The claim rests on the empirical success of an auto-encoder in learning a latent representation of SEOBNRv4 waveforms; the main unstated premise is that the training distribution adequately covers the target use cases.

free parameters (1)
  • auto-encoder architecture and training hyperparameters
    Adopted from Liao & Lin (2021) but implicitly tuned for the present waveform data and parameter space.
axioms (1)
  • domain assumption An auto-encoder can learn a compact latent representation that allows faithful reconstruction of SEOBNRv4 waveforms across the sampled mass and spin range.
    Invoked by the choice of model and the reported mismatch values.

pith-pipeline@v0.9.0 · 5675 in / 1252 out tokens · 74429 ms · 2026-05-17T22:02:35.992736+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

78 extracted references · 78 canonical work pages · 13 internal anchors

  1. [1]

    Liao and F.-L

    C.-H. Liao and F.-L. Lin, Physical Review D103, 124051 (2021)

    work page 2021

  2. [2]

    LIGO Scientific Collaboration and Virgo Collaboration, Physical Review Letters116, 061102 (2016)

    work page 2016

  3. [3]

    LIGO Scientific Collaboration and Virgo Collaboration, Physical Review X9, 031040 (2019)

    work page 2019

  4. [4]

    LIGO Scientific Collaboration and Virgo Collaboration, Physical Review X11, 021053 (2021)

    work page 2021

  5. [5]

    T. L. S. Collaboration, Physical Review X13, 041039 (2023), arXiv:2111.03606 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  6. [6]

    The LIGO Scientific Collaboration, GWTC-4.0: Updating the Gravitational-Wave Transient Catalog with Observations from the First Part of the Fourth LIGO-Virgo-KAGRA Observing Run (2025)

    work page 2025

  7. [7]

    Balasubramanian, B

    R. Balasubramanian, B. S. Sathyaprakash, and S. V. Dhurandhar, Phys. Rev. D53, 3033 (1996)

    work page 1996

  8. [8]

    Thrane and C

    E. Thrane and C. Talbot, Publications of the Astronomical Society of Australia36, e010 (2019)

    work page 2019

  9. [9]

    Hannam, S

    M. Hannam, S. Husa, F. Ohme, and P. Ajith, Physical Review D82, 124052 (2010)

    work page 2010

  10. [10]

    T. L. S. Collaboration, Physical Review Letters119, 161101 (2017), arXiv:1710.05832 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  11. [11]

    Della Monica and R

    R. Della Monica and R. Brito, Physical Review D112, 024074 (2025)

    work page 2025

  12. [12]

    A. Abac, R. Abramo, S. Albanesi, A. Albertini, A. Agapito, M. Agathos, C. Albertus, N. Andersson, T. Andrade, I. Andreoni, F. Angeloni, M. Antonelli, J. Antoniadis, F. Antonini, M. A. Sedda, M. C. Artale, S. Ascenzi, P. Auclair, M. Bachetti, C. Badger, B. Banerjee, D. Barba-Gonzalez, D. Barta, N. Bartolo, A. Bauswein, A. Begnoni, F. Beirnaert, M. Bejger, ...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  13. [13]

    E. D. Hall, K. Kuns, J. R. Smith, Y. Bai, C. Wipf, S. Biscans, R. X. Adhikari, K. Arai, S. Ballmer, L. Barsotti, Y. Chen, M. Evans, P. Fritschel, J. Harms, B. Kamai, J. G. Rollins, D. Shoemaker, B. J. J. Slagmolen, R. Weiss, and H. Yamamoto, Physical Review D103, 122004 (2021)

    work page 2021

  14. [14]

    LISA Definition Study Report

    M. Colpi, K. Danzmann, M. Hewitson, K. Holley- Bockelmann, P. Jetzer, G. Nelemans, A. Petiteau, D. Shoemaker, C. Sopuerta, R. Stebbins, N. Tanvir, H. Ward, W. J. Weber, I. Thorpe, A. Daurskikh, A. Deep, I. F. N´ u˜ nez, C. G. Marirrodriga, M. Gehler, J.-P. Halain, O. Jennrich, U. Lammers, J. Larra˜ naga, M. Lieser, N. L¨ utzgendorf, W. Martens, L. Mondin,...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  15. [15]

    Z. Xuan, S. Naoz, and X. Chen, Physical Review D107, 043009 (2023), arXiv:2210.03129 [astro-ph]

    work page arXiv 2023

  16. [16]

    Gupta et al.,Characterizing gravitational wave detector networks: FromA ♯ to Cosmic Explorer,Classical and Quantum Gravity41(2024) 245001 [2307.10421]

