{"id":"d54a0f04-975e-48fd-9f2d-8aa0328025aa","arxiv_id":"1908.06471","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"VERITAS found no gamma-ray or optical emission from 15 contemporaneous bursts of FRB 121102 or from the second repeater FRB 180814.J0422+73, setting new 95% upper limits.","lead":"Two repeating fast radio bursts were watched with the VERITAS gamma-ray telescopes while radio telescopes listened, and no gamma-ray or optical flashes were seen. The new upper limits on high-energy emission from FRB 121102 during 15 radio bursts are among the tightest yet placed.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 3.56-event 95% upper-limit convention in §4.1 is unexplained and non-standard; it directly sets the per-burst and 3.7e-8 combined limits, so the headline number is not independently reproducible without a derivation.","rationale":"The reader's CONDITIONAL verdict is appropriate. My stress-test agrees with the reader's concern about the unexplained statistical convention, but identifies it as the most load-bearing quantitative issue because it directly controls the quoted upper limits. The reader's weakest_assumption focuses on the 10 ms coincidence window; that is an acknowledged scope limitation, since the paper explicitly defers multi-window and delayed searches, so it is not an internal inconsistency. The 3.56 convention, by contrast, is an unacknowledged quantitative choice that prevents independent reproduction of the headline number. If the standard 3.00 limit were used, the combined limit would shift from 3.7e-8 to about 3.1e-8 photons cm^-2 s^-1, making the published limit conservative but not invalidating the null detection. The 10 ms window limitation remains and is already reflected in the CONDITIONAL verdict; my concern does not require changing that verdict. Because the correction would be conservative and the qualitative conclusion stands, I recommend UNCHANGED rather than REJECT.","tokens_in":5941,"tokens_out":15911,"duration_ms":169890,"concrete_test":"Recompute the 95% confidence upper limit for a Poisson mean with zero observed events and background 9e-5 using the standard Neyman construction (or Feldman-Cousins). If the resulting upper limit is 3.00 rather than 3.56, recompute the per-burst flux limit and the combined 15-burst limit from the total 150 ms exposure and total zero counts; compare with the quoted 5.6e-7 and 3.7e-8 photons cm^-2 s^-1. A shift of about 19% would confirm that the headline number needs correction or a derivation, while the null result itself would remain unchanged.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reported null detection itself is not in doubt: with a mean background of 9e-5 events per 10 ms window, observing zero gamma-ray-like events in 15 windows is fully consistent. The load-bearing issue is the numerical conversion of that null into an upper limit. The text states, 'Following [12]... the 95% confidence upper limit to the number of events per burst is 3.56 events.' For a Poisson count with zero observed events and negligible background, the standard classical 95% upper limit is 3.00 (e^-3.00 = 0.05), and Feldman-Cousins gives approximately the same value; 3.56 corresponds to roughly a 97% confidence level. No derivation or systematic-uncertainty justification is given. Because 3.56 enters linearly into the per-burst flux limit (5.6e-7 photons cm^-2 s^-1) and therefore into the combined 15-burst limit (3.7e-8 photons cm^-2 s^-1), the paper's headline quantitative claim is not reproducible as stated. In addition, the combined limit appears to be obtained by dividing the per-burst limit by 15 rather than by applying a single 95% upper limit (about 3.00 events) to the combined 150 ms exposure; these procedures agree only if the total upper limit is also 3.56. The 10 ms coincidence window is a scope limitation that the paper explicitly acknowledges, whereas the 3.56 convention is an unacknowledged quantitative choice. This statistical issue does not change the qualitative null result, but it does change the numerical limit by roughly 20%.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports VERITAS observations of the repeating FRB sources FRB 121102 and FRB 180814.J0422+73, searching for gamma-ray and optical counterparts. The central result is that no gamma-ray-like events passed selection cuts in any of the 15 FRB 121102 bursts that occurred during contemporaneous GBT observations, with an estimated background of only 9e-5 events per 10 ms window. From this null result the authors quote a 95% confidence upper limit of 3.56 events per burst, corresponding to an integral flux limit of 5.6e-7 photons cm^-2 s^-1 above 200 GeV, and a combined 15-burst limit of 3.7e-8 photons cm^-2 s^-1. The paper also gives steady-emission upper limits for both sources and reports a preliminary optical photometry search with no detected candidates. The authors explicitly note that this analysis covers only strictly contemporaneous emission in a fixed 10 ms window and that searches over multiple time windows, delayed emission, and precursor emission are deferred to a future publication.","tokens_in":6229,"tokens_out":3540,"duration_ms":36771,"significance":"If the result stands, this is a useful null measurement from a multi-burst