{"id":"131334c4-f36e-461b-9b1e-e2a22a9f06c0","arxiv_id":"2507.08131","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"No significant high-energy neutrino emission was found coincident with 2500 gravitational-wave alerts in O4 through April 2025, with upper limits set under an E^-2 spectrum assumption.","lead":"IceCube searched for neutrinos arriving within 1000 seconds of gravitational-wave alerts from LIGO/Virgo/KAGRA's O4 run and found no significant coincident signal. The paper also reports faster automated alerts for electromagnetic follow-up, cutting median response time from 56 to 22 minutes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the O4 null result is well supported; the uncorrected trial factors and missing numerical upper limits are presentation gaps, not flaws in the no-detection claim.","rationale":"The reader's weakest assumption concerned the scrambling-based background construction. I do not adopt that as load-bearing because the observed p-value distributions themselves are checked against the background expectation and are reported as consistent; if the scrambling were badly biased, this would likely show up as a deviation. The reader's other concerns are legitimate but not central to the null claim. The trial-factor caveat in Section 3 means the 22 sub-1% candidates should not be quoted as physical detections, but the paper already avoids doing so and the overall conclusion is null. The missing upper limits from the abstract are a substantive presentation gap, but they concern a secondary advertised result rather than the no-detection claim. Therefore no change to the reader's conditional verdict is needed.","tokens_in":11341,"tokens_out":7318,"duration_ms":86384,"concrete_test":"Recompute the LLAMA population p-value from the public per-event p-values in the ROC LVKNuTrackSearch archive, applying a trial-factor-corrected combination (e.g., Benjamini-Hochberg or a binomial count of p-value < 0.01); if the corrected combined p-value remains above 0.05, the null conclusion is robust.","verdict_should_be":"UNCHANGED","load_bearing_attack":"No load-bearing flaw found in the central null result. The claim that no significant high-energy neutrino emission is coincident with O4 GW alerts is supported by the reported LLAMA p-value distribution, the population test p-value of 0.15 (Section 5), and the UML observed p-values against background (Figure 4). The explicit statement in Section 3 that reported p-values are not corrected by any trial factor is a real limitation, but it cuts against overinterpreting the 22 sub-1% candidates, not against the null conclusion: these candidates are presented as follow-up alerts, not as detections, and the paper's overall conclusion explicitly says no joint emitters have been detected. The absence of the numerical upper limits promised in the abstract is a completeness gap, but it does not undermine the no-detection statement.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This ICRC proceedings paper describes IceCube's real-time follow-up of LIGO-Virgo-KAGRA O4 gravitational-wave alerts using two independent pipelines: LLAMA, a Bayesian analysis with astrophysical priors, and UML, an unbinned maximum-likelihood search. It reports cumulative results through April 1, 2025: 208 significant and 2292 low-significance alerts were processed, an automated GCN Notice stream reduced median response latency from 56 to 22 minutes, and no significant neutrino excess was found. The LLAMA p-value distribution is consistent with a uniform distribution, a population test gives p = 0.15, and the UML p-values agree with the background expectation. The paper also discusses two individual events with low p-values and states that 22 joint candidates below 1% were released for electromagnetic follow-up. The abstract claims that upper limits are set on flux and isotropic-equivalent energy, but no numerical upper limits appear in the body.","tokens_in":11436,"tokens_out":4909,"duration_ms":55879,"significance":"If the results are taken as stated, the paper provides a useful multi-messenger constraint: it extends IceCube's GW follow-up through roughly two years of O4, includes both significant and subthreshold alerts, and documents a practical improvement in real-time follow-up via automated GCN Notices. The central null claim is reasonably supported by the reported p-value distributions and by the population test p-value of 0.15, and the paper is transparent about the fact that individual p-values are not corrected for trial factors. However, the quantitative upper-limit result promised in the abstract is not actually presented, and the population test that underlies the strongest null statement is not specified. Both issues are fixable but need to be addressed before the paper can be considered complete.","major_comments":[{"comment":"The sentence reporting a population test p-value of 0.15 is the strongest quantitative null result in the paper, but the test is not defined: the text does not state the test statistic, the null distribution, how multi-event skymap updates are handled, or whether the non-uniform