REVIEW 3 major objections 5 minor 18 references
Multi-messenger Gravitational-Wave + High-Energy Neutrino Searches with LIGO, Virgo, and IceCube
T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read The LLAMA pipeline found no statistically significant joint gravitational-wave and high-energy neutrino event in the first 14 LIGO/Virgo O3 alerts, with all reported p-values above 1%.
desk verdict A short, honest proceedings status report: real operational null result for 14 O3 alerts, but the BNS p-values are missing, so the headline claim is under-supported. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the LLAMA pipeline and its Bayesian test statistic, the odds ratio $O_{gw+\nu}$. The pipeline pulls public gravitational-wave alerts and IceCube's low-latency neutrino stream, searches a $\pm 500$ s window for coincident neutrinos, and computes the odds ratio using priors of uniform sky position, uniform-in-volume distance, and log-uniform energy for both messengers, together with a Gaussian model for neutrino directional errors. The odds ratio is converted to a p-value by comparing it with a background distribution generated from scrambled neutrino data, and the significance threshold for a public alert is a p-value below 1%.
What would settle it
Recompute the joint p-values for the 14 O3 alerts once the revised binary-neutron-star source distributions are adopted. If any of the withheld BNS candidates (S190425z, S190426c, S190510g) then yields a combined p-value below 1%, or if any currently reported p-value shifts below 1% under the updated priors, the paper's claim of no significant joint detection would be overturned.
Extended reading notes
Core claim
Using a Bayesian framework that incorporates astrophysical priors for both messengers, the LLAMA pipeline computes an odds ratio $O_{gw+\nu}$ comparing the hypothesis that a gravitational wave and at least one high-energy neutrino come from the same astrophysical source against the null and chance-coincidence hypotheses. For each of the 14 O3 gravitational-wave alerts, this odds ratio is converted to a p-value through a background test-statistic distribution built from scrambled neutrino data. None of the resulting p-values fell below 1%, so the search reports no significant joint detection. The p-values for the three binary-neutron-star candidates are not reported because the BNS source distributions are being revised, and the p-values for the black-hole candidates are described as preliminary.
Load-bearing premise
The quoted significances rest on the Bayesian model and background calibration developed in the two earlier papers, which assume Gaussian neutrino direction errors, a uniform-in-volume distance prior, and log-uniform energy priors; the paper itself notes that the binary-neutron-star distributions are being revisited, so the BNS p-values could change.
Editorial extensions
If this is right
- With no event below a p-value of 1% in the first 14 O3 alerts, the search currently yields upper limits on high-energy neutrino fluence for these gravitational-wave sources.
- A future alert with a p-value below 1% will trigger GCN circulars reporting neutrino directions, angular uncertainties, time offsets, and p-values from both the LLAMA and companion maximum-likelihood searches, enabling prompt electromagnetic follow-up with sub-degree localization.
- For binary neutron star mergers, joint searches can reveal choked-jet physics through neutrinos that escape the ejecta even when electromagnetic emission is absorbed, and non-detections can place limits on choked-jet model parameters.
- The quoted discovery potential of about $0.049\ \mathrm{GeV/cm^2}$ corresponds to a 50% detection probability averaged over the sky, indicating the search is sensitive to nearby, strongly emitting events.
Reading between the lines
- Because most of the 14 O3 alerts are binary black holes, which are not expected to be strong neutrino emitters, this non-detection is not yet a strong test of neutron-star jet physics; the withheld BNS p-values carry that test.
- If the revised BNS priors shift any of the three BNS-candidate p-values below 1%, the paper's headline conclusion of no significant detection could reverse, a check that becomes available as soon as the new distributions are adopted.
- The sub-degree neutrino localization advantage suggests that the most promising near-term chance for a first joint detection is a BNS or neutron-star–black-hole merger in the northern sky, where IceCube's effective area is larger and atmospheric background is lower.
- Applying the same Bayesian search to the full completed O3 dataset, rather than only the real-time alerts, would provide a more sensitive all-sky search and could supersede the preliminary numbers reported here.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript, a proceedings contribution from the IceCube Collaboration, describes the Low-Latency Algorithm for Multi-messenger Astrophysics (LLAMA) pipeline for searching for coincident gravitational-wave (GW) and high-energy neutrino (HEN) events. The pipeline combines LIGO/Virgo public alerts with IceCube GFU data, uses a Bayesian Bayes-factor statistic introduced in prior work, and has been running in realtime since the start of O3. For the first 14 GW alerts, the paper reports p-values for 11 BBH-type candidates, with three BNS-type candidates (S190425z, S190426c, S190510g) omitted because the BNS distributions are under revision. The paper claims that no event with p-value < 1% has been found, and provides a discovery potential estimate: a fluence of 0.049 GeV/cm^2 is detectable with 50% probability at the 3-sigma level.