    I. Gupta, C. Afle, K. G. Arun, A. Bandopadhyay, M. Baryakhtar, S. Biscoveanu, S. Borhanian, F. Broekgaarden, A. Corsi, A. Dhani, M. Evans, E. D. Hall, O. A. Hannuksela, K. Kacanja, R. Kashyap, S. Khadkikar, K. Kuns, T. G. F. Li, A. L. Miller, A. H. Nitz, B. J. Owen, C. Palomba, A. Pearce, H. Phurailatpam, B. Rajbhandari, J. Read, J. D. Romano, B. S. Sathy...

    work page arXiv 2024

  17. [17]

    Collaboration, Decadal upgrade strategy for KAGRA toward post-O5 gravitational-wave astronomy (2025), arXiv:2508.03392 [gr-qc]

    KAGRA. Collaboration, Decadal upgrade strategy for KAGRA toward post-O5 gravitational-wave astronomy (2025), arXiv:2508.03392 [gr-qc]

    work page arXiv 2025

  18. [18]

    M. A. Scheel, M. Boyle, K. Mitman, N. Deppe, L. C. Stein, C. Armaza, M. S. Bonilla, L. T. Buchman, A. Ceja, H. Chaudhary, Y. Chen, M. Corman, K. Z. Csuk´ as, C. M. Ferrus, S. E. Field, M. Giesler, S. Habib, F. H´ ebert, D. A. Hemberger, D. A. B. Iozzo, T. Islam, K. Z. Jones, A. Khairnar, L. E. Kidder, T. Knapp, P. Kumar, G. Lara, O. Long, G. Lovelace, S. ...

    work page arXiv 2025

  19. [19]

    Messina and A

    F. Messina and A. Nagar, Physical Review D95, 124001 (2017)

    work page 2017

  20. [20]

    S. Husa, S. Khan, M. Hannam, M. P¨ urrer, F. Ohme, X. J. Forteza, and A. Boh´ e, Physical Review D93, 044006 (2016)

    work page 2016

  21. [21]

    S. Khan, S. Husa, M. Hannam, F. Ohme, M. P¨ urrer, X. J. Forteza, and A. Boh´ e, Physical Review D93, 044007 (2016), arXiv:1508.07253 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  22. [22]

    Dietrich, A

    T. Dietrich, A. Samajdar, S. Khan, N. K. Johnson- McDaniel, R. Dudi, and W. Tichy, Physical Review D 100, 044003 (2019), arXiv:1905.06011 [gr-qc]

    work page arXiv 2019

  23. [23]

    Pratten, S

    G. Pratten, S. Husa, C. Garc´ ıa-Quir´ os, M. Colleoni, A. Ramos-Buades, H. Estell´ es, and R. Jaume, Physical Review D102, 064001 (2020)

    work page 2020

  24. [24]

    A. Abac, T. Dietrich, A. Buonanno, J. Steinhoff, and M. Ujevic, Physical Review D109, 024062 (2024), arXiv:2311.07456 [gr-qc]

    work page arXiv 2024

  25. [25]

    Buonanno and T

    A. Buonanno and T. Damour, Physical Review D59, 084006 (1999)

    work page 1999

  26. [26]

    The Effective One Body description of the Two-Body problem

    T. Damour and A. Nagar, The Effective One Body description of the Two-Body problem (2009), arXiv:0906.1769 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2009

  27. [27]

    Time-domain effective-one-body gravitational waveforms for coalescing compact binaries with nonprecessing spins, tides and self-spin effects

    A. Nagar, S. Bernuzzi, W. D. Pozzo, G. Riemenschneider, S. Akcay, G. Carullo, P. Fleig, S. Babak, K. W. Tsang, M. Colleoni, F. Messina, G. Pratten, D. Radice, P. Rettegno, M. Agathos, E. Fauchon-Jones, M. Hannam, S. Husa, T. Dietrich, P. Cerda-Duran, J. A. Font, F. Pannarale, P. Schmidt, and T. Damour, Physical Review D98, 104052 (2018), arXiv:1806.01772 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  28. [28]

    Frequency-domain gravitational waveform models for inspiraling binary neutron stars

    K. Kawaguchi, K. Kiuchi, K. Kyutoku, Y. Sekiguchi, M. Shibata, and K. Taniguchi, Frequency-domain gravitational waveform models for inspiraling binary neutron stars (2018), arXiv:1802.06518 [gr-qc]

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  29. [29]