background expectation described for Figure 4 is accounted for. No reference is given for this test. Please specify the test construction or cite a methods paper; as written, this p-value cannot be independently assessed.","section":"§5"},{"comment":"The abstract states that the paper sets upper limits on both flux and isotropic-equivalent energy emitted in neutrinos, but no numerical upper limits, upper-limit definitions, or representative values appear anywhere in the text. Section 3 mentions a time-integrated flux sensitivity range in GCN Notices, but the paper itself does not report any of these quantities. This makes the stated deliverable unverifiable. Please add a table of representative limits (or an aggregate limit) and define the assumed spectrum, time window, and distance/energy conventions, or explicitly state that numerical limits are only available in the archived GCN Notices and do not form part of this paper.","section":"Abstract and §6"}],"minor_comments":[{"comment":"The criteria HasNS > 0.5 and HasRemnant > 0.5 are used to select events for the extended time window, but these LVK parameters are never defined or referenced. Please add a brief definition or a citation to the relevant LVK alert documentation.","section":"§2"},{"comment":"The discussion of Figure 4 explains why the background p-value expectation is non-uniform, but it would help to state explicitly how the observed black points in the right panel are combined into a global test statistic, or to state that no global test statistic is used.","section":"§5"},{"comment":"For the S230904n example, the text says LLAMA measured a p-value of 0.0037 and that the coincident neutrino was followed up by ZTF, but it does not report the released neutrino direction or angular uncertainty. A brief coordinate or a pointer to the relevant GCN Circular would make the example self-contained.","section":"§4"},{"comment":"The text contains several missing spaces and typographical issues, e.g., 'throughobservation', 'Highenergy', 'coincidingwithobservation', and 'counterpartswerealso'. A copyediting pass is recommended.","section":"Throughout"},{"comment":"The statement that 22 joint candidates had p-values below 1% is interesting, but no list, table, or distribution of these candidates is given. Even a small table with event IDs, pipeline, p-value, and release status would improve traceability without changing the null conclusion.","section":"§6"}],"recommendation":"major_revision","confidential_remarks":"This is a straightforward collaboration proceedings paper, and I see no reason to doubt the no-detection result itself. The two major gaps are both local and fixable: the under-specified population test and the missing numerical upper limits. I would accept after a revision that adds these details. The reliance on prior collaboration papers for method descriptions is normal for a proceedings format, but the population test in particular needs either a description or a proper citation, since it carries the paper's strongest aggregate null statement."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a straightforward null-result paper from a mature program. The O4 dataset is new, the LLAMA extension to subthreshold alerts is a genuine operational improvement, and the automated GCN notice stream cut median response latency from 56 to 22 minutes. Those are real, useful advances. The central claim—no significant neutrino emission coincident with LVK O4 alerts through April 1, 2025—is well supported by the LLAMA p-value distribution, the population p=0.15, and the UML background comparison. I agree with the reader that the null result should survive review.\n\nCredit where due: the paper is explicit that both pipelines were used in prior runs; the new pieces are the dataset, the subthreshold extension, and the automation. The latency improvement is quantified, and the two worked examples show the real-time system working as intended, including the skymap-update evolution that reduced significance when better localization arrived. That is a good demonstration of scientific self-correction in a real-time context.\n\nSoft spots, in proportion: the uncorrected trial factors for the 22 sub-1% candidates are a real limitation, but the paper presents them as follow-up alerts, not detections. It flags the issue in Section 3; I'd like that caveat repeated in the conclusions, but it's minor. The bigger gap is the abstract promises upper limits on flux and isotropic-equivalent energy, and the body doesn't give them numerically. For a proceedings paper that's forgivable, but a referee should ask for the numbers or a reference to a dedicated paper.\n\nThe scrambling-based background construction is the main place I'd worry about bias, but the paper notes they updated the background with O4 sky localization information, so the concern is at the level of a sensitivity check, not a demonstrated flaw. The population p=0.15 is consistent with background and not overinterpreted.