Significance. If the null result is taken at face value, it is a useful realtime constraint on joint GW+HEN emission and demonstrates the operational readiness of the LLAMA pipeline. The paper is transparent about the preliminary nature of the analysis, references the detailed derivation of the statistical method, and presents the data in a compact table. It also highlights the importance of choked-jet scenarios for BNS mergers, where neutrinos may be the only escaping messenger. However, the strength of the central claim is weakened by the omission of the BNS p-values and by the absence of systematic uncertainties on the discovery potential; these are directly relevant to the conclusions drawn in Sections 4 and 7.
major comments (3)
- [Section 4, Table 1] The central claim that 'no event with a p-value < 1% have been found to date' cannot be fully checked for the three BNS candidates (S190425z, S190426c, S190510g) because their p-values are omitted from Table 1. The table note states that BNS p-values are not reported because the BNS distributions are being revisited, but the Bayes factor in Eq. (3.1) depends on source priors such as distance (uniform in volume) and energy (log-uniform), so a change in BNS priors could shift the TS distribution and the p-values for exactly these events. The conclusion in Section 7 should be explicitly restricted to the BBH candidates, or the BNS p-values should be reported even if marked preliminary, so that the reader can independently verify the null claim.
- [Section 3, Eq. (3.1) and Section 4] The p-values quoted in Table 1 are conditional on the prior assumptions adopted in the Bayesian calculation, including the Gaussian directional-error model for neutrinos and the uniform-distance/log-uniform-energy source priors. The paper does not quantify how sensitive the p-values are to these assumptions. This is not just an academic concern, because the manuscript itself states that the BNS distributions are being revisited; if the revised priors change the TS calibration, previously borderline p-values could cross the 1% threshold. The authors should report the p-values under the current priors and also state explicitly which systematic variations are under consideration, ideally with a quantitative reassessment of the three BNS candidates.
- [Section 5, Figure 2] The discovery potential of 0.049 GeV/cm^2 is reported as a single-point estimate with no statistical or systematic uncertainty. The value depends on the assumed E^-2 spectrum, the population of injected GW sources, the background scrambling, and the prior choices, none of which are varied or discussed. The curve in Figure 2 would be much more informative if the authors added a systematic band (e.g., varying the spectral index or the source-distance prior) or at least explicitly stated the dominant sources of uncertainty. Without this, the reader cannot judge how robust the quoted sensitivity is.
minor comments (5)
- [Section 2] The phrase 'searches for HENs in a time window of 1000 s of the GW detected time' is grammatically awkward; it should read 'within 1000 s of the GW trigger time'.
- [Section 3] The sentence 'For the case of GW public alerts, H0 is zero' is ambiguous. It should be made explicit that the probability of the null hypothesis P(H0) is set to zero because the GW alert is a confirmed astrophysical event, rather than a background fluctuation.
- [Figure 1 caption] The caption states 'None of the neutrinos made a significant coincidence' — the subject-verb agreement is incorrect; it should be 'No neutrino made a significant coincidence'.
- [Table 1] The column heading 'p-value (binary merger)' is misleading; it refers to the p-value from the GW+HEN search, not to a property of the binary merger itself. A heading such as 'GW+HEN p-value (preliminary)' would be clearer.
- [Section 5] The definition of discovery potential is given, but the x-axis of Figure 2 is not described in the text. The reader has to infer the units and range from the figure; a brief description in the caption or text would improve clarity.
Circularity Check
No circular reduction: O3 p-values are measured outputs of the cited LLAMA Bayesian model, not fitted inputs or renamed predictions.
full rationale
The paper's central claim (Sec. 4, Table 1) is an empirical null result: applying LLAMA to 14 O3 LIGO/Virgo alerts, no GW+HEN coincidence reached p<1%. The Bayes factor in Eq. 3.1 is quoted from the authors' prior work [13], but it is not re-derived from the O3 data, nor is any parameter fitted from those alerts and then repackaged as a prediction. The p-values come from a fixed test statistic and a background TS distribution built from scrambled IceCube GFU events, so they are genuine measurements under the stated model assumptions. The assumptions (Gaussian neutrino directional errors, uniform distance prior, log-uniform energy prior) are declared in Sec. 3; whether they are correct is a model-validity question, not a circularity. The self-citations [11,13] supply the method but do not smuggle in the target result, because [13] is a prior method paper whose assumptions do not include the O3 alert outcomes. The paper itself flags an important limitation in Sec. 4: "The p-values for BNS candidates are not reported here due to currently revisiting the BNS distributions for this analysis." That omission weakens the completeness of the stated null claim for the three BNS candidates, but missing or preliminary p-values are evidential gaps, not definitional or fitted-input circularity. No equation in this paper is equivalent to its own output by construction. Score 1 reflects the absence of circular reduction despite heavy, but legitimate, reliance on self-cited methodology.