IACT campaign: with negligible expected background, the non-detection of gamma-ray events in 15 radio-burst windows provides the tightest per-burst and combined integral flux upper limits for FRB 121102 above 200 GeV reported by an IACT. The observational null is robust and the paper's description of the background estimate and event selection is clear. The main quantitative claim, however, depends on the unexplained use of 3.56 events as the 95% upper limit for zero observed events, which is not the standard classical or Feldman-Cousins value. Since the 3.56 factor enters directly into the headline flux limits, the numerical result is not reproducible as written and needs correction or a derivation.","major_comments":[{"comment":"The 95% confidence upper limit of 3.56 events per burst for zero observed events is stated without derivation. For a Poisson process with zero observed events and negligible background, the standard classical 95% upper limit is 2.996 events (e^-2.996 = 0.05), and the Feldman-Cousins interval also gives approximately 3.00 events; 3.56 corresponds to roughly a 97% confidence level. Because this number enters linearly into the per-burst flux limit of 5.6e-7 photons cm^-2 s^-1 and into the combined limit, the paper's principal quantitative result is not independently reproducible. Please either derive the 3.56 value explicitly, justify it as a deliberate choice (for example, a confidence level higher than 95%), or replace it with the standard 95% value.","section":"§4.1, Gamma-ray search"},{"comment":"The combined 15-burst upper limit of 3.7e-8 photons cm^-2 s^-1 appears to be obtained simply by dividing the per-burst limit by 15, which implicitly assumes that the same 3.56-event upper limit applies to the combined 150 ms exposure. A more direct calculation would apply a single 95% upper limit (about 3.00 events) to the combined exposure, which would give a limit of approximately 3.1e-8 photons cm^-2 s^-1 for the same effective area, about 20% lower than the quoted value. Please clarify whether the quoted combined limit is intended as a joint 95% confidence limit and, if so, state the statistical combination procedure explicitly.","section":"§4.1, Gamma-ray search"}],"minor_comments":[{"comment":"The sentence 'no gamma-ray like events passed the cuts' should read 'no gamma-ray-like events passed the cuts'.","section":"§4.1, Gamma-ray search"},{"comment":"The abstract states that two FRBs repeat and that VERITAS results are presented, but it does not mention that the burst-search null result applies only to a 10 ms simultaneous window; this limitation is acknowledged later in §4.1 and should be reflected in the abstract if it summarizes the burst-search result.","section":"Abstract and §4.1"},{"comment":"The caption uses 'ZFT' for the Zwicky Transient Factory and 'Large Synoptic Sky Telescope'; both should be corrected to 'ZTF' and 'Large Synoptic Survey Telescope' (now Rubin Observatory/LSST).","section":"Figure 2 caption"},{"comment":"The abbreviation 'ECM' is used in the figure ('VERITAS ECM') without definition; please define it in the caption.","section":"Figure 2 caption"},{"comment":"The phrase 'at signal/noise of > 10' would read more clearly as 'at a signal-to-noise ratio greater than 10'.","section":"§4.2, Optical search"},{"comment":"The phrase 'fewer than 100 bursts detections' should be 'fewer than 100 burst detections'.","section":"§5, Discussion"}],"recommendation":"major_revision","confidential_remarks":"The qualitative null result is sound and the paper is a legitimate IACT observing report, but the unexplained 3.56-event upper-limit convention is load-bearing for the central quantitative claim. The authors should be asked to either derive the number, cite a precise source for it, or switch to the standard 95% Poisson upper limit. This is a fixable technical point, so major revision rather than rejection is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this one before reading: the null result is solid, but the 95% upper limit is not reproducible as stated. The paper reports no gamma-ray events in 15 bursts of FRB 121102 observed simultaneously with GBT, and no steady emission from FRB 180814.J0422+73 in 8.2 hours. That is a genuinely useful observational result — the largest contemporaneous IACT burst sample so far. The background is tiny (9e-5 events per 10 ms), so the zero count is not in doubt. The paper is also honest that this is the simplest analysis only; it defers multi-window, precursor, and delayed searches to a future publication.\n\nThe real soft spot is statistical. The text says, following MAGIC, that the 95% upper limit is 3.56 events per burst for zero observed. That is not the standard Poisson 95% value of 3.00, and no derivation is given. The stress-test note is right: 3.56 enters linearly into the per-burst flux limit and hence into the combined 3.7e-8 limit, so the headline number is not independently verifiable. Also, the combined limit appears to be the per-burst limit divided by 15, which would be correct only if the combined 95% upper limit on total events were also 3.56. Those are real issues for a quantitative claim. They change the limit by roughly 20%, not by an order of magnitude, and they do not affect the qualitative conclusion that no prompt VHE emission was seen.