arxiv: 2604.13867 · v1 · submitted 2026-04-15 · 🌀 gr-qc · astro-ph.CO· hep-ph· hep-th

Recognition: unknown

Robust parameter inference for Taiji via time-frequency contrastive learning and normalizing flows

Tian-Yang Sun , Bo Liang , Ji-Yu Song , Song-Tao Liu , Shang-Jie Jin , He Wang , Ming-Hui Du , Jing-Fei Zhang

show 1 more author

Xin Zhang

Authors on Pith no claims yet

Pith reviewed 2026-05-10 12:40 UTC · model grok-4.3

classification 🌀 gr-qc astro-ph.COhep-phhep-th

keywords gravitational wavesparameter inferencenormalizing flowscontrastive learningglitchesTaijimassive black hole binariesamortized inference

0 comments

The pith

A neural framework with time-frequency contrastive learning and normalizing flows delivers more accurate parameter estimates for massive black hole binaries in glitch-contaminated Taiji data than Markov Chain Monte Carlo methods.

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

The authors introduce a method for inferring the parameters of massive black hole binary systems observed by the Taiji gravitational wave detector that remains effective even when the data includes transient noise glitches. They train conditional normalizing flows using a combination of time-frequency data processing and contrastive learning, supported by a neural network that creates realistic synthetic glitches for training purposes. Systematic tests reveal that this approach generates parameter posteriors that match the true values more closely and are better calibrated than those obtained from traditional sampling techniques like Markov Chain Monte Carlo. The work also highlights that common ways to check posterior quality fall short in such noisy conditions, recommending an additional score based on ranked probability for a fuller evaluation. If successful, this points toward faster, more reliable analysis of signals from space-based observatories.

Core claim

The paper establishes that an amortized inference framework built on conditional normalizing flows, equipped with a time-frequency multimodal fusion encoder and trained through contrastive learning on data augmented by a neural glitch generator, produces more accurate and better-calibrated posteriors for massive black hole binaries in the Taiji configuration under glitch contamination compared to conventional Markov Chain Monte Carlo sampling.

What carries the argument

The time-frequency contrastive learning setup combined with conditional normalizing flows, which enables amortized posterior estimation robust to synthetic glitch contamination in gravitational wave data.

If this is right

The proposed method yields more accurate posteriors than MCMC when glitches are present.
Ablation studies confirm that the full model with contrastive learning performs best and is robust to changes in glitch properties.
Continuous ranked probability score provides a stricter test of posterior quality than standard coverage checks alone.
This framework supports fast Bayesian estimation suitable for future space-based gravitational wave observations.

Where Pith is reading between the lines

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

The neural glitch generator could be adapted to model other types of non-stationary noise in gravitational wave detectors.
Similar amortized approaches might accelerate parameter estimation for the large event rates expected from next-generation detectors.
Training on simulated data with realistic noise could reduce reliance on expensive MCMC runs for initial analyses.
Extensions to other signal types such as extreme mass ratio inspirals may be feasible with the same architecture.

Load-bearing premise

The neural network that generates synthetic glitches must produce transients whose properties closely match those of real glitches in Taiji observations for the trained inference model to work on actual data.

What would settle it

Running the trained model on simulated Taiji data that includes real glitch waveforms extracted from existing detectors or from detailed Taiji noise simulations, then verifying whether the resulting posteriors are more accurate and have better calibration scores than MCMC results on the same data.

Figures

Figures reproduced from arXiv: 2604.13867 by Bo Liang, He Wang, Jing-Fei Zhang, Ji-Yu Song, Ming-Hui Du, Shang-Jie Jin, Song-Tao Liu, Tian-Yang Sun, Xin Zhang.