Assumptions & free parameters
free parameters (2)
- Search time window =
±500 s around GW trigger (1000 s total)
- Energy ranges for log-uniform priors =
not specified
assumptions (7)
- domain assumption A signal is equally likely to occur at any time during the observational period (time prior).
- domain assumption Sources are uniformly distributed in volume (distance prior).
- domain assumption Sky position is uniformly distributed.
- domain assumption Independent log-uniform distributions for HEN and GW energies.
- domain assumption Reconstructed neutrino direction can be described by a Gaussian distribution with the reported angular uncertainty.
- domain assumption For public GW alerts, the background hypothesis H0 has zero probability; GW events are real astrophysical events.
- domain assumption The expected number of detected HENs follows a Poisson distribution given source parameters (from ref. [13], Eqs. 17-20).
Cite this review
Pith. "Pith review of Multi-messenger Gravitational-Wave + High-Energy Neutrino Searches with LIGO, Virgo, and IceCube." pith.science (2026). https://pith.science/paper/5FVUDLBE
@misc{pith2026190804996,
author = {Pith},
title = {Pith review of: Multi-messenger Gravitational-Wave + High-Energy Neutrino Searches with LIGO, Virgo, and IceCube},
year = {2026},
howpublished = {\url{https://pith.science/paper/5FVUDLBE}},
note = {Machine review of arXiv:1908.04996}
}
read the original abstract
Multi-messenger searches for gravitational waves and high-energy neutrinos provide important insights into the dynamics of and particle acceleration by black holes and neutron stars. With LIGO's third observing period (O3), the number of gravitational wave detections has been substantially increased. The rapid identification of joint signals is crucial for electromagnetic follow-up observations of transient emission that is only detectable for short periods of time. High-energy neutrino direction can be reconstructed to sub-degree precision, making a joint detection far better localized than a standalone gravitational-wave signal. We present the latest sensitivity of joint searches and discuss the Low-Latency Algorithm for Multi-messenger Astrophysics (LLAMA) that combines LIGO/Virgo gravitational-wave candidates and searches in low-latency for coincident high-energy neutrinos from the IceCube Neutrino Observatory. We will further discuss future prospects of joint searches from the perspective of better understanding the interaction of relativistic and sub-relativistic outflows from binary neutron star mergers.
Figures
Reference graph
Works this paper leans on
-
[11]
Countryman et al., arXiv e-prints (2019) arXiv:1901.05486
S. Countryman et al., arXiv e-prints (2019) arXiv:1901.05486
arXiv 2019
-
[13]
Bartos et al., arXiv e-prints (2018) arXiv:1810.11467
I. Bartos et al., arXiv e-prints (2018) arXiv:1810.11467
arXiv 2018
-
[1]
IceCube Collaboration, M. G. Aartsen et al., Science 342 (2013) 1242856
2013
-
[2]
IceCube Collaboration, M. G. Aartsen et al., Phys. Rev. Lett 113 (2014) 101101
work page 2014
-
[3]
IceCube Collaboration, C. Haack and C. Wiebusch, PoS(ICRC2017)1005 (2017)
work page 2017
-
[4]
IceCube Collaboration, M. G. Aartsen et al., Astroparticle Physics 92 (2017) 30–41
work page 2017
-
[5]
IceCube Collaboration, M. G. Aartsen et al., Science 361 (2018) eaat1378
2018
-
[6]
IceCube Collaboration, M. G. Aartsen et al., Science 361 (2018) 147–151
2018
Show all 18 references
-
[7]
LIGO Scientific, Virgo Collaboration, B. P. Abbott et al., arXiv e-prints (2018) arXiv:1811.12907
2018 arXiv
-
[8]
LIGO Scientific, Virgo Collaboration, B. P. Abbott et al., Phys. Rev. Lett 119 (2017) 161101
2017
-
[9]
Albert et al., ApJ 870 (2019) 134
IceCube, ANTARES, LIGO Scientific, Virgo Collaboration, A. Albert et al., ApJ 870 (2019) 134
2019
-
[10]
Kimura et al., Phys
S. Kimura et al., Phys. Rev. D. 98 (2018) 043020
2018
-
[12]
Baret et al., Phys
B. Baret et al., Phys. Rev. D. 85 (2012) 103004
2012
-
[14]
S. D. Barthelmy et al., Ap&SS 231 (1995) 235–238
1995
-
[15]
Hussain et al., PoS(ICRC2019)859 (these proceedings)
IceCube Collaboration, R. Hussain et al., PoS(ICRC2019)859 (these proceedings)
-
[16]
IceCube Collaboration, M. G. Aartsen et al., Journal of Instrumentation 11 (2016) P11009
2016
-
[17]
Baret et al., Astroparticle Physics 35 (2011) 1–7
B. Baret et al., Astroparticle Physics 35 (2011) 1–7
2011
-
[18]
Adrián-Martínez et al., Phys
S. Adrián-Martínez et al., Phys. Rev. D. 93 (2016) 122010. 8
2016
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.