{"id":"61e414e8-81de-47af-a357-629dd22ee1e3","arxiv_id":"2412.07104","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"GPU-accelerated reprocessing of 87% of the HTRU-S LowLat survey data discovered 71 new pulsars, including 6 millisecond pulsars and 7 with very high dispersion measures.","lead":"Astronomers reprocessed archived radio data from a southern-sky pulsar survey using GPU-based software and found 71 previously missed pulsars, including six millisecond pulsars and seven very distant ones. The result shows how revisiting old data with faster algorithms can substantially increase the known neutron-star population.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The abstract's unqualified 'discovery of 71 pulsars' is inflated: Table 3 and Section 5 count PSR J1325–6253, already published in Sengar et al. 2022, among the 71. The headline new-pulsar count should be corrected or explicitly qualified.","rationale":"The reader's designated weakest assumption concerned the flux-density positional offset and DM-distance models used for the 'fainter and more distant' claim. Those are genuine uncertainties, but the paper explicitly hedges that claim with 'could be' and 'should be interpreted with caution', and the discovery of the pulsars does not depend on them. The count issue is more directly load-bearing for the central '71 new pulsars' claim because it is an internal inconsistency: the same object is described as new in the abstract and as already published in the conclusion. The pipeline's redetection of 792 known pulsars and follow-up confirmation of the candidates provide strong independent support that the remaining pulsars are real, so this is a correction-and-qualification issue rather than a reason to reject the survey result. The reader's CONDITIONAL verdict already anticipates revisions, so no verdict change is needed; agreement is partial because the reader mentioned the J1325–6253 inclusion in the rationale but did not make it the formal weakest assumption.","tokens_in":34090,"tokens_out":7320,"duration_ms":76127,"concrete_test":"Take Table 3's 71 names and query the ATNF PSRCAT and ADS for each name and for the discovery dates listed in Sengar et al. (2022), Sengar et al. (2023), the CRAFTS discovery page, and the MMGPS-S discovery list. Count the objects whose first published discovery precedes this paper. If J1325–6253 is the only prior publication, revise the abstract and Section 8.2 totals to '70 new pulsars plus the previously published J1325–6253'; if J1614−5218 or J1854−05 also predate submission, subtract them as well and recompute the survey-yield numbers in Table 6. As a second check, recompute the KS-test statistics in Section 7 with the corrected sample size to confirm the DM and S/N conclusions are unchanged.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4 reports 71 confirmed pulsars, and Section 5 states 'Among the 71 new pulsars reported in this work, PSR J1325–6253 is a double neutron star system', while the Table 3 entry for J1325−6253 carries the footnote 'A double neutron star system (Sengar et al. 2022)'. The conclusion then says 'With the exception of PSR J1325–6253, which has already been published...'. So at least one of the 71 objects already appeared in a dedicated discovery paper before this survey paper. The abstract presents 'discovery of 71 pulsars' with no asterisk or caveat, and Section 8.2 uses the 71/94 counts to argue survey completeness. The central quantitative claim is therefore internally inconsistent: either the count should be 70, or the text should say '71 pulsars reported from this reprocessing, one of which was published separately as PSR J1325–6253.' The same check is needed for the two entries footnoted as independent discoveries in MMGPS-S and CRAFTS (J1614−5218 and J1854−05); if either was published before submission, the 'new' count drops further. This is a factual, verifiable discrepancy, not a matter of model choice, and it directly affects the headline result.