{"id":"a0b74d7d-0b2e-4bf1-bbe8-65856c0828ff","arxiv_id":"2412.00159","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Four seasons of TAIGA-HiSCORE data show no nanosecond optical transients, setting an upper limit of about 1e-3 events per steradian per hour.","lead":"A wide-field Cherenkov detector in Siberia spent four winter seasons watching for optical flashes that last just ten nanoseconds. No astrophysical flash was found, giving an upper limit of about one event per thousand steradians per hour.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 1e-3 ster^-1 h^-1 upper limit assumes ~100% detection efficiency for plane-wave optical transients, but the efficiency is never measured; CALIPSO's anomalous reconstructed altitude (30–150 km vs 700 km) shows the proxy is unreliable, so the bound may be weaker by an unknown factor.","rationale":"I read the paper as a careful null-result search: the authors processed four winter seasons, used CALIPSO as a stand-in for distant point sources, and found no reliable astrophysical candidates. That central claim is credible: the filters are visual and quantitative, the CALIPSO events are clearly separated, and the three non-CALIPSO events that survived the final cuts are convincingly attributed to EAS wings. The load-bearing weakness is not the null result but the normalization of the upper limit. The paper derives ~1e-3 events/ster/h as 1/exposure with zero candidates, but this is only valid if detection efficiency for true plane-wave transients is near unity. The CALIPSO calibration cannot establish this because CALIPSO's reconstructed source altitude is 30–150 km rather than 700 km, a discrepancy the authors explicitly say is not understood. The filters are tuned to keep CALIPSO events, but the pass fraction for CALIPSO is not reported, and no simulation or injected-signal test is presented for astrophysical plane-wave fronts. Especially for the Z filter, a plane wave's fitted source altitude is poorly constrained by timing noise; a nontrivial fraction of real events could reconstruct below the 20–50 km thresholds and be lost. The reader's weakest assumption correctly identified this same issue, so my assessment agrees with theirs. The recommended verdict is unchanged: CONDITIONAL is appropriate because the null result is likely robust, but the numerical upper limit should be regarded as provisional until efficiency and confidence level are addressed. My proposed injection test is the concrete step that would settle the matter: it directly measures the efficiency by adding simulated plane-wave pulses to real data and running the exact pipeline. If efficiency comes out near unity and a 95% CL is stated, the paper can be promoted; if not, the bound must be raised by the inverse efficiency and the appropriate Poisson factor.","tokens_in":23858,"tokens_out":5759,"duration_ms":53601,"concrete_test":"Inject simulated plane-wave optical transients into the recorded events of each of the four seasons: take real nights with the same station configurations, add artificial pulses of 10 ns duration and amplitudes just above the 3000 quanta/m²/10 ns threshold uniformly illuminating all active stations, with timing jitter sampled from the post-calibration residual distribution (Figure 5, ~1–2 ns plus a tail to ~6 ns). Run the complete reconstruction chain (plane and spherical front fits, bad-station removal, EventSquare, Z filters) using the exact per-period thresholds from Section 4. Compute the fraction of injected events that survive. Also compute the 95% Poisson upper limit 2.996/1273 and the efficiency-corrected limit 2.996/(1273 × efficiency). If the efficiency is >0.9, the stated ~1e-3 bound is supported; if it is substantially lower, the bound must be revised upward accordingly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the upper limit ~1e-3 ster^-1 h^-1 in Section 5, derived as 1/(1273 ster·h) because zero candidates passed. This assumes every transient above the ~3000 quanta/m²/10 ns threshold triggers the array and survives the offline EventSquare and altitude-Z filters. That efficiency is never measured. The filters are calibrated on CALIPSO satellite events (Section 4), but Section 3.3.2 reports the reconstructed Z of CALIPSO is 30–150 km instead of its true 700 km, and states 'The reason is not well understood.' The Z thresholds (20–50 km) and EventSquare thresholds (0.25–0.6) are chosen to bracket CALIPSO's distributions, and the paper does not report the fraction of CALIPSO events that pass, nor any efficiency for plane-wave sources. For a true astrophysical transient the light front is planar, and the spherical-front Z fit is then dominated by timing noise; the reconstructed Z may fall below threshold, or station threshold and atmospheric effects may reduce EventSquare. The paper's own argument that true events should yield the 1000 km cap is an unvalidated assumption. Consequently, the quoted 1e-3 bound is a raw inverse-exposure number, not a confidence limit (at 95% CL it would be 2.996/1273 = 2.35e-3 even at 100% efficiency), and could be further diluted by a real efficiency below unity. The null result itself is plausible, but its normalization is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a four-season (2018–2022) search for nanosecond optical transients using the TAIGA-HiSCORE Cherenkov array in a parasitic mode. The authors describe amplitude and timing calibrations, including drone-LED and CALIPSO satellite events as proxies for distant point