Pith. sign in

REVIEW 4 major objections 5 minor 100 references

A single reversible-jump MCMC run can compare gravitational-wave source models and return odds ratios, matching standard nested sampling.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 07:17 UTC pith:OLEY7CBH

load-bearing objection A genuinely new RJMCMC sampler for CBC model selection, but the detailed-balance proof has a real gap and the validation contradicts itself on the low-SNR run. the 4 major comments →

arxiv 2607.21484 v1 pith:OLEY7CBH submitted 2026-07-23 gr-qc astro-ph.HEastro-ph.IM

A story about a tipsy kangaroo: Reversible jump MCMC for model selection in the analysis of gravitational-wave signals from the coalescence of compact objects

Anna Puecher , Tim Dietrich , Hauke Koehn , Jonathan Gair , Gregory Ashton , Luca Negri , Anuradha Samajdar This is my paper
classification gr-qc astro-ph.HEastro-ph.IM
keywords reversible jump MCMCmodel selectiongravitational wavescompact binary coalescenceneutron star classificationBayesian evidencesource identificationtransdimensional inference
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper presents t-roo, a sampler that treats the waveform model itself as a parameter to be jumped between, so one analysis returns both the odds ratio between competing source descriptions and the parameter posteriors for the favored ones. It targets the question that often decides gravitational-wave source classification: was the compact object a black hole, a neutron star, or a neutron star–black hole pair? The authors validate the sampler on simulated signals and on two real ambiguous events, finding that the model it prefers matches the ranking from a standard nested-sampling analysis. If the method holds up, it makes model comparison cheaper as data become more informative, because the sampler automatically spends most iterations in the model the data favor.

Core claim

The central claim is that transdimensional inference for compact-binary gravitational-wave signals is practical with a carefully designed reversible jump sampler. The sampler jumps between models of different parameter-space dimension, using pseudo-parameters to cache tidal deformability information between jumps, data-informed linear maps for chirp mass and mass ratio tied to tidal deformability, and auxiliary-variable constructions for which the proposal-density ratio and Jacobian cancel exactly, so the acceptance ratio reduces to likelihood and prior ratios. In every injection test the sampler's preferred model matches the reference nested sampler, and on the two ambiguous real events it

What carries the argument

The between-model move: a reversible jump between models with different numbers of tidal parameters, using storage tidal parameters in models that do not use them, a stretch move with tiny jumps for the tidal variables, and slopes learned from a short preliminary sampling phase to map chirp mass and mass ratio between models. The identity that carries the argument is Eq. (19): the ratio of auxiliary-variable densities times the Jacobian equals one, so detailed balance survives without extra factors.

Load-bearing premise

Everything rests on the exact cancellation in Eq. (19) — that the auxiliary-variable density ratio times the Jacobian equals one for every proposed jump, and on the learned mass/tidal mapping being invertible; if that fails, the sampler's stationary distribution is no longer the intended posterior and all model probabilities are biased.

What would settle it

Take a simulated signal with a known model and known evidence, run t-roo alongside a reliable evidence computation, and check whether the odds ratio from sample fractions matches the evidence ratio within sampling error — especially for signals where the learned slopes are steep or the mass/tidal support is widely separated. Alternatively, evaluate Eq. (19) numerically over the full prior range of the auxiliary variables and slopes; any deviation from 1 falsifies the sampler's correctness.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • One analysis yields both model odds and posteriors for favored models, replacing separate runs per model.
  • For informative, high-signal-to-noise events the sampler converges quickly to the favored model, cutting cost exactly where traditional analyses get expensive.
  • Source classification for ambiguous events can be read directly from sampling fractions, giving probabilities for black-hole, neutron-star, and mixed systems.
  • Comparing many models at once is feasible; disfavored models receive fewer samples, and the ranking matches a standard nested-sampling reference.
  • The approach extends to next-generation detectors, where separate analyses of loud, long signals would be computationally prohibitive.

Where Pith is reading between the lines

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

  • The design suggests a general recipe: any model-comparison problem with nested parameter spaces could use the same pseudo-parameter caching and learned linear maps, not just compact-binary source classes.
  • Because exactness of the Jacobian cancellation is asserted rather than numerically demonstrated across the full proposal range, a targeted check of the acceptance ratio against known Bayes factors would strengthen the central claim.
  • The BNS-versus-NSBH degeneracy at leading tidal order implies the sampler's classification probabilities may approximate which waveform family fits best rather than a definitive astrophysical composition; models that distinguish individual tidal deformabilities will sharpen the physical interpretation.
  • The computational advantage depends on parallelization improvements; the current implementation has overhead that limits scaling, so actual speedups may be smaller than the single-run savings suggest.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper presents t-roo, a reversible-jump MCMC sampler, built on eryn and Bilby, for transdimensional inference on gravitational-wave signals from compact binary coalescences. The sampler is designed to compare BBH, NSBH, and BNS waveform models with different parameter-space dimensionalities, returning both model probabilities/odds ratios and parameter posteriors in a single run. The between-model proposals use stretch moves for tidal parameters and data-informed linear maps for chirp mass and mass ratio, with the slopes estimated in a preliminary sampling phase. The authors validate t-roo on a set of injections and on two real events, comparing model odds and posteriors with dynesty, and claim agreement in the preferred models in all tests. The central mathematical claim is Eq. (19), that the auxiliary-variable ratio times the Jacobian equals 1, which is used to simplify the acceptance probability.

