{"id":"275e8c74-4fbe-4979-a300-f7337f227809","arxiv_id":"1908.05533","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Baikal-GVD's two independent inter-cluster timing systems agree to about 5 ns and match a laser-based geometric calibration.","lead":"The Baikal-GVD neutrino telescope synchronized its five detector clusters with two independent timing systems, reaching an inter-cluster accuracy of about 5 nanoseconds. A generalist reader should care because merging events across clusters is essential for the telescope to use its full cubic-kilometer volume for neutrino astronomy.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 5 ns absolute-accuracy claim hinges on the laser calibration, but the 18.6 m distance difference and the time conversion are quoted without uncertainties; the PPS cross-check alone establishes only relative agreement between two clocks.","rationale":"I read the paper as a detector-instrumentation report whose central claim is the 5 ns inter-cluster synchronization accuracy. The two supporting measurements are the WR-SSBT PPS comparison and the laser flash delay between clusters 2 and 3. The PPS comparison is a relative cross-check; by itself it cannot establish absolute accuracy, as both clocks could share a common offset. The laser test is therefore the load-bearing absolute validation. The reader's weakest-assumption analysis correctly identifies the unquantified geometry of the laser source as the key risk. I agree with that assessment and add that the ~85 ns expectation is itself approximate and lacks a stated refractive-index conversion, so the 4 ns measured-minus-expected residual has no significance without an uncertainty budget. This does not imply the result is wrong; it is a missing quantity that should be reported. The paper is otherwise internally coherent: the system description is detailed, the two systems are independently operated, and the in-situ operation over several months is credible. The concern is addressable by a short supplementary analysis, so no verdict change beyond the reader's conditional acceptance is needed.","tokens_in":5003,"tokens_out":8260,"duration_ms":83037,"concrete_test":"Request the surveyed coordinates of the laser source and of clusters 2 and 3 from the Baikal-GVD positioning database, together with the stated coordinate uncertainties; propagate them, using the measured wavelength-dependent group velocity of Baikal water, into a 1-sigma uncertainty on the expected 85 ns delay difference. If that uncertainty is less than about 2 ns, the laser test supports the 5 ns claim; if it is larger, the measured-minus-expected residual has low significance and the conclusion should be weakened or the system re-validated with a laser source whose position is surveyed more accurately.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The conclusion that inter-cluster synchronization accuracy does not exceed 5 ns rests on two measurements in Section 3. The PPS comparison (Fig. 3.1) measures the relative agreement between the WR and SSBT time stamps; an RMS of ~5 ns shows that the two systems track each other to 5 ns, but it does not tie the combined timeline to an absolute external reference and therefore cannot by itself support an absolute accuracy claim. The laser measurement between clusters 2 and 3 is the only absolute check: the measured delays (81 ± 3 ns WR, 83 ± 5 ns SSBT) are compared with an expected delay of ~85 ns obtained from an 18.6 m difference in path length. For this comparison to validate a 5 ns accuracy, the expected delay must be known to better than a few nanoseconds. The paper does not report the uncertainty or survey method for the 18.6 m distance difference, nor the refractive index or group velocity used to convert distance to time. A 1 m uncertainty in the distance difference corresponds to roughly 4.5 ns of light-propagation time in water, already comparable to the claimed accuracy; a similar-sized uncertainty in the optical group velocity would have the same effect. The nominal value itself is only approximate: 18.6 m at a typical water group index gives ~83 ns, not 85 ns. With the geometry error unquantified, the agreement between measured and expected delays cannot be distinguished from a chance agreement, so the absolute 5 ns bound is not established by the presented evidence.