sources, and define two new filters: the 'EventSquare' (convex hull of triggered stations) and the reconstructed source altitude Z from a spherical-front fit. Applying these filters to the 2019–2022 seasons and combining with the previously published 2018–2019 result, they find no astrophysical candidates and quote an upper bound on the event flux of about 1e-3 events/ster/h from 1273 steradian-hours of exposure.","tokens_in":24161,"tokens_out":3965,"duration_ms":34694,"significance":"If established rigorously, this bound would be a useful constraint on rare nanosecond optical transients, relevant to primordial black hole evaporation models, stellar-mass black hole accretion flares, and optical technosignature searches. The strength of the paper is the demonstration that a large-FOV, non-imaging Cherenkov array can search a wide sky area for very short pulses in a companion mode, with careful timing calibration verified by drone trajectory reconstruction and CALIPSO astrometric residuals of about 0.05 degrees. The paper also candidly reports an unexplained discrepancy in the CALIPSO altitude reconstruction, which is directly relevant to the validity of its own calibrated filters.","major_comments":[{"comment":"The upper limit of ~1e-3 ster^-1 h^-1 is computed as 1/1273, i.e., one candidate-free exposure divided by the exposure, with no stated confidence level and no division by detection efficiency. At 95% confidence and 100% efficiency the limit would be 2.996/1273 ≈ 2.35e-3, and any efficiency below unity weakens it further. The authors should present a proper Poisson upper limit with a confidence level and a clearly defined efficiency factor.","section":"§5"},{"comment":"The EventSquare and Z thresholds are tuned by hand from CALIPSO events (Zthr = 20–50 km, EventSquarethr = 0.25–0.6), but the paper does not report the fraction of CALIPSO events that pass the final selection, nor does it measure the detection efficiency for plane-wave sources. Because zero candidates divided by exposure only yields an upper limit if the selection efficiency is known, the central quantitative claim requires either an efficiency measurement from injected artificial transients or an explicit statement that the thresholds are known to pass all events of interest.","section":"§4"},{"comment":"The paper states that CALIPSO's reconstructed Z is 30–150 km instead of its true 700 km and that 'The reason is not well understood.' Since the Z filter is calibrated on these same CALIPSO events, the physical basis of the filter is not established. The argument that true astrophysical transients, having planar wavefronts, would yield Z = 1000 km is plausible but unvalidated; a synthetic plane-wave injection test would be needed to confirm that such events pass the Z threshold with known efficiency.","section":"§3.3.2"},{"comment":"The total exposure of 1273 ster·h sums 476 h from the 2018–2019 season, which was analyzed with the earlier filter set described in [5], and the three seasons analyzed with the new EventSquare and Z filters. Unless the 2018–2019 data were reprocessed with the same final filters, the combined candidate count does not correspond to a single selection efficiency, and the simple division by total exposure is not self-consistent. The authors should either re-process the first season with the new filters or conservatively quote the limit using only the seasons analyzed identically.","section":"§5"},{"comment":"The FOV is assumed to be Ω ≈ 0.6 ster, but the text states that the FOV has no sharp boundaries and depends on flash brightness. The exposure of 1273 ster·h therefore carries a systematic uncertainty that is not propagated into the quoted bound. At minimum, the authors should provide an explicit uncertainty on Ω and show its effect on the final limit.","section":"§1 and §5"}],"minor_comments":[{"comment":"The text introduces 'event square' but the parameter is inconsistently spelled 'EventSquate' in the text and 'EventSquare' in plots and mathematics; unify the spelling.","section":"§3.3.1"},{"comment":"The word 'seazon' appears repeatedly (e.g., Table 1 caption, '2021–2022 data seazon'); replace with 'season'.","section":"Table 1 and throughout"},{"comment":"The sentence 'If the remaining stations are two times as many as 3 stations for a plane front and 4 stations for a spherical front' is unclear; state the minimum number of remaining stations explicitly (e.g., at least six for plane, eight for spherical).","section":"§3.2"},{"comment":"The sentence 'There are less than 0.1% of EAS events' should read 'Fewer than 0.1% of EAS events'.","section":"§3.2"},{"comment":"The paper says the effective source altitude is detected at different altitudes in different seasons (45 km vs 90 km) but does not discuss whether the Z thresholds could bias the search differently across periods; a short comment on the stability of the Z distribution would be helpful.","section":"§3.3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is an interesting null result from a non-imaging Cherenkov array. The main issues are statistical rigor and a self-consistent treatment of the 2018–2019 season. I would encourage the editor to request a revised version that replaces the 1/exposure number with a proper confidence limit and either measures or convincingly bounds the detection efficiency for plane-wave transients."