**Figure 1.** Figure 1: FIG. 1. Overall architecture of the proposed glitch-robust inference framework. Panel (A) shows the posterior inference [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Figure 2: FIG. 2. Fidelity of the neural glitch generator across the ex [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: FIG. 3. Representative multidimensional posterior distributions for a glitch-contaminated injection. The corner plot compares [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

**Figure 4.** Figure 4: FIG. 4. P-P plots for posterior calibration in glitch-free and [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

**Figure 7.** Figure 7: FIG. 7. Robustness of inference performance to glitch dura [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗

**Figure 8.** Figure 8: FIG. 8. Illustration of marginal-coverage diagnostics in a sim [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗

read the original abstract

Transient noise artifacts, commonly referred to as glitches, pose a major challenge to parameter inference for space-based gravitational-wave (GW) observations. We develop a glitch-robust amortized inference framework for massive black hole binaries in the Taiji detector configuration by combining conditional normalizing flows, a time-frequency multimodal fusion encoder, and contrastive learning. To enable large-scale training on contaminated data, we further introduce a neural glitch generator that produces high-fidelity synthetic transients at substantially reduced computational cost. Systematic experiments show that, under glitch contamination, the proposed method yields more accurate and better-calibrated posteriors than a conventional Markov Chain Monte Carlo baseline. In ablation studies, the full time-frequency model with contrastive learning performs best overall and remains robust to variations in glitch duration and merger-relative timing. We further show that standard coverage diagnostics alone are insufficient to fully assess posterior fidelity. We therefore complement them with the continuous ranked probability score, which provides a stricter assessment of global distributional agreement in non-ideal GW data. Taken together, these results establish deep-learning-based amortized inference as a promising framework for fast and robust Bayesian parameter estimation in future space-based GW observations.

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.

Desk Editor's Note private letter to a colleague

The paper offers a practical ML approach for glitch-robust inference in Taiji but its gains depend on unverified synthetic glitch realism.

read the letter

The main takeaway is that this work builds an amortized inference pipeline for massive black hole binaries in Taiji that stays accurate when glitches are present. It uses conditional normalizing flows, a time-frequency encoder with multimodal fusion, contrastive learning, and a neural network that generates synthetic glitches so the model can train on large contaminated datasets without heavy computation. On simulated data the method beats a standard MCMC baseline on accuracy and calibration, and the ablations show the full time-frequency plus contrastive setup works best across changes in glitch length and timing. They also add the continuous ranked probability score to the usual coverage checks, which gives a stricter look at whether the posteriors are well calibrated in non-ideal conditions. That combination is new for this detector and problem, and the experiments are systematic enough to be useful. The soft spot is the neural glitch generator. Its outputs are checked only against its own training metrics, with no quantitative match to real glitch statistics from comparable detectors or to a physical model. All training and testing stay inside the synthetic distribution, so the reported robustness could shrink if the generated transients miss key time-frequency features or timing patterns that appear in actual Taiji data. The paper is aimed at people building analysis pipelines for space-based gravitational-wave missions. Anyone working on amortized methods or non-stationary noise in GW data will get something concrete from it. It deserves peer review because the core idea addresses a real operational need and the experiments are laid out clearly, even if the glitch model will need tighter validation in revision.

Referee Report

2 major / 2 minor

Summary. The manuscript develops an amortized inference framework for massive black hole binary parameters in Taiji data using conditional normalizing flows, a time-frequency multimodal fusion encoder, and contrastive learning. A neural glitch generator is introduced to produce synthetic transients for large-scale training on contaminated waveforms. Systematic experiments on simulated data claim that the method produces more accurate and better-calibrated posteriors than MCMC under glitch contamination, with ablation studies showing the full model performs best and remains robust to glitch variations; CRPS is used alongside coverage diagnostics to assess distributional fidelity.

Significance. If the central assumptions hold, the work provides a practical route to fast, glitch-robust Bayesian parameter estimation for future space-based GW missions by amortizing expensive sampling and bypassing full physical glitch modeling. Strengths include the systematic ablation studies, use of CRPS for stricter global assessment beyond coverage, and direct benchmarking against MCMC on contaminated synthetic data. These elements address a key analysis challenge in Taiji-like observations.

major comments (2)