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes the inter-cluster time synchronization system of the Baikal-GVD neutrino telescope, which consists of two independent systems: the detector-specific SSBT and the widely used White Rabbit (WR). Both systems assign time marks to trigger signals from different clusters, enabling multi-cluster event merging. The authors report in-situ checks: a PPS cross-comparison showing an RMS of ~5 ns when an external rubidium clock is used, and a laser calibration test in which the measured inter-cluster time differences (81 ± 3 ns for WR, 83 ± 5 ns for SSBT) are consistent with an expected ~85 ns delay corresponding to an 18.6 m path-length difference. The paper concludes that the accuracy of inter-cluster synchronization does not exceed 5 ns.","tokens_in":5281,"tokens_out":4415,"duration_ms":40835,"significance":"If the 5 ns claim is robust, it is an important practical result for Baikal-GVD: it demonstrates that the dual-system synchronization architecture meets the requirements imposed by the 200 MHz sampling of the detector, and it gives confidence in multi-cluster event reconstruction. The paper's strength is its use of two independent timing systems and an external rubidium reference, along with an absolute laser-based cross-check. The measurements are internally consistent and the quoted statistical errors are modest. However, the absolute accuracy claim is not fully supported by the evidence as presented: the laser test relies on an expected delay whose systematic uncertainty is not documented, and the PPS test is a relative comparison. The headline conclusion therefore overreaches what the data currently establish.","major_comments":[{"comment":"The expected delay of ~85 ns is derived from an 18.6 m path-length difference, but the manuscript does not report the survey uncertainty on this distance nor the optical group velocity used for the distance-to-time conversion. A 1 m uncertainty in the distance corresponds to roughly 4.5 ns of propagation time in water, and a comparable uncertainty in the group velocity would have the same effect; both are already at the scale of the claimed 5 ns accuracy. As a result, the agreement between the measured delays (81 ± 3 ns WR, 83 ± 5 ns SSBT) and the expected value cannot be distinguished from a chance agreement, and the absolute 5 ns bound is not established by this test.","section":"Section 3 (laser calibration, Figs. 3.2–3.3)"},{"comment":"The PPS measurement with an RMS of ~5 ns quantifies the relative disagreement between the WR and SSBT timestamps; it shows that the two systems track each other but does not tie the combined timeline to an absolute external reference. The conclusion that 'the accuracy of the inter-cluster synchronization doesn't exceed 5 ns' therefore overstates what this test alone can demonstrate. Additionally, the text states that the SSBT time resolution is ~10 ns 'with the possibility of improvement up to 5 ns' while reporting a PPS RMS of ~5 ns; the relationship between these two numbers needs clarification (for example, whether the quoted resolution refers to raw timestamps and the 5 ns RMS is obtained after some calibration).","section":"Section 3 (PPS cross-check, Fig. 3.1)"},{"comment":"The conclusion that the accuracy of the inter-cluster synchronization does not exceed 5 ns is stronger than the evidence presented. The laser test yields statistical uncertainties of 3–5 ns, and the systematic uncertainty in the expected delay is unquantified; the PPS RMS is a relative, not absolute, measure. The claim should be qualified to state what has actually been measured, for example 'the relative agreement between WR and SSBT is characterized by ~5 ns RMS, and the laser test is consistent with expectations within the quoted statistical uncertainties,' unless an explicit error budget for the absolute calibration is added.","section":"Section 5 (Conclusion)"},{"comment":"The manuscript does not report event counts or the data-taking period for the distributions shown in Figs. 3.1 and 3.3. These numbers are needed to assess the statistical robustness of the quoted RMS and mean values and to determine whether the small differences between the WR (81 ± 3 ns) and SSBT (83 ± 5 ns) results are significant. Adding this information, or explaining its absence, would strengthen the quantitative basis of the paper.","section":"Section 3 (Figs. 3.1 and 3.3)"}],"minor_comments":[{"comment":"There is a typo in the description of White Rabbit: 'SPEC (Simple PCIe