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is the best wide-field nanosecond optical transient search to date, and the null result is probably real. But the quantitative limit in the abstract is not a statistically defined bound, and I would not take 1e-3 per ster/h at face value. The reader's conditional verdict is fair; the stress-test note correctly identifies the weak underbelly.\n\nWhat is new: three additional seasons (2019–2022), a much better calibration apparatus—drone LED timing residuals near 1 ns—and two new discrimination parameters (EventSquare and reconstructed source altitude Z). The paper is careful in describing the calibration iterations and in presenting raw distributions, including the uncomfortable fact that CALIPSO events reconstruct to 30–150 km rather than the true 700 km, with an honest \"reason is not well understood.\" That transparency is real credit. The final filters are tuned on CALIPSO events, but the pre-filter that identifies CALIPSO is independent of the Z/EventSquare filters, so the circularity burden is low; I agree with the reader's score of 1 there.\n\nThe soft spots sit exactly at the main claim. The upper limit is computed as 1/(1273 ster·h) because zero candidates passed. No confidence level is stated; at 95% CL the limit is ~2.4e-3 even at 100% efficiency. More importantly, there is no measurement of detection efficiency for actual astrophysical transients. CALIPSO is a moving satellite with a narrow beam; a true transient is a plane wave. The paper assumes plane-wave events would reach Z=1000 km (the cap) and pass EventSquare, but that is not demonstrated. The CALIPSO Z mismatch shows their spherical-front fit is affected by something—scattering or beam structure—that may not apply to plane waves, but the acceptance could be lower than assumed in either case. No injected signals or Monte Carlo are used to measure efficiency. So the correct reading is: no candidates found, and the order-of-magnitude bound is plausible, but the normalization carries an unknown efficiency factor and a missing confidence level. These are addressable in revision: run signal injections, quote a Poisson limit with CL, and report what fraction of CALIPSO events pass the final selection.\n\nWho is this for? Anyone working on fast optical transients, SETI, or PBH evaporation constraints. It deserves a serious referee, not a desk reject, because the instrument is unique and the null result matters even if the quoted limit is too strong. I'd cite it as an upper bound with a caveat.\n\nRecommendation: send to review with a request for efficiency systematics and a proper confidence interval.","headline":"A credible null result with a real search, but the quoted rate limit is a raw inverse-exposure number, not a statistical bound.","tokens_in":25306,"tokens_out":2081,"would_cite":true,"duration_ms":19972,"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":"After four winter seasons and 1273 steradian-hours of wide-field sky monitoring, the TAIGA-HiSCORE Cherenkov array finds no astrophysical nanosecond optical flashes, setting an upper bound of about one event per thousand steradian-hours.","keywords":["nanosecond optical transients","Cherenkov array","wide-field search","upper limit","primordial black holes","CALIPSO calibration","event square filter","air shower background"],"falsifier":"Fly or place a pulsed point-like light source at a known distance, for example a calibrated laser on a satellite or high-altitude balloon at known range, and run the full EventSquare and Z pipeline: if its detection efficiency differs substantially from the CALIPSO-derived efficiency, the quoted upper limit would need rescaling by that efficiency factor.","tokens_in":23662,"feed_emoji":"🔭","tokens_out":7158,"duration_ms":60239,"temperature":0.7,"pith_summary":"Using the TAIGA-HiSCORE Cherenkov array, a cosmic-ray instrument with a roughly one-steradian field of view, the authors search for nanosecond optical flashes from astrophysical sources. Across four winter observing seasons (2018–2022) they accumulate 1273 steradian-hours of exposure and find no reliable astrophysical candidate. This yields an upper bound of roughly $10^{-3}$ events per steradian per hour for flashes of $10\\,\\mathrm{ns}$ or longer with flux above about $3000$ quanta/m$^2$/10$\\,\\mathrm{ns}$. The result matters because nanosecond optical transients are predicted from evaporating primordial black holes, magnetic reconnection near isolated black holes, and hypothetical extraterrestrial lasers, and no previous optical search combined such short timescales with such a wide field of view.","feed_headline":"Four-year sky scan finds zero nanosecond flashes","feed_subtitle":"Upper limit: fewer than one flash per thousand steradian-hours at fluxes above 3,000 quanta per square meter per 10 ns.","key_machinery":"The load-bearing analysis is a two-stage filter tuned on satellite calibration events. Stage one removes air-shower background with EventSquare, the normalized area of the convex hull of triggered stations; stage two keeps only events whose spherical-front source altitude Z falls above a season-dependent threshold near 20–50 km. Calibration is carried out with an iterative timing correction using air-shower fronts and drone LED pulses, plus amplitude equalization based on power-law spectra, so that a distant point source yields uniform illumination and a plane wavefront. The paper explicitly treats the CALIPSO satellite lidar transits as a proxy for a distant point-source flash, and notes that their reconstructed altitudes of 30–150 km are far below the satellite's true 