Significance. If correct, this would be a valuable new tool for gravitational-wave model selection, potentially reducing the cost of comparing many models or highly informative signals by performing one transdimensional run instead of separate nested-sampling runs. The paper includes useful details, such as pseudo-parameters that cache tidal information and a careful argument that priors on those pseudo-parameters must be included in the acceptance ratio (Appendix B4). The approach is well-motivated and the comparison infrastructure is standard. However, the central reversibility argument is not established: the Jacobian computation omits a term, the inverse between-model maps are not shown to be true inverses, and the tidal-proposal equations contain apparent typos. These issues directly affect the sampler's stationary distribution and therefore every reported odds ratio. The validation overclaim in Section V (and the abstract) is also contradicted by Table II. The underlying idea is promising, but the mathematical core needs substantial correction before the sampler's outputs can be trusted.

major comments (4)
  1. [Appendix B2, Eq. (B18) and Eq. (19)] The Jacobian derivation omits the dependence of the mass-ratio proposal on the mass ratio itself. For the BNS→BBH map in Eq. (16), q_bbh = q_bns − s_q \tildeΛ(Mc,q,Λ1,Λ2), so ∂q_bbh/∂q_bns = 1 − s_q ∂\tildeΛ/∂q, not 1. The Mc-row also has ∂Mc_bbh/∂q_bns = −s_Mc ∂\tildeΛ/∂q. Including these terms gives det J = (1 − s_q ∂\tildeΛ/∂q)/(w1 w2), not 1/(w1 w2). Hence Eq. (19) should read g'/g·J = 1 − s_q ∂\tildeΛ/∂q, which is not 1 in general (e.g., for unequal tidal deformabilities). This directly biases the acceptance probability in Eq. (18) and all model probabilities in Table II. The authors must provide the corrected Jacobian or prove that ∂\tildeΛ/∂q = 0.
  2. [Appendix B1, Eqs. (B12)–(B16)] The inverse transformations in Eqs. (B15)–(B16) are not demonstrated to be the true inverses of Eqs. (15)–(17). For example, the inverse BNS→BBH defines M'_c,bns via M'_c,bns = M_c,bbh + s_Mc \tildeΛ(M'_c,bns, q'_bns, ...), which is a fixed-point equation in the new masses. The condition u1w1 = 1 recovers the old tidal parameters, but it does not imply M'_c,bns = M_c,bns or q'_bns = q_bns; the statement in Eq. (B13) that reversibility follows from \tildeΛ_bns = \tildeΛ'_bns is circular. No contraction condition or uniqueness proof is given, and no invertibility range for s_Mc and s_q is provided. Without a genuine diffeomorphism h, the proposal pair does not satisfy the measure-matching requirement of Green's reversible-jump construction, so detailed balance is broken regardless of the Jacobian factor.
  3. [Eqs. (16)–(17) and (B15)–(B16)] The tidal-parameter proposals appear to contain typos. In Eq. (16), Λ2,bbh,i+1 is proposed as Λ2,j,i + w2(Λ1,bns,i − Λ2,j,i) rather than using Λ2,bns,i; the same pattern appears in Eq. (17) and in the inverse formulas (B15)–(B16), where Λ1 is used for both Λ1 and Λ2. If implemented literally, the transformation is not reversible: with u2=1/w2 the inverse gives Λ2,bns' = Λ2,j + u2(Λ1,bns − Λ2,j), which equals the original Λ2,bns only when Λ1,bns = Λ2,bns. The stated reversibility conditions u2w2=1 and v2w2=1 are therefore insufficient. Please correct the equations and verify that the code uses the intended component tidal deformability.
  4. [Section V, Table II, abstract] The claim 'in all the tests, the model preferred by t-roo and dynesty is the same' is contradicted by Table II. For nsbh lam600 lowsnr, t-roo gives lnO_{bbh}^{nsbh} = −0.86 (NSBH preferred) while dynesty gives +0.40 (BBH preferred). In the bns lam600 mm row, t-roo prefers IMRPhenomD NRTidalv2 over IMRPhenomXAS NRTidalv3 (lnO = +0.065) whereas dynesty prefers XAS NRTidalv3 (lnO = −0.22). The text in Sec. V C acknowledges the second discrepancy but the abstract and Sec. V overstate the agreement. This validation claim must be qualified, and the low-SNR discrepancy explained.
minor comments (5)
  1. [Section I, footnote 1] The GitHub URL 'https://github.com/AnnaPuecher/t roo' contains a space and is not clickable; it should be a valid repository link (e.g., with a hyphen or underscore). This hampers reproducibility.
  2. [Eq. (9) and Eq. (4)] The proposal distributions g(z) are written as 1/√z without a normalization constant. The normalization cancels in the acceptance ratio, but as written they are not probability densities. Please state that they are defined up to normalization, or include the normalizing factor.
  3. [Appendix B2, Eq. (B17)] The matrix notation is ambiguous: Θ is a vector, so the displayed block is not literally 6×6; the absolute-value bars inside individual entries of the Jacobian are also nonstandard. Please present the Jacobian with clear block notation and take the absolute value only of the final determinant.
  4. [General] There are several typographical errors, e.g., 'starey' after Eq. (18), 'dynetsyrecover' in Sec. V B, and inconsistent notation for the tidal parameters in the equations noted above. A careful proofreading pass is needed.
  5. [Section V E] The computational-advantage argument is weakened by the authors' own statement that 'eryn does not scale properly when running on more than 8 cores' and that dynesty can run in parallel on many CPUs. This is acknowledged, but it should be more prominently reflected in the abstract and conclusions.