[Methods (neural glitch generator description)] The headline performance gains rest on the neural glitch generator producing transients whose statistical properties (time-frequency content, amplitude envelopes, duration, and merger-relative timing) match those expected in real Taiji data. No quantitative external validation against real or physics-based glitches is reported; all diagnostics remain internal to the synthetic training distribution, leaving open the possibility of distribution shift that could inflate the reported accuracy and calibration improvements over MCMC.
[Experiments and results] The MCMC baseline comparison is performed exclusively on data contaminated by the same neural generator. While this is internally consistent, the paper should demonstrate that the baseline MCMC is not disadvantaged by using a mismatched glitch model, and clarify whether the reported superiority persists when the MCMC is given access to the identical synthetic glitch realizations used in training.

minor comments (2)

Notation for the multimodal fusion encoder and contrastive loss could be clarified with explicit equations linking the time-frequency representations to the flow conditioning.
Figure captions should explicitly state the number of realizations and parameter ranges used in the ablation studies for reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their detailed and constructive report. The comments highlight important aspects of validation and experimental design that we address below. We have prepared revisions to improve clarity and acknowledge limitations where appropriate.

read point-by-point responses

Referee: The headline performance gains rest on the neural glitch generator producing transients whose statistical properties (time-frequency content, amplitude envelopes, duration, and merger-relative timing) match those expected in real Taiji data. No quantitative external validation against real or physics-based glitches is reported; all diagnostics remain internal to the synthetic training distribution, leaving open the possibility of distribution shift that could inflate the reported accuracy and calibration improvements over MCMC.

Authors: We agree that external validation against real or physics-based glitches would provide stronger evidence. Since Taiji has not yet collected flight data, all training and testing rely on simulated transients whose statistical properties were chosen to reflect current models of space-based detector noise. We performed extensive internal checks, including time-frequency spectrogram comparisons and moment matching, but these remain within the synthetic distribution. In the revised manuscript we will add an explicit limitations subsection discussing the assumptions of the neural glitch generator, the risk of distribution shift, and the implications for generalization to real data. We will also report additional quantitative diagnostics (e.g., Wasserstein distances on selected time-frequency features) that were computed during generator training. revision: partial
Referee: The MCMC baseline comparison is performed exclusively on data contaminated by the same neural generator. While this is internally consistent, the paper should demonstrate that the baseline MCMC is not disadvantaged by using a mismatched glitch model, and clarify whether the reported superiority persists when the MCMC is given access to the identical synthetic glitch realizations used in training.

Authors: The MCMC baseline uses the standard likelihood that does not include explicit glitch modeling, which mirrors the practical situation in which glitch parameters are unknown a priori. Supplying the MCMC sampler with the exact synthetic glitch realizations would amount to providing oracle noise information that is unavailable in real analyses; such a comparison would therefore not test the method under realistic conditions. Our amortized approach is trained to marginalize over unknown glitches without requiring their explicit parameterization. In the revision we will add a clarifying paragraph in the experimental section that states the comparison is performed under unmodeled glitch contamination and explains why an oracle-glitch MCMC experiment would not be representative. We will also note that, were perfect glitch knowledge available, both methods could in principle be improved, but our framework does not rely on it. revision: partial

Circularity Check

0 steps flagged

No significant circularity; empirical validation against independent MCMC baseline

full rationale

The paper's central claims rest on systematic experiments comparing the proposed time-frequency contrastive learning and normalizing flows framework against a conventional MCMC baseline on simulated Taiji data with synthetic glitches. No equations, derivations, or self-citations reduce the reported accuracy, calibration, or CRPS metrics to quantities defined by the model itself. The neural glitch generator is introduced as a practical tool for training data generation, but performance metrics are computed externally via comparison to MCMC and standard diagnostics, with no reduction by construction. The derivation chain for the amortized inference method is self-contained and does not rely on fitted inputs renamed as predictions or uniqueness theorems imported from the authors' prior work.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central performance claim depends on the fidelity of the invented neural glitch generator and on standard machine-learning assumptions about generalization from synthetic to real data; these are not independently verified beyond the reported experiments.

free parameters (1)

neural network architecture and training hyperparameters
Model depth, width, learning rates, and contrastive loss weights are selected to achieve the reported performance on contaminated data.

axioms (1)

domain assumption Synthetic glitches generated by the neural network have statistical properties sufficiently close to real Taiji glitches for the inference model to generalize.
Invoked to justify large-scale training on synthetic contaminated data.