FMC сarrier) joint' contains a duplicated word and a misspelled word; also 'DIOWRSPEC' should be written as 'WR SPEC DIO board' or 'DIO WR SPEC'.","section":"Section 2"},{"comment":"The sentence 'the clock travel difference is 50 ns per 1 s' is unclear; it should specify whether this is the relative drift rate between the WR and SSBT clocks and how it was measured.","section":"Section 3"},{"comment":"The expected delay of ~85 ns for an 18.6 m path difference is only approximate: using a typical water group index of 1.34–1.35 gives about 82–84 ns. The paper should state the value and uncertainty of the group velocity or refractive index used for this conversion.","section":"Section 3"},{"comment":"An explicit error budget table summarizing the contributions to the synchronization accuracy (trigger time reference, WR and SSBT resolutions, clock stability, fiber calibration, geometry) would greatly improve the transparency of the central claim.","section":"General"},{"comment":"Reference [6] appears to concatenate two conference papers (ICALEPCS 2009 and ISPCS 2018); these should be separate references with distinct URLs/DOIs.","section":"References"},{"comment":"The statement that 'During the period from April to June 2019 there weren’t any failures of any system components' is relevant to reliability but not to the accuracy claim; consider moving it to a reliability section or providing a quantitative context.","section":"Section 2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a short conference proceedings article, and the main quantitative claim appears in both the abstract and the conclusion. The strongest concern is the unquantified systematic uncertainty in the laser-based absolute calibration; if the survey team can provide the geometry uncertainty and the group-velocity value, a major revision with a qualified conclusion would be appropriate. Otherwise, the authors should downgrade the claim from 'accuracy doesn't exceed 5 ns' to a statement about relative agreement and consistency with expectations. The paper is otherwise within the scope of the journal/proceedings and reports useful in-situ data."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis is a short ICRC2019 proceedings paper on the inter-cluster time synchronization for Baikal-GVD: two independent systems, White Rabbit (WR) and SSBT, now deployed on all five clusters. What's actually new is the in-situ calibration data: PPS cross-checks giving ~5 ns RMS when WR uses an external rubidium reference, and a laser test between clusters 2 and 3 giving 81±3 ns (WR) and 83±5 ns (SSBT) against an expected ~85 ns from the 18.6 m geometric path difference.\n\nThe engineering itself is credible. The paper describes the two systems, how they are integrated in the cable caps, the data flow, and the mutual-control scheme. The numbers are internally consistent, and the PPS check is a sensible way to compare two clocks. For a proceedings paper it is a solid, honest report.\n\nThe soft spot is the absolute accuracy claim. The conclusion states that the inter-cluster synchronization accuracy \"doesn't exceed 5 ns.\" The PPS measurement alone establishes only that the two systems track each other to 5 ns; it doesn't anchor the combined timeline to an absolute external reference. That anchor is supposed to be the laser test, but the expected delay of ~85 ns is quoted without any survey uncertainty for the 18.6 m distance difference or the group velocity used. A 1 m distance error corresponds to about 4.5 ns of light-propagation time in water, already comparable to the claimed accuracy. Also, with a typical water group index, 18.6 m gives about 83 ns, not 85 ns; the measured 81/83 ns values actually sit closer to 83. So the agreement is reassuring but not a rigorous validation of a sub-5 ns absolute bound. The paper itself doesn't claim a rigorous bound; the conclusion is a bit loose.\n\nThe citation pattern looks standard—White Rabbit, DESY software, prior Baikal-GVD papers—and the work doesn't hide its own limitations. This is a conference contribution rather than a full instrument paper; the missing details (measurement statistics, survey method, group index) likely belong in a longer paper later.