{"id":"3880b047-9638-4986-9664-ea8a9b9ae4a3","arxiv_id":"1908.05990","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"AMON's joint-likelihood analysis of HAWC and IceCube subthreshold data is ready to issue coincidence alerts at a false alarm rate of one per year.","lead":"This paper describes a new real-time alert system that combines subthreshold gamma-ray detections from HAWC with neutrino events from IceCube to flag possible multimessenger sources. A generalist might read it to see how observatory networks are turning archival searches into live follow-up triggers for the next era of multimessenger astronomy.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The abstract's 'AMON is ready to produce real-time coincidence alerts' is contradicted by Sec. 6, which describes the real-time search as future work; no online test, latency measurement, or alert distribution is presented.","rationale":"The reader's RA-scrambling concern is legitimate but secondary: the long-baseline exposure argument makes RA-uniformity plausible, and the FAR can be validated by time-shift tests. The more directly load-bearing issue is that the paper's own conclusions describe the real-time search as future work, while the abstract asserts operational readiness. The body provides no evidence of an online system, no measured latency, and no alert distribution, so the central claim as stated is unsupported by the presented material. This does not invalidate the underlying coincidence method; it means the paper should be accepted conditionally, with the readiness claim either demonstrated or softened to 'designed for' or 'ready pending online validation.' The reader's CONDITIONAL verdict therefore remains appropriate, but for a different, more immediate reason than the one emphasized in the reader's rationale.","tokens_in":4781,"tokens_out":18988,"duration_ms":207112,"concrete_test":"Replay the full two-year archival HAWC/IceCube dataset through the AMON real-time software in blind mode, issuing test GCN notices for every coincidence above the FAR=1/yr threshold and logging timestamps at each stage (hotspot receipt, neutrino matching, statistic computation, alert transmission). If the median end-to-end delay exceeds six hours, or any alert is missed or requires manual intervention, the 'ready' claim is not supported; otherwise the abstract is verified and only the FAR caveat remains.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 6 states: \"This method will serve as a base for a real-time search for multimessenger sources of neutrinos... The search would consist of sending alerts...\" That is future tense, whereas the abstract claims readiness to distribute alerts with a median delay of six hours. The body reports only an offline archival analysis of 2016-2017 data; it contains no description of an operating online pipeline, no end-to-end latency measurement, no automated GCN transmission, and no monitoring of livetime or data quality in real time. The six-hour median delay is \"anticipated,\" not measured. Therefore the strongest claim—that AMON is ready to issue operational alerts—rests on an asserted capability the paper does not demonstrate. Even if the FAR calibration is correct, a real-time system requires validated data streams, alert generation, and follow-up distribution; none are evidenced. This is an internal inconsistency, not a consensus disagreement, and it is precisely the kind of missing support that should be flagged.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an AMON joint-likelihood analysis that searches for coincidences between HAWC 'hotspot' gamma-ray excesses observed during a transit and IceCube through-going muon neutrino events. The selection requires temporal overlap with the HAWC transit and an angular separation below 3.5 degrees; a Fisher-style ranking statistic combines the spatial, spectral, and temporal p-values. The false alarm rate is calibrated with right-ascension-scrambled data corresponding to about 729 years, and a threshold ranking statistic of 7.3 gives one false alarm per year. Applied to 1.85 years of archival livetime from 2016-2017, the analysis finds one coincidence with ranking statistic 7.34 and one close to threshold at 7.27. The conclusions describe the method as a basis for a future real-time search, while the abstract states that AMON is ready to produce real-time alerts with a median delay of six hours.","tokens_in":4907,"tokens_out":5247,"duration_ms":54612,"significance":"The analysis addresses a timely and important problem: finding sub-threshold multimessenger coincidences between TeV gamma rays and neutrinos. Its strengths are clear: the false-alarm calibration uses a large scrambled background sample equivalent to 729 years, the ranking statistic combines relevant information through a standard Fisher-type method, and the archival search yields two interesting candidate events whose sky maps are shown. If the claimed real-time alert system were actually operational, this would be a valuable contribution to the multimessenger follow-up ecosystem. However, the body of the paper establishes only an offline archival search, not an operating alert pipeline. The unblinded result is also a weak validation because the threshold is chosen to enforce the false alarm rate rather than used as an independent prediction. With appropriate revisions that scope the claims to the offline analysis and quantify the calibration uncertainties, the method itself is a