    D. P. Mihaylov, S. Ossokine, A. Buonanno, H. Estelles, L. Pompili, M. P¨ urrer, and A. Ramos-Buades, pySEOBNR: A software package for the next generation of effective-one-body multipolar waveform models (2023), arXiv:2303.18203 [gr-qc]

    work page arXiv 2023

  30. [30]

    van de Meent, A

    M. van de Meent, A. Buonanno, D. P. Mihaylov, S. Ossokine, L. Pompili, N. Warburton, A. Pound, B. Wardell, L. Durkan, and J. Miller, Enhancing the SEOBNRv5 effective-one-body waveform model with second-order gravitational self-force fluxes (2023), arXiv:2303.18026 [gr-qc]

    work page arXiv 2023

  31. [31]

    Albanesi, R

    S. Albanesi, R. Gamba, S. Bernuzzi, J. Fontbut´ e, A. Gonzalez, and A. Nagar, Effective-one-body modeling for generic compact binaries with arbitrary orbits (2025), arXiv:2503.14580 [gr-qc]

    work page arXiv 2025

  32. [32]

    A. J. K. Chua, C. R. Galley, and M. Vallisneri, Physical 11 Review Letters122, 211101 (2019)

    work page 2019

  33. [33]

    Blackman, S

    J. Blackman, S. E. Field, M. A. Scheel, C. R. Galley, D. A. Hemberger, P. Schmidt, and R. Smith, Physical Review D95, 104023 (2017)

    work page 2017

  34. [34]

    Blackman, S

    J. Blackman, S. E. Field, M. A. Scheel, C. R. Galley, C. D. Ott, M. Boyle, L. E. Kidder, H. P. Pfeiffer, and B. Szil´ agyi, Physical Review D96, 024058 (2017)

    work page 2017

  35. [35]

    Pathak, A

    L. Pathak, A. Reza, and A. S. Sengupta, Physical Review D110, 064022 (2024)

    work page 2024

  36. [36]

    S. E. Field, V. Varma, J. Blackman, B. Gadre, C. R. Galley, T. Islam, K. Mitman, M. P¨ urrer, A. Ravichandran, M. A. Scheel, L. C. Stein, and J. Yoo, Journal of Open Source Software10, 7073 (2025), arXiv:2504.08839 [astro-ph]

    work page arXiv 2025

  37. [37]

    P. J. Nee, A. Ravichandran, S. E. Field, T. Islam, H. P. Pfeiffer, V. Varma, M. Boyle, A. Ceja, N. Ghadiri, L. E. Kidder, P. Kumar, A. Maurya, M. Morales, A. Ramos- Buades, A. Ravishankar, K. Rink, H. R. R¨ uter, M. A. Scheel, M. A. Shaikh, and D. Tellez, Eccentric binary black holes: A new framework for numerical relativity waveform surrogates (2025), ar...

    work page arXiv 2025

  38. [38]

    M. Dax, S. R. Green, J. Gair, J. H. Macke, A. Buonanno, and B. Sch¨ olkopf, Physical Review Letters127, 241103 (2021), arXiv:2106.12594 [gr-qc]

    work page arXiv 2021

  39. [39]

    Dax, Physical Review Letters130, 10.1103/PhysRevLett.130.171403 (2023)

    M. Dax, Physical Review Letters130, 10.1103/PhysRevLett.130.171403 (2023)

    work page doi:10.1103/physrevlett.130.171403 2023

  40. [40]

    H. Xia, L. Shao, J. Zhao, and Z. Cao, Physical Review D103, 024040 (2021), arXiv:2011.04418 [astro-ph]

    work page arXiv 2021

  41. [41]

    Gabbard, C

    H. Gabbard, C. Messenger, I. S. Heng, F. Tonolini, and R. Murray-Smith, Nature Physics18, 112 (2022), arXiv:1909.06296 [astro-ph]

    work page arXiv 2022

  42. [42]

    C. Ma, S. Li, Z. Cao, and M. Jia, Physical Review D110, 063010 (2024)

    work page 2024

  43. [43]

    Cuoco, M

    E. Cuoco, M. Cavagli` a, I. S. Heng, D. Keitel, and C. Messenger, Living Reviews in Relativity28, 2 (2025)

    work page 2025

  44. [44]

    Adhikari, and Erik Katsavounidis, Phys

    R. Ormiston, T. Nguyen, M. Coughlin, R. X. Adhikari, and E. Katsavounidis, Phys. Rev. Res.2, 033066 (2020), arXiv:2005.06534 [astro-ph.IM]

    work page arXiv 2020

  45. [45]