Circularity Check

0 steps flagged

No significant circularity: the sampler's target distribution and validation are independent of its data-informed proposal design.

full rationale

The paper's derivation chain is not circular. The target joint posterior p(θ_k,k|d) is defined by Bayes' theorem (Eq. A5) and is independent of the sampler's output; the RJMCMC acceptance rule (Eq. 18) is the standard Metropolis-Hastings correction, and Eq. (19) is a stated mathematical identity about the designed between-model maps, not a fitted result relabeled as a prediction. The slopes s_Mc and s_q (Eqs. 13-14) are proposal-tuning constants obtained during a preliminary phase; they are not the quantities being predicted and do not define the model probabilities, which are estimated from the chain's visit counts (Eq. 22). Validation against dynesty is an external benchmark using independent nested-sampling evidence estimates, so the agreement is evidence rather than circular confirmation. Self-citations (e.g., Refs. [21,22]) are contextual and not load-bearing for the central claims. Even if Eq. (19) failed to hold or the inverse maps were not true diffeomorphisms, that would be a correctness or verification defect, not circularity. No step in the manuscript reduces a claimed prediction to its own input by construction.

Axiom & Free-Parameter Ledger

5 free parameters · 7 axioms · 1 invented entities

The central scientific output is the sampler and its validation; the ledger is dominated by proposal-tuning parameters and auxiliary variables rather than physical constants. The most consequential entries are the data-fitted slopes s_Mc and s_q, the pseudo-parameters, and the detailed-balance identity on which the method's validity depends.

free parameters (5)
  • s_Mc(mod1,mod2) = computed from data; not quoted
    Slope of chirp-mass proposal in between-model jumps; derived from likelihood center-of-mass of last 200 preliminary samples (Eqs. 13). It is fitted to the same data being analyzed, so it is a data-dependent proposal tuning parameter.
  • s_q(mod1,mod2) = computed from data; not quoted
    Same as s_Mc but for the mass-ratio proposal (Eq. 14).
  • epsilon = 0.01 (default)
    Width of the stretch move for tidal pseudo-parameters (Eq. 9), chosen by hand to keep jumps small.
  • step_rw = 10
    Random-walk step size for pseudo-parameters in within-model moves; chosen by hand so their posterior approximates the prior.
  • center-of-mass window = last 200 iterations
    Number of preliminary samples used to estimate the likelihood center-of-mass and slopes; chosen by hand.
axioms (7)
  • standard math Reversible-jump acceptance probability (Green 1995) with dimension-matching maps yields the joint model-parameter posterior as stationary distribution.
    Used throughout; the sampler's model probabilities p(k|d)=n_k/Σn_k (Eq. 2) are valid only if detailed balance holds.
  • standard math The between-model maps (Eqs. 15-17) are invertible diffeomorphisms with dimension matching.
    Needed for the Jacobian computation; the paper derives reversibility conditions u_i w_i=1 and v_i w_i=1 in Appendix B.1.
  • domain assumption Priors on pseudo-parameters are uniform on [0,5000] and normalized, so they do not change the marginal model probabilities.
    Appendix B.4 shows these priors must be included in the acceptance ratio to preserve detailed balance.
  • domain assumption All models have equal prior probability p(k); astrophysical priors can be applied a-posteriori.
    Section III.A: 'currently the models' priors can be incorporated a-posteriori by multiplying the final odds ratios.'
  • domain assumption Waveform approximants and prior ranges (Mc windows, q in [0.16,1], Λ in [0,5000], aligned spins ≤0.9) are valid for the masses and signals analyzed.
    Section IV.B; invalid waveform evaluations would artificially disfavor models, as the paper notes for masses near the NS maximum.
  • domain assumption The data-informed slope parameters are fixed after the preliminary sampling phase and the proposal kernel is stationary.
    Required for the proposal distribution to be a fixed kernel for which detailed balance is checked; Section III.A.2 computes slopes from last 200 iterations.
  • domain assumption Multiband likelihood and phase/distance marginalization provide a sufficiently accurate likelihood for model selection.
    Section II.C; these approximations affect both t-roo and dynesty runs.
invented entities (1)
  • Pseudo-parameters Λ1,bbh, Λ2,bbh, Λ1,nsbh no independent evidence
    purpose: Storage/cache of tidal-deformability information when the sampler is in a BBH or NSBH model that does not use those parameters in the likelihood; they are updated by random walk and used to seed tidal parameters on jumps into BNS/NSBH models.
    They are auxiliary variables with no physical meaning. Their priors are included in the acceptance ratio to preserve detailed balance (Appendix B.4).