useful template for subthreshold coincidence searches.","major_comments":[{"comment":"The abstract states 'AMON is ready to produce real-time coincidence alerts' and gives a 'median anticipated delay of six hours,' but Section 6 describes the real-time search only in the future tense: 'This method will serve as a base for a real-time search for multimessenger sources of neutrinos... The search would consist of sending alerts...' The paper presents no running online pipeline, no end-to-end latency measurement, no alert-distribution test through GCN, and no live data-quality monitoring. The six-hour delay is labeled 'anticipated,' not measured. This discrepancy is load-bearing because the title and abstract promise an operational alert system, whereas the analysis demonstrates only an offline archival search. Please either present evidence of the operational system or revise the abstract and conclusions to state that the real-time implementation is planned.","section":"Abstract; Section 6"},{"comment":"The false alarm rate curve in Fig. 1 is shown without statistical uncertainties. With about 729 years of scrambled data, the Poisson uncertainty on the FAR at one per year is roughly 4%, but in the high-ranking-statistic tail the effective number of trials is much smaller and the uncertainty is correspondingly larger. Because the alert threshold of 7.3 is derived from this curve, the curve should include confidence bands, and the text should state the number of independent scrambles and how trials are handled. Without this, the one-per-year FAR and the threshold are not fully quantified.","section":"Section 4, Fig. 1"},{"comment":"The RA-scrambling procedure assumes that each detector's acceptance is effectively uniform in right ascension. The text says 'The scrambling consisted on randomly permuting the right ascension information of the events. Parameters that are declination dependent were kept together with their respective declination.' If HAWC's transit pattern or IceCube's seasonal or Earth-blocking effects create RA-dependent acceptance, the scrambled background could bias the false alarm rate. Please state explicitly why the RA-uniform assumption holds, and ideally validate the calibration by comparing with an alternative scrambling (for example, a full time-shift or time-randomization).","section":"Section 4"},{"comment":"The comparison between the unblinded ranking-statistic distribution and the scrambled background in Fig. 2 is only qualitative. The observation of one event above the threshold in 1.85 years is consistent with the calibrated rate of one per year, but because the threshold was chosen to enforce that rate, this single event does not independently validate the analysis or indicate a signal. The paper should include a quantitative goodness-of-fit test for the distribution and should report the expected number of events above threshold for the actual livetime as an explicit check.","section":"Section 5, Fig. 2"}],"minor_comments":[{"comment":"The Introduction contains a typo: 'The purpose of the this analysis' should read 'The purpose of this analysis.'","section":"Section 1"},{"comment":"The IceCube paragraph says 'The parameters consists of' and should be 'The parameters consist of.' In addition, 'signal acceptance' is listed as an input but is never defined; please define it.","section":"Section 2"},{"comment":"The text refers to the 'Galactic Coordinates Network [10],' but the referenced network is the GRB Coordinates Network (GCN); please correct the name.","section":"Section 4"},{"comment":"The definition of the ranking statistic should state explicitly the null distributions and the number of degrees of freedom, including the formula for p_cluster, which is described only in words.","section":"Section 3, Eq. (3.1)"},{"comment":"Reference [6] is incomplete: 'arXiv:180' does not identify the paper. Please provide the full arXiv identifier or journal reference.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"This is a short proceedings contribution, and the underlying offline analysis appears sound. The main problem is that the abstract and conclusions overstate readiness: the paper describes a planned real-time system, not a demonstrated one. This is fixable in revision by aligning the claims with what is shown. The FAR calibration would also benefit from error bars and a validation of the RA-scrambling assumption. I recommend major revision rather than rejection because the method and the archival results are useful and the central claims can be corrected without new data."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this with an eye to what it actually demonstrates. The novel bit is the AMON joint-likelihood search pairing HAWC hotspot transits with IceCube through-going tracks, using Fisher's method plus a spatial overlap likelihood. That is an incremental extension within AMON, not a new framework, but it is a real combination that has not been published before. The paper does the calibration work properly: 729 years of RA-scrambled data to set the FAR, a pre-defined threshold of 1/yr, then an unblinded search of 1.85 years livetime with one event at 7.34 and a second at 7.27. The authors do not over-interpret; they say the event is consistent with background. Credit where due.\n\nThe main soft spot is the abstract/body mismatch. The abstract says AMON is ready to produce real-time alerts with a median six-hour delay, but Sec. 6 says the method \"will serve as a base for a real-time search\" and \"the search would consist of sending alerts.