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports a GPU-accelerated coherent reprocessing of 87.3% of the HTRU-S LowLat survey data using the PEASOUP pipeline, searching the full 72-minute observations at native time resolution with an acceleration range up to |50| m/s^2. The authors report 71 confirmed pulsar discoveries, including 6 millisecond pulsars and 7 pulsars with DM > 800 pc cm^-3, and compare the new sample with previous HTRU-S LowLat pulsars and the background pulsar population in period, DM, signal-to-noise, flux density, distance, and luminosity. The paper also evaluates survey yield against population-synthesis predictions and discusses why the pulsars were missed in earlier processing passes. The core discovery claim is supported by follow-up confirmations, while the statistical characterization rests on minimum flux densities and model-dependent distances.","tokens_in":34395,"tokens_out":9371,"duration_ms":86784,"significance":"If the discovery-count issue is corrected, this is a valuable survey paper. The 71 pulsars were confirmed in follow-up observations, the pipeline was validated by redetecting 792 unique known pulsars, the search parameters are explicitly tabulated, and the open-source PEASOUP code is used. The demonstration that folding candidates below the FFT false-alarm threshold yields real pulsars, and the identification of the incorrect downsampling scaling as a cause of missed pulsars, are concrete and useful technical contributions. The paper also provides a careful comparison with population-synthesis predictions and a detailed account of the reasons for earlier non-detections. These strengths make the manuscript worth publishing after the counting and statistical-claim issues below are resolved.","major_comments":[{"comment":"The headline 'discovery of 71 pulsars' is internally inconsistent with the paper's own statements. Section 5 says 'Among the 71 new pulsars reported in this work, PSR J1325–6253 is a double neutron star system', while Table 3 footnotes PSR J1325–6253 as published in Sengar et al. (2022), and Section 10 says 'with the exception of PSR J1325–6253, which has already been published'. In addition, Section 5.3 states that PSR J1638−47 and PSR J1710−3946 'were reported as an independent discovery by Sengar et al. (2023)', and Section 6 states that 10 of the PMPS blind-FFT detections 'are included as independent discoveries in the PMPS reprocessing (Sengar et al. 2023) (they were originally discovered for the first time in this work)'. Because Sengar et al. (2023) is published, these objects already have public discovery papers. The abstract's unqualified 'discovery of 71 pulsars' and the conclusion's 'reported for the first time in this study' must be corrected, and the counts in Section 8.2 and Table 6 must be re-derived consistently.","section":"§4, Table 3, §5, §10"},{"comment":"The claim that the new sample probes fainter pulsars is based on minimum flux densities computed under the assumption that the discovery position is the beam centre, with a single median positional offset of 3.94 arcmin transferred from the previous 88-pulsar sample to the new sample. Because the actual positional offsets of the new pulsars are unknown, and because S_min is a lower bound that increases when a pulsar is off-axis, the KS-test comparison with p=0.014 is conditional on an unverified assumption. The authors should either propagate the uncertainty in the offset distribution, demonstrate that the conclusion is robust to the plausible range of offsets, or explicitly state in the abstract and conclusions that the 'fainter' claim depends on the beam-centre assumption.","section":"§7.2"},{"comment":"The claim that the new pulsars are on average more distant rests entirely on NE2001 and YMW16 dispersion-measure distances. The text cautions that DM-derived distances are not well constrained, but the abstract presents the greater average distance as a headline result. The KS p-values of 0.007 and 0.009 are model-dependent and do not include systematic errors in the electron-density models. The authors should add a systematic-error treatment or soften the summary and conclusions to make the model dependence and the 'could be' caveat explicit, rather than presenting model-dependent distances as a firm population difference.","section":"§7.3"}],"minor_comments":[{"comment":"The pulsar is called PSR J1445−63 in the text and Table 3, but PSR J1449−63 in Figure 2 and Table 4; the listed right ascension 14:49.8 suggests J1449−63 is the correct name, and the inconsistency should be fixed.","section":"§5.1.1, Table 3, Figure 2, Table 4"},{"comment":"The text refers to 'PSR J1651–42' with DM ~931 pc cm^-3, but Table 3 lists PSR J1652−4237 with DM 943 pc cm^-3; the intended pulsar should be identified consistently.","section":"§5.3"},{"comment":"The search radius is given as 'one beam-width (14.4◦)'; this should be 14.4 arcminutes, not