\n\nWho is this for? Anyone working on large underwater neutrino detectors or on time synchronization for distributed detectors. It is a useful data point: the WR+SSBT combination works for Baikal-GVD, and the laser cross-check is a reasonable method. I would send it to peer review for the proceedings; it is competent and worth recording. Just don't treat the 5 ns as a demonstrated absolute bound until the laser-geometry uncertainty is quantified.","headline":"Useful engineering report on Baikal-GVD's inter-cluster timing, with a plausible 5 ns claim that the paper's own cross-checks support only loosely.","tokens_in":6179,"tokens_out":2763,"would_cite":true,"duration_ms":25482,"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":"Baikal-GVD's five operating clusters share one timeline to better than 5 nanoseconds, according to in-situ tests reported in this paper.","keywords":["Baikal-GVD","neutrino telescope","time synchronization","White Rabbit","SSBT","Cherenkov detector","inter-cluster calibration"],"falsifier":"Re-measure the laser-to-cluster distances with independent survey and check whether the 18.6 m difference is accurate to about 1 m; if its uncertainty exceeds roughly 1 m (about 5 ns in water), the 81–83 ns measured delays cannot by themselves substantiate the claimed 5 ns absolute synchronization accuracy.","tokens_in":4809,"feed_emoji":"🔭","tokens_out":8732,"duration_ms":78311,"temperature":0.7,"pith_summary":"Baikal-GVD is a cubic-kilometre neutrino telescope being deployed in Lake Baikal as an array of independent clusters, and this paper is about making those clusters share one clock. A high-energy neutrino can deposit light in more than one cluster, so merging such events requires the cluster timestamps to agree. The paper argues that the inter-cluster synchronization accuracy does not exceed 5 ns, using two independent systems: White Rabbit and the detector-specific SSBT. This matters because 5 ns matches the detector's 200 MHz sampling step, so events spanning cluster boundaries can be reconstructed on a single timeline. If true, the five currently operating clusters—and the nine planned for the first stage—can be treated as one detector for event reconstruction.","feed_headline":"Baikal-GVD clusters stay synchronized to 5 nanoseconds","feed_subtitle":"Two independent time systems let the underwater telescope merge events that cross cluster boundaries.","key_machinery":"The machinery is a pair of independent timestamping networks installed in the cable cap of each cluster. White Rabbit, a fibre-optic synchronization technology, stamps each cluster trigger signal through a DIO board, while SSBT assigns its own time marks using a thermally stabilised rubidium master clock. A per-second pulse generated by White Rabbit is fed into SSBT, letting the two systems measure each other's clock offsets; connecting an external rubidium reference to White Rabbit reduces the PPS timing spread from about 10 ns RMS to about 5 ns RMS. The absolute check is a laser source whose flashes reach clusters 2 and 3 with a known 18.6 m path difference, so the measured inter-cluster delay should be about 85 ns.","core_discovery":"The central claim is that the two independent synchronization systems installed on all five operating clusters of Baikal-GVD keep the cluster trigger timestamps aligned to better than 5 ns. The evidence is an in-situ calibration: a laser source illuminates clusters 2 and 3, whose distances from the laser differ by 18.6 m, corresponding to about 85 ns of light travel time. White Rabbit records $81 \\pm 3$ ns between the two clusters and SSBT records $83 \\pm 5$ ns, both consistent with the geometric expectation. The paper also measures the SSBT fiber-optic delay between the two clusters as $9.32 \\times 10^{-7}$ s using the per-second pulse output of the White Rabbit board. Taken together, the authors conclude that the relative and absolute inter-cluster timing accuracy does not exceed 5 ns.","pith_inferences":["The paper's 5 ns statement is an upper bound set partly by the SSBT time resolution; a higher-rate reference would likely show the White Rabbit chain holds the clusters to tighter alignment, but that is not demonstrated here.","Instead of a single laser position, coincident atmospheric muons crossing several clusters would map all pairwise cluster delays and show whether the 5 ns accuracy holds across the whole array, not only the cluster 2–3 pair.","When the array grows to nine clusters, each new cluster needs its own survey tie; if the survey uncertainty is kept below about 1 m, the same per-second pulse and laser checks should transfer