    Reissel, S

    C. Reissel, S. Soni, M. Saleem, M. Coughlin, P. Harris, and E. Katsavounidis, Coherence DeepClean: Toward autonomous denoising of gravitational-wave detector data (2025), arXiv:2501.04883 [gr-qc]

    work page arXiv 2025

  46. [46]

    Oshino, Y

    S. Oshino, Y. Sakai, M. Meyer-Conde, T. Uchiyama, Y. Itoh, Y. Shikano, Y. Terada, and H. Takahashi, Glitch noise classification in KAGRA O3GK observing data using unsupervised machine learning (2025), arXiv:2510.14291 [gr-qc]

    work page arXiv 2025

  47. [47]

    McLeod, D

    A. McLeod, D. Jacobs, C. Chatterjee, L. Wen, and F. Panther, Rapid Mass Parameter Estimation of Binary Black Hole Coalescences Using Deep Learning (2022), arXiv:2201.11126 [astro-ph]

    work page arXiv 2022

  48. [48]

    E. Marx, W. Benoit, A. Gunny, R. Omer, D. Chatterjee, R. C. Venterea, L. Wills, M. Saleem, E. Moreno, R. Raikman, E. Govorkova, D. Rankin, M. W. Coughlin, P. Harris, and E. Katsavounidis, A machine- learning pipeline for real-time detection of gravitational waves from compact binary coalescences (2024), arXiv:2403.18661 [gr-qc]

    work page arXiv 2024

  49. [49]

    J. Lee, S. H. Oh, K. Kim, G. Cho, J. J. Oh, E. J. Son, and H. M. Lee, Physical Review D103, 11 (2021), arXiv:2101.05685 [astro-ph]

    work page arXiv 2021

  50. [50]

    Khan and R

    S. Khan and R. Green, Physical Review D103, 064015 (2021)

    work page 2021

  51. [51]

    A. J. K. Chua, M. L. Katz, N. Warburton, and S. A. Hughes, Physical Review Letters126, 051102 (2021)

    work page 2021

  52. [52]

    Schmidt, M

    S. Schmidt, M. Breschi, R. Gamba, G. Pagano, P. Rettegno, G. Riemenschneider, S. Bernuzzi, A. Nagar, and W. Del Pozzo, Physical Review D103, 043020 (2021)

    work page 2021

  53. [53]

    L. M. Thomas, G. Pratten, and P. Schmidt, Physical Review D106, 104029 (2022)

    work page 2022

  54. [54]

    R. Shi, Y. Zhou, T. Zhao, Z. Ren, and Z. Cao, arXiv e-prints 10.48550/arXiv.2411.14893 (2024)

    work page doi:10.48550/arxiv.2411.14893 2024

  55. [55]

    Grimbergen, S

    T. Grimbergen, S. Schmidt, C. Kalaghatgi, and C. van den Broeck, Physical Review D109, 104065 (2024)

    work page 2024

  56. [56]

    M. Sun, J. Wu, J. Li, B. Mccane, N. Yang, X. Ma, B. Wang, and M. Zhang, Conditional Autoencoder for Generating BNS Waveforms with Tidal and Precession Effects (2025), arXiv:2503.19512 [astro-ph]

    work page arXiv 2025

  57. [57]

    Garg, inProceedings of 39th International Cosmic Ray Conference — PoS(ICRC2025), Vol

    S. Garg, inProceedings of 39th International Cosmic Ray Conference — PoS(ICRC2025), Vol. 501 (SISSA Medialab, 2025) p. 927

    work page 2025

  58. [58]

    O. G. Freitas, A. Theodoropoulos, N. Villanueva, T. Fernandes, S. Nunes, J. A. Font, A. Onofre, A. Torres- Forn´ e, and J. D. Martin-Guerrero, Physical Review D 112, 043026 (2025), arXiv:2412.06946 [gr-qc]

    work page arXiv 2025

  59. [59]

    Tissino, G

    J. Tissino, G. Carullo, M. Breschi, R. Gamba, S. Schmidt, and S. Bernuzzi, Physical Review D107, 084037 (2023)

    work page 2023

  60. [60]

    Gadre, M

    B. Gadre, M. P¨ urrer, S. E. Field, S. Ossokine, and V. Varma, Physical Review D110, 124038 (2024)

    work page 2024

  61. [61]

    Cuoco, M

    E. Cuoco, M. Cavagli` a, I. S. Heng, D. Keitel, and C. Messenger, arXiv e-prints 10.48550/arXiv.2412.15046 (2024)

    work page doi:10.48550/arxiv.2412.15046 2024

  62. [62]

    LIGO Scientific Collaboration, LIGO Algorithm Library, LIGO Scientific Collaboration (2018)

    work page 2018

  63. [63]