\n\nThe other limitations are stated clearly: the coincidence window is fixed at 10 ms, so longer or delayed emission would be missed; the optical photometry was partially commissioned and not accurately timestamped. That is fine for a status report, but it does mean the optical limits are preliminary.\n\nThe citation pattern looks fine. The VERITAS instrument and analysis references are standard. The comparison to MAGIC is appropriate.\n\nWho is this for? Someone working on FRB multiwavelength campaigns, particularly the repeater population. It is a conference proceedings, so the stakes are lower than a journal paper, but the data are genuinely new and the null is worth having on the record. If this were submitted as a journal article, I would send it to review but insist that the 3.56-event convention be derived or replaced with a standard Poisson/Feldman-Cousins value, and that the combined limit be computed from the total exposure rather than by dividing. As a proceedings, it is acceptable but the upper limit should be cited with caution until the statistical note is clarified.","headline":"A clean null result with the largest contemporaneous IACT burst sample to date, but the 95% upper-limit convention behind the headline number is unexplained and should be fixed before the flux limits are quoted.","tokens_in":6835,"tokens_out":1411,"would_cite":false,"duration_ms":17322,"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":"Coordinated observations of the repeating fast radio burst source FRB 121102 found no gamma-ray emission within 10 milliseconds of any of 15 radio bursts.","keywords":["fast radio bursts","gamma-ray counterparts","VERITAS","imaging atmospheric Cherenkov telescopes","FRB 121102","FRB 180814.J0422+73","upper limits","millisecond transients"],"falsifier":"A direct check of the VERITAS event database for the 15 bursts: if any event survives the same soft cuts within 10 ms of a reported burst time, the central null claim is false. Alternatively, a future repeating-FRB burst detected simultaneously by a radio telescope and VERITAS with a gamma-ray-like event in the 10 ms window would refute the conclusion that these bursts are TeV-silent at this level.","tokens_in":5696,"feed_emoji":"📡","tokens_out":8059,"duration_ms":77615,"temperature":0.7,"pith_summary":"Coordinated observations of FRB 121102, a source of repeating millisecond radio flashes from outside the Milky Way, found no gamma-ray counterpart to any of the 15 radio bursts that occurred during VERITAS exposures. The paper's central result is this null detection: after standard analysis cuts, zero gamma-ray-like events fell inside a 10-millisecond window centered on each burst time. Because the expected background in such a window is only $9\\times10^{-5}$ events, the absence is statistically compatible with no signal, and the data set a 95% confidence upper limit of $5.6\\times10^{-7}$ photons cm$^{-2}$ s$^{-1}$ per burst above 200 GeV, improving to $3.7\\times10^{-8}$ photons cm$^{-2}$ s$^{-1}$ when the 15 bursts are combined. The paper also reports limits on steady very-high-energy emission from both known repeating FRB sources and a preliminary optical search, and argues that Cherenkov telescope arrays are well matched to millisecond-timescale counterpart searches.","feed_headline":"No gamma-ray flash accompanies any of 15 fast radio bursts","feed_subtitle":"VERITAS sets the tightest multi-burst limit yet on very-high-energy emission from FRB 121102.","key_machinery":"The load-bearing mechanism is a strict temporal coincidence search: a 10 ms window centered on each reported radio burst time, with the expected number of background gamma-rays estimated by the ring-background method. At this timescale the array is essentially background-free, since only $9\\times10^{-5}$ background events are expected per window, so even a single event passing the standard soft gamma-ray selection cuts would have constituted a detection. The event selection uses boosted decision trees trained on simulated gamma rays and real cosmic-ray backgrounds, and the same machinery produces the steady-emission upper limits via standard significance skymaps.","core_discovery":"The discovery is an absence of a prompt gamma-ray signal. During 115 minutes of simultaneous VERITAS and Green Bank Telescope observations on 25 November 2017, 17 radio bursts were detected from FRB 121102; the 15 bursts with overlapping VERITAS exposures produced no gamma-ray-like events above 200 GeV within a 10 ms coincidence window, against an expected background of $9\\times10^{-5}$ events per window. The per-burst 95% upper limit is $5.6\\times10^{-7}$ photons cm$^{-2}$ s$^{-1}$, and the combined limit over all 15 bursts is $3.7\\times10^{-8}$ photons cm$^{-2}$ s$^{-1}$, the tightest constraint from a multi-burst imaging Cherenkov campaign to date. No steady emission was detected toward FRB 121102 or FRB 180814.J0422+73, with 95% flux limits above 200-500 GeV. A partially commissioned optical channel found no sub-second flashes, with sensitivity to optical fluences ranging from about 1% of the radio fluence at 10 ms duration up to equality with the radio fluence at 0.1 ms duration.","pith_inferences":["The data can be re-analyzed