\n\nWho is this for? People in multimessenger time-domain astronomy and anyone tracking cumulative GW–neutrino null constraints. It deserves a serious referee, but with the expectation that requested changes are minor and mostly completeness-driven. I'd accept it as a proceedings paper and ask for the upper-limit numbers and a trial-factor caveat in the conclusions.","headline":"O4 GW-neutrino null result from IceCube is solid and the latency improvements are real; the paper needs upper-limit numbers and a trial-factor caveat, but it deserves review.","tokens_in":12025,"tokens_out":3147,"would_cite":true,"duration_ms":32349,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"IceCube's real-time follow-up of LIGO-Virgo-KAGRA's fourth observing run found no high-energy neutrino emission coincident with any gravitational-wave alert through April 1, 2025, with a population-level $p$-value of 0.15 consistent with…","keywords":["high-energy neutrinos","gravitational-wave astronomy","multimessenger astronomy","real-time follow-up","compact binary mergers","IceCube","neutrino upper limits","LIGO-Virgo-KAGRA"],"falsifier":"A confirmed gravitational-wave alert with a coincident high-energy neutrino whose joint $p$-value falls far below the background expectation, or a full-O4 population test with $p$-value well below 0.01, would refute the no-emission conclusion. A simpler technical check is a signal-injection run that adds fake neutrino events at known positions to the GFU stream and tests whether both pipelines recover them at the claimed false-alarm rate.","tokens_in":1739,"feed_emoji":"🔭","tokens_out":8002,"duration_ms":139625,"temperature":0.7,"pith_summary":"This paper reports that IceCube's real-time follow-up of gravitational-wave alerts from the LIGO-Virgo-KAGRA fourth observing run (O4) found no high-energy neutrino emission coincident with any of the 208 significant and 2,292 low-significance alerts issued through April 1, 2025. The central statistical claim is that the distribution of $p$-values from the Bayesian LLAMA pipeline is consistent with the uniform distribution expected from background, with a population-level $p$-value of 0.15. Individual events did show low $p$-values, including two that triggered electromagnetic follow-up campaigns, but further analysis showed the apparent coincidences were not associated with the gravitational-wave events. The same null result holds for an extended search window of 14 days applied to the five alerts likely to contain a neutron star. These searches set upper limits on the neutrino flux and on the isotropic-equivalent energy emitted in neutrinos.","feed_headline":"No neutrinos from O4 gravitational-wave alerts","feed_subtitle":"Two pipelines scanned 2,500 alerts; combined population p-value of 0.15 looks like pure background.","key_machinery":"The argument is carried by two independent coincidence searches over the Gamma-ray Follow-Up (GFU) track-event stream, which provides low-latency high-energy neutrino candidates. LLAMA computes a Bayesian odds ratio using astrophysical priors (uniform time and sky location, $r^{-2}$ distance prior, log-uniform isotropic energies) and converts it to a frequentist $p$-value against a background built from scrambled neutrinos and the O4 gravitational-wave skymap set. The UML search builds an extended likelihood over the HEALPix pixelized sky with the GW skymap applied as a spatial penalty, and computes $p$-values from background-only pseudo-experiments. The population statement comes from combining per-event $p$-values and comparing the distribution to the uniform expectation, using an updated background that incorporates O4 sky localization information.","core_discovery":"The central claim is that none of the gravitational-wave alerts from the first part of O4 is accompanied by high-energy neutrino emission distinguishable from background. Two independent real-time searches, LLAMA and the unbinned maximum likelihood (UML) 'generic transient' search, examined each alert in a $\\pm 500$ s window around the merger time; for five neutron-star-containing candidates, UML also searched a $[-0.1, +14]$ day window. The observed $p$-value distributions and the population test ($p=0.15$) are consistent with background, and upper limits are placed on the neutrino flux and isotropic-equivalent energy for each alert. The paper also documents the switch to fully automated GCN notices, which reduced the median response latency from 56 minutes to 22 minutes.","pith_inferences":["If the null result continues to the end of O4, the combined population test could reach sensitivity to exclude optimistic models where a few percent of mergers produce bright neutrino flares; the paper does not extrapolate this far.","The scrambling-based background could be validated independently by injecting simulated neutrino events into the GFU stream and checking that both pipelines recover them at their claimed significance; the paper reports upper limits but does not show such an end-to-end test.","The 22-minute latency is dominated by the fixed $\\pm 500$ s window and by waiting for the next GFU track; an early-warning or shorter-window variant would cut this further, though with reduced sensitivity.","Joint GW+neutrino