degrees, since the Parkes multibeam receiver's beam FWHM is approximately 14.4 arcmin.","section":"§6"},{"comment":"The text says the sky temperature was 'scaled at 1400GHz as ν^-2.6'; this should be 1400 MHz.","section":"§7.2"},{"comment":"The text states that 457 unique pulsars were selected, while the Figure 8 caption says 467; the correct number should be used consistently.","section":"§8.1 and Figure 8"},{"comment":"The column header 'S/NPMPS' is repeated in Table 3, and the table caption contains the typo 'The paramteres'; also, the period and DM units in the Table 3 continuation header are inconsistent with the first part of the table.","section":"Table 3 and Table 4"},{"comment":"The text says the total observing bandwidth went 'from 288 MHz for PMPS to 3402 MHz for HTRU-S LowLat'; this should be 340 MHz, consistent with the effective bandwidth stated elsewhere.","section":"§2"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the count discrepancy is factual and easy to fix, but it affects the headline result and the survey-yield accounting in Section 8.2. The paper's own text acknowledges prior publication of PSR J1325−6253 and reports that PSR J1638−47 and PSR J1710−3946 were independently reported in Sengar et al. (2023), so the 'first time in this study' wording cannot stand. The rest of the paper is technically sound and the discovery claims are well supported by follow-up observations; with the counts corrected and the statistical claims appropriately qualified, the paper should be suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the punchline: this is a solid reprocessing paper that delivers a big haul of new pulsars, but the headline \"71 new pulsars\" is not quite right — PSR J1325–6253 was already published by the same group in 2022, and the paper admits this in the conclusion while still counting it among the 71 elsewhere. That is a one-number correction, but it needs to be made in the abstract, not just in a footnote.\n\nWhat is genuinely new and good: the paper applies the PEASOUP GPU pipeline to the full-resolution HTRU-S LowLat data, covering 87% of the survey with coherent acceleration searches up to ±50 m/s². The 71 confirmed pulsars (70 genuinely new after removing the DNS) include 6 MSPs, 5 binaries, and 7 high-DM pulsars, all confirmed with follow-up observations. The analysis of why these were missed — the downsampling bug that effectively produced 1-bit data, the S/N threshold of 8, and the restriction to 16 harmonics — is clear and convincing. The redetection of 34 of the new pulsars in PMPS gives independent support. The scattering analysis of 9 pulsars is a nice extra.\n\nThe soft spots are real but not fatal. First, the count issue above. Second, the claim that the new sample is fainter and more distant rests on beam-center flux densities and NE2001/YMW16 distances; the authors acknowledge the distance caveat, but the abstract states it as a result. A median positional offset correction is a crude fix, not a substitute for timing positions. Third, the data are available only \"on reasonable request\"; that is a limitation for anyone wanting to verify or reuse the sample. Fourth, the CRAFTS and MMGPS footnoted discoveries need to be checked for prior publication — if either was public before submission, the new count drops further.\n\nWho is this for? Pulsar astronomers and anyone doing survey reprocessing. It deserves refereeing; with the count clarified and the statistical language tightened, it would be a strong contribution.","headline":"Solid reprocessing haul, but the '71 new pulsars' headline double-counts PSR J1325–6253, already published in 2022; fix the count and the paper is a strong contribution.","tokens_in":35027,"tokens_out":3882,"would_cite":true,"duration_ms":37523,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Reprocessing archived southern-sky pulsar survey data with full time resolution and coherent folding reveals 71 previously missed pulsars, including six millisecond pulsars and seven high-dispersion objects.","keywords":["pulsar surveys","HTRU-S LowLat","GPU acceleration","coherent folding","millisecond pulsars","binary pulsars","dispersion measure","Galactic plane"],"falsifier":"Obtain precise interferometric positions and independent distance estimates (for example via parallax or HI absorption) for a representative subset of the 71 pulsars; if the recalculated 1400-MHz flux densities and distances no longer show a statistically