invented entities (1)

neural glitch generator no independent evidence
purpose: Produce high-fidelity synthetic transients at reduced computational cost to enable large-scale training.
New component introduced to address the data scarcity problem for glitch-contaminated training.

pith-pipeline@v0.9.0 · 5534 in / 1482 out tokens · 45653 ms · 2026-05-10T12:40:57.815341+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

129 extracted references · 115 canonical work pages · 9 internal anchors

[1]

B. P. Abbottet al.(LIGO Scientific, Virgo), Phys. Rev. Lett.116, 221101 (2016), [Erratum: Phys.Rev.Lett. 121, 129902 (2018)], arXiv:1602.03841 [gr-qc]

work page Pith review arXiv 2016
[2]

B. P. Abbottet al.(LIGO Scientific, Virgo, 1M2H, Dark Energy Camera GW-E, DES, DLT40, Las Cumbres Observatory, VINROUGE, MASTER), Nature551, 85 (2017), arXiv:1710.05835 [astro-ph.CO]

work page Pith review arXiv 2017
[3]

Implications of the Neutron Star Merger GW170817 for Cosmological Scalar-Tensor Theories

J. Sakstein and B. Jain, Phys. Rev. Lett.119, 251303 (2017), arXiv:1710.05893 [astro-ph.CO]

work page Pith review arXiv 2017
[4]

B. P. Abbottet al.(LIGO Scientific, Virgo), Phys. Rev. Lett.123, 011102 (2019), arXiv:1811.00364 [gr-qc]

work page Pith review arXiv 2019
[5]

B. P. Abbottet al.(LIGO Scientific, Virgo), Phys. Rev. D100, 104036 (2019), arXiv:1903.04467 [gr-qc]

work page internal anchor Pith review arXiv 2019
[6]

Tests of General Relativity with Binary Black Holes from the second LIGO-Virgo Gravitational-Wave Transient Catalog

R. Abbottet al.(LIGO Scientific, Virgo), Phys. Rev. D 103, 122002 (2021), arXiv:2010.14529 [gr-qc]

work page internal anchor Pith review arXiv 2021
[7]

Abbottet al.(LIGO Scientific, Virgo,, KAGRA, VIRGO), Constraints on the Cosmic Expansion His- tory from GWTC–3, Astrophys

R. Abbottet al.(LIGO Scientific, Virgo, KAGRA), Astrophys. J.949, 76 (2023), arXiv:2111.03604 [astro- ph.CO]

work page arXiv 2023
[8]

Tests of General Relativity with GWTC-3

R. Abbottet al.(LIGO Scientific, VIRGO, KAGRA), (2021), arXiv:2112.06861 [gr-qc]

work page internal anchor Pith review arXiv 2021
[9]

A. G. Abacet al.(LIGO Scientific, VIRGO, KAGRA), (2025), arXiv:2509.04348 [astro-ph.CO]

work page arXiv 2025
[10]

B. F. Schutz, Nature323, 310 (1986)

1986
[11]

Tensions between the Early and the Late Universe

L. Verde, T. Treu, and A. G. Riess, Nature Astron.3, 891 (2019), arXiv:1907.10625 [astro-ph.CO]

work page internal anchor Pith review arXiv 2019
[12]

First measurement of the Hubble constant from a dark standard siren using the Dark Energy Survey galaxies and the LIGO/Virgo binary-black-hole merger GW170814

M. Soares-Santoset al.(DES, LIGO Scientific, Virgo), Astrophys. J. Lett.876, L7 (2019), arXiv:1901.01540 [astro-ph.CO]

work page Pith review arXiv 2019
[13]

A statistical standard siren measurement of the Hubble constant from the LIGO/Virgo gravitational wave compact object merger GW190814 and Dark Energy Survey galaxies,

A. Palmeseet al.(DES), Astrophys. J. Lett.900, L33 (2020), arXiv:2006.14961 [astro-ph.CO]

work page arXiv 2020
[14]

Palmese, C

A. Palmese, C. R. Bom, S. Mucesh, and W. G. Hartley, Astrophys. J.943, 56 (2023), arXiv:2111.06445 [astro- ph.CO]

work page arXiv 2023
[15]