pith-pipeline@v1.3.0-alltime-deepseek · 38146 in / 18106 out tokens · 167487 ms · 2026-08-01T07:17:55.117270+00:00 · methodology

0 comments
read the original abstract

Bayesian inference is commonly employed in the analysis of gravitational-wave signals not only to estimate the source parameters, but also for model selection. The latter provides insight into the physics of the source and has the potential to inform the direction for future model development. Although model comparison is usually performed by analyzing the data separately with different models and comparing the obtained Bayesian evidences, an alternative approach consists in sampling directly over the model itself. Here, we present t-roo, a reversible jump Markov chain Monte Carlo sampler capable of performing transdimensional inference on gravitational-wave signals from compact binary coalescences. Employing t-roo, a single analysis provides simultaneously the model odds ratio and the parameter posteriors for the favored models, hence yielding a potentially substantial computational advantage, particularly when comparing many models or analyzing highly informative data. t-roo is built on the sampler eryn and is specifically designed to compare models describing different kinds of sources, i.e., binary black hole, binary neutron star, or neutron star-black hole systems, as well as multiple models for the same source class. We validate the sampler on a set of injections, finding agreement with the results obtained with the nested sampler dynesty. We then use t-roo to analyze the real events GW190425 and GW230529, for which the system's parameters alone do not provide conclusive evidence of the presence of a neutron star component. t-roo can be adapted to any model-comparison scenario, thus providing a valuable tool in particular for next-generation detectors, where analyzing data separately with competing models becomes computationally even more demanding.

Figures

Figures reproduced from arXiv: 2607.21484 by Anna Puecher, Anuradha Samajdar, Gregory Ashton, Hauke Koehn, Jonathan Gair, Luca Negri, Tim Dietrich.

Figure 1
Figure 1. Figure 1: FIG. 1. Schematic representation of the moves implemented in [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: shows an example of the distribution of Mc and Λ samples after the preliminary sampling, illustrat- ˜ ing how the NSBH and BBH model find disjoint chirp mass intervals. To identify a mapping between these different intervals, and similarly for the mass ratio, the sampler considers the Mc, q, and Λ samples from each ˜ model after the preliminary sampling phase has finished, with Λ being the mass-weighted (o… view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Top panel: Model probabilities, computed as the [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Parameter posterior distributions recovered by [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Model probabilities for the [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Parameter posterior distributions recovered by [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Parameter posterior distributions recovered by [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8. Comparison between odds ratios for the [PITH_FULL_IMAGE:figures/full_fig_p029_8.png] view at source ↗

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

100 extracted references · 67 linked inside Pith

  1. [1]

    Bayesian model selection According to Bayes’ theorem, we can infer the parameters ⃗θof an observed system from the observed datadas p(⃗θ|d,H) = p(d|⃗θ,H)p( ⃗θ|H) p(d|H) ,(A1) wherep( ⃗θ|d,H) is the posterior probability of the parameters ⃗θgiven the datadand a model or hypothesisHto describe the data,p(d| ⃗θ,H) is thelikelihood, andp( ⃗θ|H) is the prior p...

  2. [2]

    positive recurrance

    MCMC review Markov chains are stochastic processes that define a sequence of states of a system, in which each state depends on the previous one, and only on that. A Markov chain is characterized by a transition kernelT, which describes the transition probability from one state to the other — i.e.,T(x, y) denotes the conditional probability of the next st...

    1993

  3. [3]

    Suppose we are in a state xi in the Markov Chain, to move to the next statex i+1 one needs to:

    Metropolis-Hastings algorithm The Metropolis-Hastings algorithm for MCMC sampling can be summarized as follows. Suppose we are in a state xi in the Markov Chain, to move to the next statex i+1 one needs to:

  4. [4]

    Propose the new statex i+1, generating it from the proposal distributionq(x i+1|xi), which describes the proba- bility ofx i+1 givenx i (and onlyx i, since Markov chains depend only on the previous point)

  5. [5]

    Decide whether to accept the new state or not, based on theacceptance probabilityα. To decide whether the proposed point is accepted, a random valuerbetween [0,1] is drawn: ifα≥r, the proposedx i+1 is accepted and saved as the next state in the chain, otherwise it is rejected andx i is saved again as the chain state. The acceptance ratio is computed as α=...