700 km, a discrepancy attributed tentatively to atmospheric scattering and described as not well understood.","core_discovery":"The paper's central claim is that the rate of nanosecond astrophysical optical transients above the detector threshold is below one per thousand steradian-hours per hour. After calibrating station amplitudes and trigger times, the authors filter events by two new signatures: EventSquare, the area of the smallest convex polygon covering all triggered stations, and Z, the source altitude reconstructed from a spherical light-front fit. Satellite lidar pulses passing through the array set the filter thresholds; three residual events from the 2019–2020 season survived the cuts, and all three are identified as air-shower wings, not point-source flashes. With no true candidates in 1273 steradian-hours of exposure, the upper limit follows as $1 \\times 10^{-3}$ events per steradian per hour, corresponding to an energy flux density of at least $10^{-4}$ erg/s/cm$^2$ in 10 ns pulses.","pith_inferences":["Converting the flux upper limit into a bound on primordial black hole evaporation requires assuming an optical energy fraction in the final explosion; the paper does not make that conversion, but the flux limit is available for such models.","The CALIPSO altitude anomaly implies the Z filter's efficiency for real astrophysical transients is uncertain: if atmospheric scattering bends the light fronts of collimated beams more than plane fronts, a real distant flash might be reconstructed at a different effective altitude than a satellite pulse.","A sky survey with several such arrays, or a larger station spacing, could push the rate limit down by another order of magnitude within a few seasons."],"forward_implications":["Future seasons of the same array can continue the search without hardware changes, since transient selection runs on the existing event database.","Any genuine nanosecond flash above 3000 quanta per square meter per 10 ns occurs less often than once per thousand steradian-hours, so detecting one requires either a larger field of view, longer exposure, or a lower threshold.","The EventSquare and Z filters reduce the extensive-air-shower background to nearly zero over the full field of view, not only near the zenith.","A detection in the same instrument could be cross-confirmed by the array's gamma-ray channel, since a primordial-black-hole explosion should also emit multi-TeV gamma rays in the last milliseconds before evaporation."],"supporting_citations":[{"why":"Established the first-season search method, the 0.6 sr FOV estimate, and the prior null result that this paper extends.","marker":"[5]"},{"why":"Documents the design and construction of the HiSCORE stations whose trigger amplitudes and times are used here.","marker":"[2]"},{"why":"Supplies the working principle of the array as a detector of Cherenkov light fronts.","marker":"[6]"},{"why":"Provides the amplitude-calibration technique adapted here to equalize station responses.","marker":"[59]"}],"fun_headline_variants":["Four-year search finds no nanosecond optical bursts","Zero nanosecond flashes in four-year sky scan","TAIGA-HiSCORE: no flashes in 1273 sr-h exposure","Four years, no flashes: upper limit on nanosecond optical transients"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The search assumes that CALIPSO satellite pulses are a faithful proxy for astrophysical nanosecond transients, so thresholds derived from CALIPSO distributions leave near-unity acceptance for the searched events; the paper itself reports that CALIPSO's reconstructed altitude is 30–150 km instead of the true 700 km and says the reason is not well understood.","fun_headline_variants_meta":{"raw":{"variants":["Four-year search finds no nanosecond optical bursts","Zero nanosecond flashes in four-year sky scan","TAIGA-HiSCORE: no flashes in 1273 sr-h exposure","Four years, no flashes: upper limit on nanosecond optical transients"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001247,"raw_usage":{"total_tokens":5136,"prompt_tokens":988,"completion_tokens":4148,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":604,"completion_tokens_details":{"reasoning_tokens":4085}},"tokens_in":604,"tokens_out":4148,"duration_ms":21978,"temperature":1.0,"reasoning_tokens":4085,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T06:00:22.167207+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fly or place a pulsed point-like light source at a known distance, for example a calibrated laser on a satellite or high-altitude balloon at known range, and run the full EventSquare and Z pipeline: if its detection efficiency differs substantially from the CALIPSO-derived efficiency, the quoted upper limit would need rescaling by that efficiency factor.","supporting_citations":[{"cited_title":"Angles θ >60◦ were outside the FOV of the HiSCORE array","cited_arxiv_id":null,"evidence_quote":"Established the first-season search method, the 0.6 sr FOV estimate, and the prior null result that this paper extends."},{"cited_title":"We distinguish four categories of problems","cited_arxiv_id":null,"evidence_quote":"Documents the design and construction of the HiSCORE stations whose trigger amplitudes and times are used here."},{"cited_title":"Such a source may be the evaporation of the remnant of a primordial black hole (PBH)","cited_arxiv_id":null,"evidence_quote":"Supplies the working principle of the array as a detector of Cherenkov light fronts."}],"review_version":1}