\" The body contains no online pipeline, no latency measurement, no GCN transmission, no data-quality monitoring. The six-hour number is anticipated, not measured. That is an internal inconsistency, and it is load-bearing for the abstract's strongest claim. The offline analysis is fine; the readiness claim is not.\n\nOther weaknesses are minor but real. There are no uncertainties on the FAR curve, and the comparison of the unblinded ranking distribution with the scrambled expectation (Fig. 2) is only visual; a KS test or a p-value would have made the consistency check quantitative. The RA-scrambling assumption is standard for IceCube point-source analyses, and keeping declination-dependent parameters together mitigates the main worry, but the paper does not show any validation that the scrambled ensemble reproduces the data's null distribution (e.g., by declination band or temporal correlation). I would expect that as a check, not a blocking issue. Also, as a proceedings, the reference list is appropriate, with AMON and detector papers cited; no obvious citation problem.\n\nWho this is for: people building or using multimessenger coincidence frameworks; follow-up network users who need to know what an AMON alert means. It deserves referee time if pushed to a journal, with the explicit request to either demonstrate the online system or soften the abstract. I would not cite it as evidence that real-time alerts are flowing, but I would cite the calibrated FAR method and the null result.","headline":"Useful AMON methods proceedings with a clean FAR calibration and an honest null result; the abstract's real-time readiness claim is ahead of the body's future-tense outlook.","tokens_in":5489,"tokens_out":2415,"would_cite":false,"duration_ms":24262,"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":"This paper reports that AMON is ready to issue real-time coincidence alerts from HAWC gamma-ray hotspots and IceCube neutrinos, with a median delay of six hours, localization near 0.2 degrees, and a false-alarm rate of one per year.","keywords":["multimessenger astrophysics","AMON","HAWC","IceCube","gamma-ray neutrino coincidence","false alarm rate calibration","subthreshold transient search"],"falsifier":"Recompute the false-alarm-rate curve using a background ensemble that preserves each detector's real sky-exposure pattern, for example by scrambling event times rather than coordinates, and compare the ranking-statistic value that gives one false alarm per year. If that threshold moves away from 7.3 by more than the statistical uncertainty of the scrambled sample, the alert calibration and its quoted rate are wrong.","tokens_in":4534,"feed_emoji":"🔭","tokens_out":10079,"duration_ms":89162,"temperature":0.7,"pith_summary":"This paper reports that the Astrophysical Multimessenger Observatory Network (AMON) is ready to issue real-time coincidence alerts pairing HAWC TeV gamma-ray hotspots with IceCube astrophysical neutrino events. The goal is to catch transient or variable sources, such as blazars and radio galaxies, that produce both gamma rays and neutrinos, something the recent gravitational-wave and flaring-blazar multimessenger detections suggest should happen more often. To calibrate the alerts, the authors scramble right ascension in two years of archival data and find that a ranking statistic of 7.3 corresponds to one false alarm per year. In the unblinded 1.85 years of livetime, one coincidence exceeds that threshold, consistent with the expected background, and two more sit just below it. If correct, the system would begin sending such alerts to follow-up observatories with a median delay of six hours and angular resolution near 0.2 degrees.","feed_headline":"AMON ready to issue HAWC-IceCube coincidence alerts in real time","feed_subtitle":"Alerts arrive within six hours, localize to about 0.2 degrees, and trigger one false alarm per year","key_machinery":"The machinery is a joint-likelihood ranking statistic built on Fisher's method. For a candidate coincidence, the statistic is $\\chi^2_{6+2n_\\nu} = -2\\ln[p_\\lambda p_{\\mathrm{HAWC}} p_{\\mathrm{cluster}} \\prod_i p_{\\mathrm{IC},i}]$, where $p_\\lambda$ comes from a maximum-likelihood fit for the best overlap position of the HAWC hotspot and IceCube event uncertainties on the sphere, $p_{\\mathrm{HAWC}}$ is the probability that the hotspot is a background fluctuation, $p_{\\mathrm{cluster}}$ is the probability of seeing the observed number of neutrinos within the HAWC transit time from background, and each $p_{\\mathrm{IC},i}$ is the background probability of an IceCube track event. The $\\chi^2$ value is converted to a p-value with $6+2n_\\nu$ degrees of freedom and then to $\\chi'{}^2 = -\\log p$, the number used to rank coincidences. The false-alarm-rate curve built from right-ascension-scrambled data converts this ranking into an alert threshold.","core_discovery":"The central working claim is that a joint-likelihood search combining subthreshold HAWC hotspots and IceCube through-going muon-neutrino track events can be run in real time as an AMON alert stream. The analysis ranks coincidences by Fisher's method, combining a spatial-overlap likelihood and background probabilities for the gamma-ray excess and each neutrino, with degrees of freedom adjusted for the number of neutrinos found inside the HAWC transit window and a 3.5-degree search radius. Scrambling right