C. Gong, T. Zhu, R. Niu, Q. Wu, J.-L. Cui, X. Zhang, W. Zhao, and A. Wang, Phys. Rev. D105, 044034 (2022), arXiv:2112.06446 [gr-qc]

work page arXiv 2022
[16]

C. Gong, T. Zhu, R. Niu, Q. Wu, J.-L. Cui, X. Zhang, W. Zhao, and A. Wang, Phys. Rev. D107, 124015 (2023), arXiv:2302.05077 [gr-qc]

work page arXiv 2023
[17]

Model-independent H 0 within FLRW: Joint Constraints from GWTC-3 Standard Sirens and Strong Lensing Time Delays,

J.-Y. Song, J.-Z. Qi, J.-F. Zhang, and X. Zhang, Astro- phys. J. Lett.985, L44 (2025), arXiv:2503.10346 [astro- 12 ph.CO]

work page arXiv 2025
[18]

Song, G.-H

J.-Y. Song, G.-H. Du, T.-N. Li, L.-F. Wang, J.-Z. Qi, J.-F. Zhang, and X. Zhang, Sci. China Phys. Mech. Astron.69, 240413 (2026), arXiv:2511.12017 [astro- ph.CO]

work page arXiv 2026
[19]

W. Zhao, C. Van Den Broeck, D. Baskaran, and T. G. F. Li, Phys. Rev. D83, 023005 (2011), arXiv:1009.0206 [astro-ph.CO]

work page arXiv 2011
[20]

Testing General Relativity with Present and Future Astrophysical Observations

E. Bertiet al., Class. Quant. Grav.32, 243001 (2015), arXiv:1501.07274 [gr-qc]

work page internal anchor Pith review arXiv 2015
[21]

Estimating cosmological parameters by the simulated data of gravitational waves from the Einstein Telescope

R.-G. Cai and T. Yang, Phys. Rev. D95, 044024 (2017), arXiv:1608.08008 [astro-ph.CO]

work page Pith review arXiv 2017
[22]

H.-Y. Chen, M. Fishbach, and D. E. Holz, Nature562, 545 (2018), arXiv:1712.06531 [astro-ph.CO]

work page Pith review arXiv 2018
[23]

Impacts of gravitational-wave standard siren observation of the Einstein Telescope on weighing neutrinos in cosmology

L.-F. Wang, X.-N. Zhang, J.-F. Zhang, and X. Zhang, Phys. Lett. B782, 87 (2018), arXiv:1802.04720 [astro- ph.CO]

work page Pith review arXiv 2018
[24]

Improving cosmological parameter estimation with the future gravitational-wave standard siren observation from the Einstein Telescope

X.-N. Zhang, L.-F. Wang, J.-F. Zhang, and X. Zhang, Phys. Rev. D99, 063510 (2019), arXiv:1804.08379 [astro-ph.CO]

work page Pith review arXiv 2019
[25]

Prospect for constraining holographic dark energy with gravitational wave standard sirens from the Einstein Telescope,

J.-F. Zhang, H.-Y. Dong, J.-Z. Qi, and X. Zhang, Eur. Phys. J. C80, 217 (2020), arXiv:1906.07504 [astro- ph.CO]

work page arXiv 2020
[26]

Zhang, M

J.-F. Zhang, M. Zhang, S.-J. Jin, J.-Z. Qi, and X. Zhang, JCAP09, 068 (2019), arXiv:1907.03238 [astro-ph.CO]

work page arXiv 2019
[27]

Li, D.-Z

H.-L. Li, D.-Z. He, J.-F. Zhang, and X. Zhang, JCAP 06, 038 (2020), arXiv:1908.03098 [astro-ph.CO]

work page arXiv 2020
[28]

Zhang, Sci

X. Zhang, Sci. China Phys. Mech. Astron.62, 110431 (2019), arXiv:1905.11122 [astro-ph.CO]

work page arXiv 2019
[29]

Maggiore, C

M. Maggioreet al.(ET), JCAP03, 050 (2020), arXiv:1912.02622 [astro-ph.CO]

work page arXiv 2020
[30]