  6. [6]

    Generalizing Metropolis-Hastings To facilitate the extension to the trans-dimensional case, we first consider the Metropolis-Hastings MCMC in a more general way. We now consider two statesxandx ′, belonging to the setsAandA ′, respectively, which for now we assume are subsets (more precisely Borel sets) of a larger parameter space, e.g.,A, A ′ ⊆R n. We im...

  7. [7]

    The chain is a statex∈A, it proposes a move tox ′ ∈A ′, and the move is accepted

  8. [8]

    The chain is in a statex∈A ′, it proposes a move anywhere but the move is not accepted, therefore the chain remains in the setA ′. This can be written as P(x, A′) = Z A′ q(x′|x)α(x, x′)dx′ | {z } case 1 + Z q(x′|x)[1−α(x, x′)]dx′ ·I {x∈A′} | {z } case 2 ,(A20) where [1−α(x, x′)] denotes the probability ofnotaccepting the move tox ′, andI {x∈A′} is the ind...

  9. [9]

    solution

    The trans-dimensional case So far we have assumed that the Markov chains always move between spaces with the same dimensionality. However, especially in the case where we want to compare different models, the states involved might belong to spaces with different dimensions. A “solution” to this issue, proposed by Green in Ref. [42], consists in introducin...

  10. [10]

    standard

    Reversible Jump MCMC As discussed earlier, when using RJMCMC algorithms in Bayesian inference to compare models or hypothesis is to sample the joint posterior probabilityp( ⃗θk, k|d) of the parametersandthe model describing the data. Therefore, the chains move between states given byx= ( ⃗θk, k), withkbeing the label denoting the model and ⃗θk the paramet...

  11. [11]

    original

    F ull proposal derivation and inverse moves In the following, we show how the move proposals in Eqs. (15) to (17) in Sec. III are derived, and how we obtain the proposals for the inverse move. Let us start from the proposal for the tidal parameters and look at the case BNS→BBH as an example. As discussed in the main text, we update the tidal parameter thr...

  12. [12]

    A 5, it is essential to include the Jacobian in the acceptance-ratio computation for the between- model move

    Jacobians As discussed in Sec. A 5, it is essential to include the Jacobian in the acceptance-ratio computation for the between- model move. In the following we compute the Jacobian specifically for the move BNS→BBH, i.e., for the mapping in Eq. (16). The Jacobians for all the other transformations can be computed with the same procedure. The Jacobian for...

  13. [13]

    Symmetry conditions for stretch moves The stretch move employed int-rooproposals is described in Sec. II A. The acceptance ratio for the stretch move shown in Eq. (5) includes a term to preserve the symmetry condition in the proposal. Note that this is different from the detailed balance condition, which includes in its definition the acceptance ratio its...

  14. [14]

    empirical

    Priors on pseudo-parameters As explained in Sec. III, int-roowe add thepseudo-parameters Λ 1,bbh, Λ2,bbh, and Λ1,nsbh to store the information about tidal parameters in models that would not include such parameters. In this section, we show that the prior on thesepseudo-parameters has to (crucially) be included in the acceptance ratio in order to preserve...

    2000

  15. [15]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2605.27225 (2026), arXiv:2605.27225 [gr-qc]

    Pith/arXiv arXiv 2026

  16. [16]

    Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2605.27226 (2026), arXiv:2605.27226 [astro-ph.HE]

    N. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2605.27226 (2026), arXiv:2605.27226 [astro-ph.HE]

    Pith/arXiv arXiv 2026

  17. [17]

    B. P. Abbottet al.(LIGO Scientific, Virgo), PRL119, 161101 (2017), arXiv:1710.05832 [gr-qc]

    Pith/arXiv arXiv 2017

  18. [18]

    B. P. Abbottet al.(LIGO Scientific, Virgo, Fermi GBM, INTEGRAL, IceCube, AstroSat Cadmium Zinc Telluride Imager Team, IPN, Insight-Hxmt, ANTARES, Swift, AGILE Team, 1M2H Team, Dark Energy Camera GW-EM, DES, DLT40, GRA WITA, Fermi-LAT, ATCA, ASKAP, Las Cumbres Observatory Group, OzGrav, DWF (Deeper Wider Faster Program), AST3, CAAS- TRO, VINROUGE, MASTER, ...