skymaps, mentioned as ongoing work, would likely reduce the number of misleading low-$p$-value alerts by suppressing chance coincidences between unassociated events."],"forward_implications":["If the null result persists through the remainder of O4, the accumulated population will place tighter upper limits on the neutrino flux and isotropic-equivalent energy emitted by compact binary mergers.","The automated GCN notice stream means that electromagnetic observatories now receive neutrino coincidence information within roughly 22 minutes of a merger, a latency short enough to enable prompt follow-up of joint candidates.","The consistency of the $p$-value distributions with background constrains the fraction of O4 compact binary mergers that emit detectable high-energy neutrinos, limiting hadronic jet models.","The two sub-1% events illustrate the real-time program in action: low-$p$-value alerts are circulated and followed up quickly, even when later skymap updates or optical observations show the coincidence was not a true association."],"supporting_citations":[{"why":"Established the previous O3 follow-up with the UML method that this search extends and shares.","marker":"[4]"},{"why":"Reported the earlier O1/O2 LLAMA search, providing the precedent and baseline for the Bayesian pipeline.","marker":"[5]"},{"why":"Defines the Gamma-ray Follow-Up (GFU) event stream whose track candidates form the search dataset.","marker":"[6]"},{"why":"Provides the GRB emission model on which the ±500 s coincidence time window is based.","marker":"[7]"},{"why":"Introduces the LLAMA method (Bayesian odds ratio with astrophysical priors) used here.","marker":"[8]"},{"why":"Supplies the unbinned maximum likelihood formalism that the UML search applies.","marker":"[11]"},{"why":"Extends that formalism to transient searches, including background estimation used for the UML p-values.","marker":"[12]"},{"why":"Defines the HEALPix pixelization scheme used to apply GW skymaps as spatial penalties.","marker":"[13]"}],"fun_headline_variants":["IceCube finds no neutrinos in 2,500 O4 GW alerts","No neutrino coincidences in O4 gravitational-wave alerts","O4 GW alerts yield no neutrino signals for IceCube","IceCube’s O4 search: no neutrinos from GW alerts"],"cache_read_input_tokens":14208,"weakest_assumption_plain":"The entire null result rests on the scrambled-neutrino background: if scrambling the GFU track sample against the O4 gravitational-wave skymaps does not faithfully reproduce the spatial and temporal correlations of real background, then the quoted $p$-values, including the 0.15 population result, could be biased.","fun_headline_variants_meta":{"raw":{"variants":["IceCube finds no neutrinos in 2,500 O4 GW alerts","No neutrino coincidences in O4 gravitational-wave alerts","O4 GW alerts yield no neutrino signals for IceCube","IceCube’s O4 search: no neutrinos from GW alerts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000185,"raw_usage":{"total_tokens":1334,"prompt_tokens":973,"completion_tokens":361,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":589,"completion_tokens_details":{"reasoning_tokens":288}},"tokens_in":589,"tokens_out":361,"duration_ms":4672,"temperature":1.0,"reasoning_tokens":288,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:27:36.933597+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A confirmed gravitational-wave alert with a coincident high-energy neutrino whose joint $p$-value falls far below the background expectation, or a full-O4 population test with $p$-value well below 0.01, would refute the no-emission conclusion. A simpler technical check is a signal-injection run that adds fake neutrino events at known positions to the GFU stream and tests whether both pipelines recover them at the claimed false-alarm rate.","supporting_citations":[{"cited_title":"Abbasiet al., ApJ 944 no","cited_arxiv_id":null,"evidence_quote":"Established the previous O3 follow-up with the UML method that this search extends and shares."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reported the earlier O1/O2 LLAMA search, providing the precedent and baseline for the Bayesian pipeline."},{"cited_title":"Abbasiet al., ApJ 910 no","cited_arxiv_id":null,"evidence_quote":"Defines the Gamma-ray Follow-Up (GFU) event stream whose track candidates form the search dataset."},{"cited_title":"Baretet al.,Astropart","cited_arxiv_id":null,"evidence_quote":"Provides the GRB emission model on which the ±500 s coincidence time window is based."},{"cited_title":"Bartos, D","cited_arxiv_id":null,"evidence_quote":"Introduces the LLAMA method (Bayesian odds ratio with astrophysical priors) used here."},{"cited_title":"Braun, J","cited_arxiv_id":null,"evidence_quote":"Supplies the unbinned maximum likelihood formalism that the UML search applies."},{"cited_title":"Braun, M","cited_arxiv_id":null,"evidence_quote":"Extends that formalism to transient searches, including background estimation used for the UML p-values."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the HEALPix pixelization scheme used to apply GW skymaps as spatial penalties."}],"review_version":1}