Jin, D.-Z

S.-J. Jin, D.-Z. He, Y. Xu, J.-F. Zhang, and X. Zhang, JCAP03, 051 (2020), arXiv:2001.05393 [astro-ph.CO]

work page arXiv 2020
[31]

Song, L.-F

J.-Y. Song, L.-F. Wang, Y. Li, Z.-W. Zhao, J.-F. Zhang, W. Zhao, and X. Zhang, Sci. China Phys. Mech. Astron. 67, 230411 (2024), arXiv:2212.00531 [astro-ph.CO]

work page arXiv 2024
[32]

Han, J.-F

T. Han, J.-F. Zhang, and X. Zhang, Eur. Phys. J. C 86, 8 (2026), arXiv:2504.17741 [astro-ph.CO]

work page arXiv 2026
[33]

Jin, J.-Y

S.-J. Jin, J.-Y. Song, T.-Y. Sun, S.-R. Xiao, H. Wang, L.-F. Wang, J.-F. Zhang, and X. Zhang, Sci. China Phys. Mech. Astron.69, 220401 (2026), arXiv:2507.12965 [astro-ph.CO]

work page arXiv 2026
[34]

Synergy between CSST and third-generation gravitational-wave detectors: Inferring cosmological parameters using cross-correlation of dark sirens and galaxies

Y.-N. Du, J.-Y. Song, Y. Li, S.-J. Jin, L.-F. Wang, J.-F. Zhang, and X. Zhang, (2025), arXiv:2510.21521 [astro- ph.CO]

work page internal anchor Pith review Pith/arXiv arXiv 2025
[35]

Laser Interferometer Space Antenna

P. Amaro-Seoaneet al.(LISA), (2017), arXiv:1702.00786 [astro-ph.IM]

work page internal anchor Pith review Pith/arXiv arXiv 2017
[36]

TianQin: a space-borne gravitational wave detector

J. Luoet al.(TianQin), Class. Quant. Grav.33, 035010 (2016), arXiv:1512.02076 [astro-ph.IM]

work page Pith review arXiv 2016
[37]

Hu and Y.-L

W.-R. Hu and Y.-L. Wu, Natl. Sci. Rev.4, 685 (2017)

2017
[38]

Science with the space-based interferometer eLISA. I: Supermassive black hole binaries

A. Kleinet al., Phys. Rev. D93, 024003 (2016), arXiv:1511.05581 [gr-qc]

work page Pith review arXiv 2016
[39]

Wang, Z.-W

L.-F. Wang, Z.-W. Zhao, J.-F. Zhang, and X. Zhang, JCAP11, 012 (2020), arXiv:1907.01838 [astro-ph.CO]

work page arXiv 2020
[40]

Zhao, L.-F

Z.-W. Zhao, L.-F. Wang, J.-F. Zhang, and X. Zhang, Sci. Bull.65, 1340 (2020), arXiv:1912.11629 [astro- ph.CO]

work page arXiv 2020
[41]

Wang, S.-J

L.-F. Wang, S.-J. Jin, J.-F. Zhang, and X. Zhang, Sci. China Phys. Mech. Astron.65, 210411 (2022), arXiv:2101.11882 [gr-qc]

work page arXiv 2022
[42]

Jin, L.-F

S.-J. Jin, L.-F. Wang, P.-J. Wu, J.-F. Zhang, and X. Zhang, Phys. Rev. D104, 103507 (2021), arXiv:2106.01859 [astro-ph.CO]

work page arXiv 2021
[43]

Jin, S.-S

S.-J. Jin, S.-S. Xing, Y. Shao, J.-F. Zhang, and X. Zhang, Chin. Phys. C47, 065104 (2023), arXiv:2301.06722 [astro-ph.CO]

work page arXiv 2023
[44]

Jin, Y.-Z

S.-J. Jin, Y.-Z. Zhang, J.-Y. Song, J.-F. Zhang, and X. Zhang, Sci. China Phys. Mech. Astron.67, 220412 (2024), arXiv:2305.19714 [astro-ph.CO]

work page arXiv 2024
[45]