    Pith/arXiv arXiv 2017

  19. [19]

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

    Pith/arXiv arXiv 2017

  20. [20]

    Dietrich, M

    T. Dietrich, M. W. Coughlin, P. T. H. Pang, M. Bulla, J. Heinzel, L. Issa, I. Tews, and S. Antier, Science370, 1450 (2020), arXiv:2002.11355 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  21. [21]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2605.27227 (2026), arXiv:2605.27227 [astro-ph.CO]

    Pith/arXiv arXiv 2026

  22. [22]

    Abbottet al.(LIGO Scientific, VIRGO, KAGRA), Phys

    R. Abbottet al.(LIGO Scientific, VIRGO, KAGRA), Phys. Rev. D112, 084080 (2025), arXiv:2112.06861 [gr- qc]

    Pith/arXiv arXiv 2025

  23. [23]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2603.19019 (2026), arXiv:2603.19019 [gr-qc]

    Pith/arXiv arXiv 2026

  24. [24]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arXiv:2603.19020 (2026), arXiv:2603.19020 [gr-qc]

    Pith/arXiv arXiv 2026

  25. [25]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arxiv:2603.19021 (2026), arXiv:2603.19021 [gr-qc]

    Pith/arXiv arXiv 2026

  26. [26]

    Skilling, Bayesian Analysis1, 833 (2006)

    J. Skilling, Bayesian Analysis1, 833 (2006)

    2006

  27. [29]

    Aasiet al.(LIGO Scientific), Class

    J. Aasiet al.(LIGO Scientific), Class. Quant. Grav.32, 074001 (2015), arXiv:1411.4547 [gr-qc]

    Pith/arXiv arXiv 2015

  28. [30]

    Acerneseet al.(VIRGO), Class

    F. Acerneseet al.(VIRGO), Class. Quant. Grav.32, 024001 (2015), arXiv:1408.3978 [gr-qc]

    Pith/arXiv arXiv 2015

  29. [31]

    Akutsu, M

    T. Akutsu, M. Ando, K. Arai, Y. Arai, S. Araki, A. Araya, N. Aritomi, Y. Aso, S. Bae, Y. Bae, L. Baiotti, R. Bajpai, M. A. Barton, K. Cannon, E. Capocasa, M. Chan, C. Chen, K. Chen, Y. Chen, H. Chu, Y. K. Chu, S. Eguchi, Y. Enomoto, R. Flaminio, Y. Fujii, M. Fukunaga, M. Fukushima, G. Ge, A. Hagiwara, S. Haino, K. Hasegawa, H. Hayakawa, K. Hayama, Y. Hime...

    2021

  30. [32]

    J. S. Speagle, Mon. Not. Roy. Astron. Soc.493, 3132 (2020), arXiv:1904.02180 [astro-ph.IM]

    Pith/arXiv arXiv 2020

  31. [33]

    Ashton and C

    G. Ashton and C. Talbot, Mon. Not. Roy. Astron. Soc. 507, 2037 (2021), arXiv:2106.08730 [gr-qc]

    Pith/arXiv arXiv 2037

  32. [34]

    Ashtonet al., Astrophys

    G. Ashtonet al., Astrophys. J. Suppl.241, 27 (2019), arXiv:1811.02042 [astro-ph.IM]

    Pith/arXiv arXiv 2019

  33. [35]

    Ashton and T

    G. Ashton and T. Dietrich, Nature Astron.6, 961 (2022), arXiv:2111.09214 [gr-qc]

    Pith/arXiv arXiv 2022

  34. [36]

    Puecher, A

    A. Puecher, A. Samajdar, G. Ashton, C. Van Den Broeck, and T. Dietrich, Phys. Rev. D109, 023019 (2024), arXiv:2310.03555 [gr-qc]

    Pith/arXiv arXiv 2024

  35. [37]

    A. G. Abacet al.(LIGO Scientific, VIRGO, KA- GRA), arXiv preprint arxiv:2508.18083 (2025), arXiv:2508.18083 [astro-ph.HE]. [24]https://emfollow.docs.ligo.org/userguide/(2026), LIGO/Virgo Public Alerts User Guide

    Pith/arXiv arXiv 2025

  36. [38]

    Abbottet al.(LIGO Scientific, Virgo), Astrophys

    R. Abbottet al.(LIGO Scientific, Virgo), Astrophys. J. Lett.896, L44 (2020), arXiv:2006.12611 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  37. [39]

    A. G. Abacet al.(LIGO Scientific, KAGRA, VIRGO), Astrophys. J. Lett.970, L34 (2024), arXiv:2404.04248 [astro-ph.HE]

    Pith/arXiv arXiv 2024

  38. [40]

    B. P. Abbottet al.(LIGO Scientific, Virgo), Astrophys. J. Lett.892, L3 (2020), arXiv:2001.01761 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  39. [41]

    Barbieri, O

    C. Barbieri, O. S. Salafia, M. Colpi, G. Ghirlanda, and A. Perego, Astron. Astrophys.654, A12 (2021), arXiv:2002.09395 [astro-ph.HE]

    Pith/arXiv arXiv 2021

  40. [42]

    R. Dudi, A. Adhikari, B. Br¨ ugmann, T. Dietrich, K. Hayashi, K. Kawaguchi, K. Kiuchi, K. Kyutoku, M. Shibata, and W. Tichy, Phys. Rev. D106, 084039 (2022), arXiv:2109.04063 [astro-ph.HE]