    A. Nitz, I. Harry, D. Brown, C. M. Biwer, J. Willis, T. D. Canton, C. Capano, T. Dent, L. Pekowsky, G. S. C. Davies, S. De, M. Cabero, S. Wu, A. R. Williamson, B. Machenschalk, D. Macleod, F. Pannarale, P. Kumar, S. Reyes, dfinstad, S. Kumar, M. T´ apai, L. Singer, P. Kumar, veronica-villa, maxtrevor, B. U. V. Gadre, S. Khan, S. Fairhurst, and A. Tolley, ...

    work page 2024

  64. [64]

    Schmidt,Searching for Precessing Black Hole Binaries in Gravitational-wave Data, Doctoral thesis 1 (Research UU / Graduation UU), Utrecht University, Utrecht (2025)

    S. Schmidt,Searching for Precessing Black Hole Binaries in Gravitational-wave Data, Doctoral thesis 1 (Research UU / Graduation UU), Utrecht University, Utrecht (2025)

    work page 2025

  65. [65]

    N. T. Nam and P. D. Hung, inProceedings of the 3rd International Conference on Machine Learning and Soft Computing, ICMLSC ’19 (Association for Computing Machinery, New York, NY, USA, 2019) pp. 138–142

    work page 2019

  66. [66]

    M. A. Islam, M. Kowal, S. Jia, K. G. Derpanis, and N. D. B. Bruce, Position, Padding and Predictions: A Deeper Look at Position Information in CNNs (2021), arXiv:2101.12322 [cs]

    work page arXiv 2021

  67. [67]

    Gravitational Waves from Mergin Compact Binaries: How Accurately Can One Extract the Binary's Parameters from the Inspiral Waveform?

    C. Cutler and E. Flanagan, Physical Review D49, 2658 (1994), arXiv:gr-qc/9402014

    work page internal anchor Pith review Pith/arXiv arXiv 1994

  68. [68]

    B. S. Sathyaprakash and B. F. Schutz, Living Reviews in Relativity12, 2 (2009)

    work page 2009

  69. [69]

    D. P. Kingma and M. Welling, Auto-Encoding Variational Bayes (2022), arXiv:1312.6114 [stat]

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  70. [70]

    Note that, degeneracies exist in the inverse problem of estimating parameters posteriors given some known waveform, in the sense that identical waveforms can belong to different sets of source parameters. An example for this is the case of waveforms from highly precessing binary systems that can have orbital timescale frequency 12 modulations that are sim...

    work page

  71. [71]

    D. P. Kingma and M. Welling, Foundations and Trends® in Machine Learning12, 307 (2019), arXiv:1906.02691 [cs]

    work page arXiv 2019

  72. [72]

    Doersch, (2016), arXiv:1606.05908 [stat.ML]

    C. Doersch, Tutorial on Variational Autoencoders (2021), arXiv:1606.05908 [stat]

    work page arXiv 2021

  73. [73]

    PyTorch: An Imperative Style, High-Performance Deep Learning Library

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. K¨ opf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala, PyTorch: An Imperative Style, High-Performance Deep Learning Library (2019), arXiv:1912.01703 [cs]

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  74. [74]

    Nickolls, I

    J. Nickolls, I. Buck, M. Garland, and K. Skadron, Queue 6, 40 (2008)

    work page 2008

  75. [75]

    D. P. Kingma and J. Ba, Adam: A Method for Stochastic Optimization (2017), arXiv:1412.6980 [cs]

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  76. [76]

    S. Garg, S. Sasaoka, D. Dominguez, Y. Hou, N. Koyama, K. Somiya, H. Takahashi, and M. Ohashi, inProceedings of 38th International Cosmic Ray Conference—PoS (ICRC2023)(Sissa Medialab, Nagoya, Japan, 2023) (2023)

    work page 2023

  77. [77]

    T. L. S. Collaboration, GWTC-4.0: Population Properties of Merging Compact Binaries (2025), arXiv:2508.18083 [astro-ph]

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  78. [78]

    Wette, SoftwareX12, 100634 (2020), arXiv:2012.09552 [astro-ph]

    K. Wette, SoftwareX12, 100634 (2020), arXiv:2012.09552 [astro-ph]. Appendix: 2C2E1D CVAE model We use a 2-conditional, 2-encoder, 1-decoder model for generating gravitational waveform approximations corresponding to a specific set of parameters. Our task can be formally stated as follows. Given some inputsx= [A(t), f(t)] t∈{1...T} consisting of normalized...

    work page arXiv 2020