Dong, J.-Y

Y.-Y. Dong, J.-Y. Song, S.-J. Jin, J.-F. Zhang, and X. Zhang, JCAP05, 046 (2025), arXiv:2404.18188 [astro-ph.CO]

work page arXiv 2025
[46]

Dong, J.-Y

Y.-Y. Dong, J.-Y. Song, J.-F. Zhang, and X. Zhang, JCAP03, 046 (2026), arXiv:2507.10165 [gr-qc]

work page arXiv 2026
[47]

Song, Y.-Y

J.-Y. Song, Y.-Y. Dong, S.-J. Jin, S.-R. Xiao, J.-F. Zhang, and X. Zhang, (2026), arXiv:2603.13080 [astro- ph.CO]

work page arXiv 2026
[48]

Dong, J.-Y

Y.-Y. Dong, J.-Y. Song, J.-F. Zhang, and X. Zhang, (2026), arXiv:2603.15999 [astro-ph.CO]

work page arXiv 2026
[49]

Marsat, J

S. Marsat, J. G. Baker, and T. Dal Canton, Phys. Rev. D103, 083011 (2021), arXiv:2003.00357 [gr-qc]

work page arXiv 2021
[50]

Gao, Y.-M

J. Gao, Y.-M. Hu, E.-K. Li, J.-d. Zhang, and J. Mei, Phys. Rev. D111, 023039 (2025), arXiv:2401.12813 [astro-ph.GA]

work page arXiv 2025
[51]

M. Du, P. Wang, Z. Luo, W.-B. Han,et al., Sci. China Phys. Mech. Astron.69, 249501 (2026)

2026
[52]

Powell, D

J. Powell, D. Trifir` o, E. Cuoco, I. S. Heng, and M. Cavagli` a, Class. Quant. Grav.32, 215012 (2015), arXiv:1505.01299 [astro-ph.IM]

work page arXiv 2015
[53]

Zevin et al

M. Zevinet al., Class. Quant. Grav.34, 064003 (2017), arXiv:1611.04596 [gr-qc]

work page arXiv 2017
[54]

M. C. Edwards, P. Maturana-Russel, R. Meyer, J. Gair, N. Korsakova, and N. Christensen, Phys. Rev. D102, 084062 (2020), arXiv:2004.07515 [gr-qc]

work page arXiv 2020
[55]

Baghi, N

Q. Baghi, N. Korsakova, J. Slutsky, E. Castelli, N. Kar- nesis, and J.-B. Bayle, Phys. Rev. D105, 042002 (2022), arXiv:2112.07490 [gr-qc]

work page arXiv 2022
[56]

Armanoet al.(LISA Pathfinder), Phys

M. Armanoet al.(LISA Pathfinder), Phys. Rev. D106, 062001 (2022), arXiv:2205.11938 [astro-ph.IM]

work page arXiv 2022
[57]

Spadaro, R

A. Spadaro, R. Buscicchio, D. Vetrugno, A. Klein, S. Vi- tale, R. Dolesi, W. J. Weber, and M. Colpi, Phys. Rev. D108, 123029 (2023), arXiv:2306.03923 [gr-qc]

work page arXiv 2023
[58]

Houba, L

N. Houba, L. Ferraioli, and D. Giardini, Phys. Rev. D 109, 083027 (2024), arXiv:2401.00846 [astro-ph.IM]

work page arXiv 2024
[59]

Armanoet al.(LISA Pathfinder), In-depth analy- sis of LISA Pathfinder performance results: Time evo- lution, noise projection, physical models, and impli- cations for LISA, Phys

M. Armanoet al.(LISA Pathfinder), Phys. Rev. D110, 042004 (2024), arXiv:2405.05207 [astro-ph.IM]

work page arXiv 2024
[60]

J. R. Gair, S. Deng, and S. Babak, (2025), arXiv:2510.16770 [gr-qc]

work page arXiv 2025
[61]

Zevinet al., The European Physical Journal Plus139, 100 (2024), arXiv:2308.15530

M. Zevinet al., Eur. Phys. J. Plus139, 100 (2024), arXiv:2308.15530 [gr-qc]

work page arXiv 2024
[62]