significant difference from the previous 100-pulsar sample in a Kolmogorov-Smirnov test, the paper's central statistical claim would be falsified.","tokens_in":33875,"feed_emoji":"📡","tokens_out":8147,"duration_ms":79473,"temperature":0.7,"pith_summary":"The paper claims that reprocessing 87.3% of the archival High Time Resolution Universe South Low Latitude (HTRU-S LowLat) survey at full 64-microsecond time resolution with a GPU-accelerated coherent search recovers 71 pulsars that earlier processing passes missed, among them six millisecond pulsars (five in binary systems) and seven pulsars with dispersion measures above 800 pc $cm^{-3}$. It argues that the earlier passes lost sensitivity by downsampling the data — effectively reducing two-bit data to one-bit precision and lowering signal-to-noise by roughly 20–25% — and by imposing a fixed false-alarm threshold that rejected many narrow-duty-cycle pulsars whose folded signal-to-noise ratio greatly exceeds their Fourier-domain signal-to-noise ratio. The new sample is statistically fainter and more distant than the previous 100 pulsars from the same survey, and the reprocessing brings the survey's normal-pulsar yield into line with population-synthesis predictions. A sympathetic reader should care because the result shows that archival survey data, searched coherently at native resolution, still holds large numbers of undiscovered pulsars, especially faint, distant, and binary ones.","feed_headline":"71 pulsars found hiding in old southern-sky survey data","feed_subtitle":"Coherent folding of 72-minute observations uncovers faint, distant pulsars and five binary millisecond pulsars.","key_machinery":"The load-bearing mechanism is coherent folding of low-significance Fourier candidates: instead of accepting or rejecting pulsar candidates by a fixed spectral signal-to-noise cutoff, the pipeline folds every plausible candidate — including those near the Fourier noise floor — at the candidate period, dispersion measure, and acceleration, so the full pulse energy is summed in the time domain. This is paired with a GPU-accelerated time-domain resampling acceleration search that covers the complete 72-minute observation at 64-microsecond native sampling over accelerations up to |50| m $s^{-2}$, which is most sensitive to binary pulsars whose orbital acceleration is nearly constant across the observation. The FFT search itself uses up to 32 harmonics, which helps recover narrow-duty-cycle pulsars whose power is spread across many harmonics.","core_discovery":"The paper's central claim is that a coherent, GPU-accelerated reprocessing of the full 72-minute HTRU-S LowLat observations — searched at native sampling, with an acceleration range up to |50| m $s^{-2}$, 2778 trial dispersion measures up to 2000 pc $cm^{-3}$, and 32-harmonic summing — discovers 71 new pulsars, increasing the survey's new-pulsar yield by roughly 75% over the 100 found in the previous two passes. The key evidence is that many of the new pulsars were found by folding candidates with Fourier signal-to-noise ratios near or below the conventional detection threshold; for narrow-duty-cycle pulsars the folded signal-to-noise ratio can be 1.5 to 2.5 times higher than the incoherent harmonic sum, so a fixed threshold had been discarding real signals. The paper further shows that the new sample has higher dispersion measures, lower flux densities, and larger DM-derived distances than the earlier sample, and that the survey's normal-pulsar detection count now matches population-synthesis expectations.","pith_inferences":["An editorial extension: other archived pulsar surveys that used data downsampling and fixed S/N thresholds may harbour a comparable hidden population; re-running them with coherent folding at native resolution could yield a similar fractional increase in discoveries.","An editorial extension: the paper's emphasis on full-length coherent acceleration searches suggests that a GPU-reimplemented segmented or jerk-aware search, applied to the same data, might still find short-orbit relativistic binaries that constant-acceleration searches miss.","An editorial extension: the measured scattering timescales that exceed NE2001 predictions for many high-DM pulsars could be used to identify foreground H II regions and to test improved electron-density models, since these pulsars trace particularly complex sightlines.","An editorial extension: if future large surveys adopt the 'fold