with wider coincidence windows, such as 50 ms to 1 s, without new observations, since the paper already holds the event lists; such an analysis would test models in which the gamma-ray emission is delayed by propagation or by a post-burst flare.","If the forthcoming CHIME-VERITAS campaign catches a non-repeating FRB in the overlapping field of view, the same background-free method would apply to a much larger burst population, and a single coincident detection would be far more informative than the stacked limit.","The per-burst upper limit of 3.56 events at 95% confidence could in principle be converted into a constraint on the gamma-ray luminosity per burst under an assumed distance and spectrum, an extension the paper leaves implicit."],"forward_implications":["If FRB 121102 emits very-high-energy gamma rays in step with its radio bursts, each burst must carry fewer than about 3.6 detectable gamma rays above 200 GeV, implying a radio-to-gamma-ray energy ratio that rules out simple one-to-one TeV counterparts at this sensitivity.","The combined 15-burst upper limit of $3.7\\times10^{-8}$ photons cm$^{-2}$ s$^{-1}$ is the most restrictive limit yet from a multi-burst IACT campaign and demonstrates that Cherenkov arrays can monitor repeating FRB sources at millisecond timescales with negligible background.","The null result applies only to simultaneous emission within 10 ms; precursor, delayed, or longer-duration gamma-ray emission is not constrained by this analysis and awaits the paper's planned multi-window search.","The optical limits show that the VERITAS optical channel can constrain optical-counterpart fluence to within a factor of about 10 of the best otherwise planned surveys, making IACTs competitive for sub-second optical photometry of FRBs."],"supporting_citations":[{"why":"Supplies the earlier VERITAS observations of FRB 121102 and the archival steady-emission limits that this campaign extends.","marker":"[18]"},{"why":"Sets the multi-window burst-search methodology and the earlier MAGIC multi-burst limits that the VERITAS result is compared with.","marker":"[12]"},{"why":"Reports the discovery of FRB 121102 as a repeating source, making targeted simultaneous observations possible.","marker":"[4]"},{"why":"Reports the second repeating FRB FRB 180814.J0422+73, the target of the VERITAS 2018-2019 observations.","marker":"[5]"},{"why":"Provides the boosted decision tree gamma/hadron separation used for the soft and moderate analysis cuts.","marker":"[19]"},{"why":"Describes the ring-background method used to estimate the expected background in the 10 ms coincidence window.","marker":"[21]"},{"why":"Gives the optical-versus-radio fluence scheme used to express the VERITAS optical sensitivity limits.","marker":"[20]"},{"why":"Describes the CHIME radio system whose continuous northern-sky coverage enables coordinated searches for non-repeating FRBs.","marker":"[25]"}],"fun_headline_variants":["No gamma-ray flash in 15 fast radio bursts, tightest limit yet","15 FRBs, zero gamma-ray flashes: tightest limit yet","VERITAS: no very-high-energy gamma rays from 15 FRBs","Absence of gamma rays in 15 FRBs sets record limit","No gamma-ray counterpart to 15 bursts from repeating FRB"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The limits assume that any gamma-ray counterpart arrives at the same time as the radio burst, within a strict 10-millisecond window; emission that is delayed, that lasts longer than 10 milliseconds, or that precedes the radio burst would escape the search entirely.","fun_headline_variants_meta":{"raw":{"variants":["No gamma-ray flash in 15 fast radio bursts, tightest limit yet","15 FRBs, zero gamma-ray flashes: tightest limit yet","VERITAS: no very-high-energy gamma rays from 15 FRBs","Absence of gamma rays in 15 FRBs sets record limit","No gamma-ray counterpart to 15 bursts from repeating FRB"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001098,"raw_usage":{"total_tokens":4568,"prompt_tokens":914,"completion_tokens":3654,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":530,"completion_tokens_details":{"reasoning_tokens":3560}},"tokens_in":530,"tokens_out":3654,"duration_ms":24777,"temperature":1.0,"reasoning_tokens":3560,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:45:01.840739+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct check of the VERITAS event database for the 15 bursts: if any event survives the same soft cuts within 10 ms of a reported burst time, the central null claim is false. Alternatively, a future repeating-FRB burst detected simultaneously by a radio telescope and VERITAS with a gamma-ray-like event in the 10 ms window would refute the conclusion that these bursts are TeV-silent at this level.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the second repeating FRB FRB 180814.J0422+73, the target of the VERITAS 2018-2019 observations."},{"cited_title":"How else can we detect Fast Radio Bursts?","cited_arxiv_id":"1605.01468","evidence_quote":"Gives the optical-versus-radio fluence scheme used to express the VERITAS optical sensitivity limits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the CHIME radio system whose continuous northern-sky coverage enables coordinated searches for non-repeating FRBs."}],"review_version":1}