Robson and N

T. Robson and N. J. Cornish, Phys. Rev. D99, 024019 (2019), arXiv:1811.04490 [gr-qc]

work page arXiv 2019
[63]

Castelli, Q

E. Castelli, Q. Baghi, J. G. Baker, J. Slutsky, J. Bobin, N. Karnesis, A. Petiteau, O. Sauter, P. Wass, and W. J. Weber, Class. Quant. Grav.42, 065018 (2025), arXiv:2411.13402 [gr-qc]

work page arXiv 2025
[64]

N. J. Cornish and T. B. Littenberg, Class. Quant. Grav. 32, 135012 (2015), arXiv:1410.3835 [gr-qc]

work page Pith review arXiv 2015
[65]

N. J. Cornish, T. B. Littenberg, B. B´ ecsy, K. Chatzi- ioannou, J. A. Clark, S. Ghonge, and M. Millhouse, Phys. Rev. D103, 044006 (2021), arXiv:2011.09494 [gr- qc]

work page arXiv 2021
[66]

Alvey, U

J. Alvey, U. Bhardwaj, V. Domcke, M. Pieroni, and C. Weniger, Phys. Rev. D111, 102006 (2025), arXiv:2408.00832 [gr-qc]

work page arXiv 2025
[67]

Muratore, J

M. Muratore, J. Gair, and L. Speri, Phys. Rev. D109, 042001 (2024), arXiv:2308.01056 [gr-qc]. 13

work page arXiv 2024
[68]

P. Wu, M. Du, and P. Xu, Opt. Express32, 43249 (2024), arXiv:2206.10664 [gr-qc]

work page arXiv 2024
[69]

Blelly, J

A. Blelly, J. Bobin, and H. Moutarde, Mon. Not. Roy. Astron. Soc.509, 5902 (2021), arXiv:2104.05250 [gr-qc]

work page arXiv 2021
[70]

Muratore, O

M. Muratore, O. Hartwig, D. Vetrugno, S. Vitale, and W. J. Weber, Phys. Rev. D107, 082004 (2023), arXiv:2207.02138 [gr-qc]

work page arXiv 2023
[71]

Sauter, P

O. Sauter, P. Wass, W. Sanchez, and H. Inchausp´ e, (2025), arXiv:2501.06315 [gr-qc]

work page arXiv 2025
[72]

Malz and J

A.-K. Malz and J. Veitch, Phys. Rev. D112, 024071 (2025), arXiv:2505.00657 [gr-qc]

work page arXiv 2025
[73]

Muratore, J

M. Muratore, J. Gair, O. Hartwig, M. L. Katz, and A. Toubiana, Phys. Rev. D112, 063041 (2025), arXiv:2505.19870 [gr-qc]

work page arXiv 2025
[74]

Wu, Y.-M

Z. Wu, Y.-M. Hu, H.-M. Fan, and I. S. Heng, Phys. Rev. D109, 103004 (2024), arXiv:2308.15354 [gr-qc]

work page arXiv 2024
[75]

Pankow et al

C. Pankowet al., Phys. Rev. D98, 084016 (2018), arXiv:1808.03619 [gr-qc]

work page arXiv 2018
[76]

Davis, T

D. Davis, T. B. Littenberg, I. M. Romero-Shaw, M. Mill- house, J. McIver, F. Di Renzo, and G. Ashton, Class. Quant. Grav.39, 245013 (2022), arXiv:2207.03429 [astro-ph.IM]

work page arXiv 2022
[77]

Baldi, P

P. Baldi, P. Sadowski, and D. Whiteson, Nat. Commun. 5, 4308 (2014), arXiv:1402.4735 [hep-ph]

work page arXiv 2014
[78]

Carleo, I

G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborov´ a, Rev. Mod. Phys.91, 045002 (2019), arXiv:1903.10563 [physics.comp-ph]

work page arXiv 2019
[79]

Cranmer, J

K. Cranmer, J. Brehmer, and G. Louppe, Proc. Nat. Acad. Sci.117, 30055 (2020), arXiv:1911.01429 [stat.ML]

work page arXiv 2020
[80]

G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, and L. Yang, Nat. Rev. Phys.3, 422 (2021)

2021

Showing first 80 references.