    Pith/arXiv arXiv 2022

  41. [43]

    R. J. Foley, D. A. Coulter, C. D. Kilpatrick, A. L. Piro, E. Ramirez-Ruiz, and J. Schwab, Mon. Not. Roy. Astron. Soc.494, 190 (2020), arXiv:2002.00956 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  42. [44]

    Kyutoku, S

    K. Kyutoku, S. Fujibayashi, K. Hayashi, K. Kawaguchi, K. Kiuchi, M. Shibata, and M. Tanaka, Astrophys. J. Lett.890, L4 (2020), arXiv:2001.04474 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  43. [45]

    Han, S.-P

    M.-Z. Han, S.-P. Tang, Y.-M. Hu, Y.-J. Li, J.-L. Jiang, Z.-P. Jin, Y.-Z. Fan, and D.-M. Wei, Astrophys. J. Lett. 891, L5 (2020), arXiv:2001.07882 [astro-ph.HE]

    Pith/arXiv arXiv 2020

  44. [46]

    Khadkikar and D

    S. Khadkikar and D. Singh, Phys. Rev. D113, 023029 (2026), arXiv:2507.07895 [astro-ph.HE]

    Pith/arXiv arXiv 2026

  45. [47]

    I. Tews, P. T. H. Pang, T. Dietrich, M. W. Coughlin, S. Antier, M. Bulla, J. Heinzel, and L. Issa, Astrophys. J. Lett.908, L1 (2021), arXiv:2007.06057 [astro-ph.HE]

    Pith/arXiv arXiv 2021

  46. [48]

    L. L. Lopes and D. P. Menezes, Astrophys. J.936, 41 (2022), arXiv:2111.02247 [hep-ph]

    Pith/arXiv arXiv 2022

  47. [49]

    B. K. K. Lee, M.-c. Chu, and L.-M. Lin, Astrophys. J. 922, 242 (2021), arXiv:2110.05538 [astro-ph.HE]

    Pith/arXiv arXiv 2021

  48. [50]

    Biswas, R

    B. Biswas, R. Nandi, P. Char, S. Bose, and N. Ster- gioulas, Mon. Not. Roy. Astron. Soc.505, 1600 (2021), arXiv:2010.02090 [astro-ph.HE]

    Pith/arXiv arXiv 2021

  49. [51]

    Bombaci, A

    I. Bombaci, A. Drago, D. Logoteta, G. Pagliara, and I. Vida˜ na, Phys. Rev. Lett.126, 162702 (2021), arXiv:2010.01509 [nucl-th]

    arXiv 2021

  50. [52]

    Koehn, T

    H. Koehn, T. Wouters, H. Rose, P. T. H. Pang, R. Soma- sundaram, I. Tews, and T. Dietrich, Phys. Rev. D110, 103015 (2024), arXiv:2407.07837 [astro-ph.HE]

    Pith/arXiv arXiv 2024

  51. [53]

    Janquartet al., Mon

    J. Janquartet al., Mon. Not. Roy. Astron. Soc.537, 1001 (2025), arXiv:2409.07298 [gr-qc]

    Pith/arXiv arXiv 2025

  52. [54]

    Markin, A

    I. Markin, A. Puecher, M. Bulla, and T. Dietrich, Phys. Rev. D113, 024031 (2026), arXiv:2508.08750 [gr-qc]

    arXiv 2026

  53. [55]

    P. J. GREEN, Biometrika82, 711 (1995), https://academic.oup.com/biomet/article- pdf/82/4/711/699533/82-4-711.pdf

    1995

  54. [56]

    Hastie and P

    D. Hastie and P. Green, Statistica Neerlandica66(2012), 10.1111/j.1467-9574.2012.00516.x

    arXiv 2012

  55. [57]

    Karnesis, M

    N. Karnesis, M. L. Katz, N. Korsakova, J. R. Gair, and N. Stergioulas, Mon. Not. Roy. Astron. Soc.526, 4814 (2023), arXiv:2303.02164 [astro-ph.IM]

    Pith/arXiv arXiv 2023

  56. [58]

    Umstatter, N

    R. Umstatter, N. Christensen, M. Hendry, R. Meyer, V. Simha, J. Veitch, S. Vigeland, and G. Woan, Class. Quant. Grav.22, S901 (2005), arXiv:gr-qc/0503121

    Pith/arXiv arXiv 2005

  57. [59]

    Umstatter, N

    R. Umstatter, N. Christensen, M. Hendry, R. Meyer, V. Simha, J. Veitch, S. Vigeland, and G. Woan, Phys. Rev. D72, 022001 (2005), arXiv:gr-qc/0506055

    Pith/arXiv arXiv 2005

  58. [60]

    W. M. Farr, I. Mandel, and D. Stevens, Royal Society Open Science2, 150030 (2015), arXiv:1104.0984 [astro- ph.IM]

    Pith/arXiv arXiv 2015

  59. [61]