ascension in the archived two-year dataset yields a false-alarm-rate curve; a ranking statistic of 7.3 corresponds to one alert per year. Unblinding 1.85 years of livetime produces one coincidence at 7.34 and a near-miss at 7.27, matching the background expectation. The paper concludes that the method is ready to broadcast alerts with median six-hour delay, roughly 0.2-degree combined localization, and a public alert threshold set at a one-per-year false alarm rate.","pith_inferences":["The paper does not raise the option of requiring multiple neutrinos inside the transit window; doing so with the same ranking statistic would trade alert rate for purity and could be calibrated against the identical scrambled-background ensemble.","If either near-threshold coincidence at ranking 7.34 or 7.27 is later tied to a flaring blazar or radio galaxy, the archival test would retroactively become a multimessenger discovery, because both are consistent with background by the paper's own numbers.","One extension the paper leaves implicit is using the calibrated false-alarm rate as a prior when combining these alerts with gravitational-wave or other multimessenger triggers to prioritize follow-up observations."],"forward_implications":["If the system runs as calibrated, AMON will broadcast a HAWC-plus-IceCube coincidence alert with a median delay of six hours, short enough for a full HAWC transit to pass but still timely for deep follow-up by optical, X-ray, and very-high-energy instruments.","Alerts will carry a combined localization of about 0.2 degrees at 1-sigma for a 2D Gaussian, making them suitable for pointed electromagnetic follow-up.","With a one-per-year false alarm rate, essentially every issued alert should be treated as a real multimessenger candidate worth immediate observation.","The single candidate above threshold in 1.85 years of livetime is consistent with the expected background, so the first alerts are a calibrated stream rather than claimed detections.","The same pipeline can be extended to search for sources of ultra-high-energy cosmic rays by combining neutrino alerts with data from additional observatories."],"supporting_citations":[{"why":"Supplies the HAWC detector response, field of view, and angular resolution that define the gamma-ray hotspot input.","marker":"[3]"},{"why":"Supplies the IceCube instrumentation and track-event selection that define the neutrino input.","marker":"[4]"},{"why":"Describes the AMON framework that hosts the joint-likelihood analysis and alert distribution.","marker":"[5]"},{"why":"Updates the AMON framework that this analysis builds on.","marker":"[6]"},{"why":"Identifies blazars and radio galaxies as the gamma-ray-plus-neutrino source class that this search targets.","marker":"[7]"}],"fun_headline_variants":["AMON sends HAWC+IceCube coincidence alerts in 6 hours","Real-time AMON: 6-hour alerts with 0.2° localization","AMON's joint HAWC-IceCube search: subthreshold to one alert/year","AMON live: HAWC+IceCube subthreshold coincidences, 0.2°, 6 hours"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The false-alarm calibration assumes that randomly shuffling the east-west sky positions of the events, while keeping their north-south positions, produces a faithful picture of random coincidences; if either telescope's sky exposure depends on east-west position, the one-per-year alert threshold could be wrong.","fun_headline_variants_meta":{"raw":{"variants":["AMON sends HAWC+IceCube coincidence alerts in 6 hours","Real-time AMON: 6-hour alerts with 0.2° localization","AMON's joint HAWC-IceCube search: subthreshold to one alert/year","AMON live: HAWC+IceCube subthreshold coincidences, 0.2°, 6 hours"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001166,"raw_usage":{"total_tokens":4853,"prompt_tokens":1004,"completion_tokens":3849,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":620,"completion_tokens_details":{"reasoning_tokens":3757}},"tokens_in":620,"tokens_out":3849,"duration_ms":21423,"temperature":1.0,"reasoning_tokens":3757,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:58:23.060807+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the false-alarm-rate curve using a background ensemble that preserves each detector's real sky-exposure pattern, for example by scrambling event times rather than coordinates, and compare the ranking-statistic value that gives one false alarm per year. If that threshold moves away from 7.3 by more than the statistical uncertainty of the scrambled sample, the alert calibration and its quoted rate are wrong.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the HAWC detector response, field of view, and angular resolution that define the gamma-ray hotspot input."},{"cited_title":"U., et al., Observation of the Crab Nebula with the HAWC Gamma-Ray Observatory, ApJ, 843, 39, 2017","cited_arxiv_id":null,"evidence_quote":"Supplies the IceCube instrumentation and track-event selection that define the neutrino input."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the AMON framework that hosts the joint-likelihood analysis and alert distribution."},{"cited_title":"Smith, et al., The Astrophysical Multimessenger Observatory Network (AMON), Astro","cited_arxiv_id":null,"evidence_quote":"Updates the AMON framework that this analysis builds on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies blazars and radio galaxies as the gamma-ray-plus-neutrino source class that this search targets."}],"review_version":1}