the demonstrated accuracy."],"forward_implications":["If the 5 ns accuracy holds, events depositing light in several clusters can be merged into one reconstructed event on a common timeline.","The two independent synchronization systems provide mutual verification, so a fault in one does not silently corrupt the other's timestamps.","The achieved accuracy matches the detector's 200 MHz sampling step, meaning the existing electronics can use the cross-cluster alignment without extra corrections.","The same deployment method extends to the planned nine-cluster array, since each cluster receives the same per-second pulse cross-check and laser calibration."],"supporting_citations":[{"why":"Defines the Baikal-GVD detector and its cluster structure, giving the objects whose timelines must be synchronized.","marker":"[1]"},{"why":"Documents the early construction phase that sets the cluster geometry and lake deployment conditions.","marker":"[2]"},{"why":"Reports Baikal-GVD status and results, against which the multi-cluster event reconstruction need is established.","marker":"[3]"},{"why":"Supplies the White Rabbit time-synchronization software and experience that the Baikal-GVD WR nodes use.","marker":"[7]"},{"why":"Provides the time-stamping design for SPEC/ZEN platforms that records the White Rabbit trigger time marks.","marker":"[8]"},{"why":"Describes the external time reference option whose connection reduces the measured synchronization RMS from about 10 ns to about 5 ns.","marker":"[9]"}],"fun_headline_variants":["Baikal-GVD inter-cluster sync hits 5 ns precision","Laser test proves Baikal-GVD sync to 5 ns","Baikal-GVD clusters timed to better than 5 ns","Two systems keep Baikal-GVD sync under 5 ns","Inter-cluster time sync at Baikal-GVD: <5 ns"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The laser calibration assumes the 18.6 m difference in distance from the light source to clusters 2 and 3 is known to better than the claimed few-nanosecond accuracy, but the paper does not report the survey uncertainty of that distance.","fun_headline_variants_meta":{"raw":{"variants":["Baikal-GVD inter-cluster sync hits 5 ns precision","Laser test proves Baikal-GVD sync to 5 ns","Baikal-GVD clusters timed to better than 5 ns","Two systems keep Baikal-GVD sync under 5 ns","Inter-cluster time sync at Baikal-GVD: <5 ns"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000176,"raw_usage":{"total_tokens":1217,"prompt_tokens":798,"completion_tokens":419,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":414,"completion_tokens_details":{"reasoning_tokens":329}},"tokens_in":414,"tokens_out":419,"duration_ms":3803,"temperature":1.0,"reasoning_tokens":329,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:10:16.181884+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-measure the laser-to-cluster distances with independent survey and check whether the 18.6 m difference is accurate to about 1 m; if its uncertainty exceeds roughly 1 m (about 5 ns in water), the 81–83 ns measured delays cannot by themselves substantiate the claimed 5 ns absolute synchronization accuracy.","supporting_citations":[{"cited_title":"Avrorin, V","cited_arxiv_id":null,"evidence_quote":"Defines the Baikal-GVD detector and its cluster structure, giving the objects whose timelines must be synchronized."},{"cited_title":"Avrorin, A.D","cited_arxiv_id":null,"evidence_quote":"Documents the early construction phase that sets the cluster geometry and lake deployment conditions."},{"cited_title":"Avrorin, A.D","cited_arxiv_id":null,"evidence_quote":"Reports Baikal-GVD status and results, against which the multi-cluster event reconstruction need is established."},{"cited_title":"Time Synchronization with White Rabbit - Experience from Tunka-HiSCORE","cited_arxiv_id":null,"evidence_quote":"Supplies the White Rabbit time-synchronization software and experience that the Baikal-GVD WR nodes use."},{"cited_title":"Brückner and R","cited_arxiv_id":null,"evidence_quote":"Provides the time-stamping design for SPEC/ZEN platforms that records the White Rabbit trigger time marks."},{"cited_title":"Wlostowski, https://www.ohwr.org/project/whiterabbit/uploads/ae3282acd8f9f6c5a9067b061202277d/ wr_external_reference.pdf 7","cited_arxiv_id":null,"evidence_quote":"Describes the external time reference option whose connection reduces the measured synchronization RMS from about 10 ns to about 5 ns."}],"review_version":1}