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

    Pith/arXiv arXiv 2015

  60. [62]

    T. B. Littenberg and N. J. Cornish, Phys. Rev. D91, 084034 (2015), arXiv:1410.3852 [gr-qc]

    Pith/arXiv arXiv 2015

  61. [63]

    Tonget al., Astrophys

    H. Tonget al., Astrophys. J. Suppl.276, 50 (2025), arXiv:2404.04460 [gr-qc]

    Pith/arXiv arXiv 2025

  62. [64]

    Chatziioannou, N

    K. Chatziioannou, N. Cornish, M. Wijngaarden, and T. B. Littenberg, Phys. Rev. D103, 044013 (2021), arXiv:2101.01200 [gr-qc]

    Pith/arXiv arXiv 2021

  63. [65]

    S. Y. Cheung, R. Udall, D. Davis, P. D. Lasky, and E. Thrane, arXiv preprint arxiv:2607.09111 (2026), arXiv:2607.09111 [gr-qc]. 32

    Pith/arXiv arXiv 2026

  64. [66]

    W. K. Hastings, Biometrika57, 97 (1970), https://academic.oup.com/biomet/article- pdf/57/1/97/23940249/57-1-97.pdf

    1970

  65. [67]

    Foreman-Mackey, D

    D. Foreman-Mackey, D. W. Hogg, D. Lang, and J. Good- man, Publications of the Astronomical Society of the Pa- cific125, 306 (2013), arXiv:1202.3665 [astro-ph.IM]

    Pith/arXiv arXiv 2013

  66. [68]

    Goodman and J

    J. Goodman and J. Weare, Communications in Applied Mathematics and Computational Science5, 65 (2010)

    2010

  67. [69]

    Wang and R

    J. Wang and R. H. Swendsen, Progress of Theoreti- cal Physics Supplement157, 317 (2005), arXiv:cond- mat/0407273 [cond-mat.stat-mech]

    arXiv 2005

  68. [70]

    W. R. Gilks, G. O. Roberts, and S. K. Sahu, Journal of the American Statistical Association93, 1045 (1998)

    1998

  69. [71]

    D. J. Earl and M. W. Deem, Physical Chemistry Chemi- cal Physics (Incorporating Faraday Transactions)7, 3910 (2005), arXiv:physics/0508111 [physics.comp-ph]

    Pith/arXiv arXiv 2005

  70. [72]

    W. D. Vousden, W. M. Farr, and I. Mandel, Mon. Not. Roy. Astron. Soc.455, 1919 (2016), arXiv:1501.05823 [astro-ph.IM]

    Pith/arXiv arXiv 1919

  71. [73]

    Thrane and C

    E. Thrane and C. Talbot, Publications of the As- tronomical Society of the Pacific36, e010 (2019), arXiv:1809.02293 [astro-ph.IM]

    Pith/arXiv arXiv 2019

  72. [74]

    Morisaki, Phys

    S. Morisaki, Phys. Rev. D104, 044062 (2021), arXiv:2104.07813 [gr-qc]

    Pith/arXiv arXiv 2021

  73. [75]

    Zackay, L

    B. Zackay, L. Dai, and T. Venumadhav, arXiv preprint arxiv:1806.08792 (2018), arXiv:1806.08792 [astro-ph.IM]

    Pith/arXiv arXiv 2018

  74. [76]

    Smith, S

    R. Smith, S. E. Field, K. Blackburn, C.-J. Haster, M. P¨ urrer, V. Raymond, and P. Schmidt, Phys. Rev. D94, 044031 (2016), arXiv:1604.08253 [gr-qc]

    Pith/arXiv arXiv 2016

  75. [77]

    Favata, Phys

    M. Favata, Phys. Rev. Lett.112, 101101 (2014), arXiv:1310.8288 [gr-qc]

    Pith/arXiv arXiv 2014

  76. [78]

    LIGO Scientific Collaboration,Noise Curves Used for Simulations in the Update of the Observing Scenarios Paper, Tech. Rep. LIGO-T2000012-v2 (LIGO Scientific Collaboration, 2022)

    2022

  77. [79]

    LIGO Scientific Collaboration, Virgo Collaboration, and KAGRA Collaboration, Living Reviews in Relativity23, 3 (2020)

    2020

  78. [80]

    J. E. Thompson, E. Fauchon-Jones, S. Khan, E. Nitoglia, F. Pannarale, T. Dietrich, and M. Hannam, Phys. Rev. D101, 124059 (2020), arXiv:2002.08383 [gr-qc]

    Pith/arXiv arXiv 2020

  79. [81]

    Pannarale, E

    F. Pannarale, E. Berti, K. Kyutoku, B. D. Lackey, and M. Shibata, Phys. Rev. D92, 084050 (2015), arXiv:1509.00512 [gr-qc]

    Pith/arXiv arXiv 2015

  80. [82]

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

    Pith/arXiv arXiv 2016

Showing first 80 references.