everything near the noise floor' strategy, GPU-accelerated folding will become the practical bottleneck, since CPU-side folding of millions of candidates will not scale to SKA-era data volumes."],"forward_implications":["The known pulsar census in the southern Galactic plane grows by 71 objects, including five binary millisecond pulsars and a double neutron star, improving statistics for binary-evolution and population studies.","The survey's normal-pulsar yield now essentially matches the predicted ~890 detections from population synthesis, indicating that the original survey design and sensitivity estimates were correct.","Seven new pulsars with dispersion measures above 800 pc cm^-3, two among the top ten highest-DM pulsars known, provide new probes of the Galactic electron distribution and scattering along tangential spiral-arm directions.","Since 34 of the 71 pulsars are also detectable in older multibeam-survey data, the result implies that comparable archived surveys reprocessed this way could yield similar numbers of hidden pulsars.","The five binary millisecond pulsars found through the acceleration search demonstrate the value of searching full-length observations without segmentation or downsampling for finding mildly accelerated binary systems."],"supporting_citations":[{"why":"Supplies the GPU pulsar-search pipeline, the candidate-sifting and folding strategy, and the harmonic-sum/false-alarm analysis that the reprocessing is built on.","marker":"Sengar et al. (2023)"},{"why":"Reports the first processing of half the HTRU-S LowLat survey; its 60 pulsars and segmented acceleration search define the baseline the new search surpasses.","marker":"Ng et al. (2015)"},{"why":"Reports the second processing of 44% of the survey, adding 40 pulsars and PSR J1757−1854; its sample and yield statistics are compared throughout.","marker":"Cameron et al. (2020)"},{"why":"Describes the HTRU-S LowLat survey design, receiver parameters, and the original prediction of 900 detectable normal pulsars.","marker":"Keith et al. (2010)"},{"why":"Provides the NE2001 Galactic electron-density model used to convert dispersion measures to distances and to benchmark scattering timescales.","marker":"Cordes & Lazio (2002)"},{"why":"Provides the YMW16 electron-density model used as the alternative distance estimate and as the maximum-DM reference along each line of sight.","marker":"Yao et al. (2017)"},{"why":"Predicts 64 millisecond-pulsar detections for the survey, the upper benchmark against which the observed MSP yield is judged.","marker":"Levin et al. (2013)"},{"why":"Presents the timing solution for PSR J1325−6253, the double neutron star system counted among the new discoveries.","marker":"Sengar et al. (2022)"}],"fun_headline_variants":["GPU reprocessing of old survey data finds 71 pulsars","Coherent GPU fold of old data yields 71 pulsars","71 pulsars found by coherent re-fold of survey data","GPU re-analysis reveals 71 pulsars, 5 binary MSPs"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that the new pulsars are systematically fainter and more distant assumes that each discovery position, corrected only by the median 3.94 arcminute offset, is close enough to the true position that the derived minimum flux densities are unbiased, and that the NE2001 and YMW16 electron-density models give reliable distances.","fun_headline_variants_meta":{"raw":{"variants":["GPU reprocessing of old survey data finds 71 pulsars","Coherent GPU fold of old data yields 71 pulsars","71 pulsars found by coherent re-fold of survey data","GPU re-analysis reveals 71 pulsars, 5 binary MSPs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000295,"raw_usage":{"total_tokens":1784,"prompt_tokens":1083,"completion_tokens":701,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":699,"completion_tokens_details":{"reasoning_tokens":629}},"tokens_in":699,"tokens_out":701,"duration_ms":8109,"temperature":1.0,"reasoning_tokens":629,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:06:19.363932+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Obtain precise interferometric positions and independent distance estimates (for example via parallax or HI absorption) for a representative subset of the 71 pulsars; if the recalculated 1400-MHz flux densities and distances no longer show a statistically significant difference from the previous 100-pulsar sample in a Kolmogorov-Smirnov test, the paper's central statistical claim would be falsified.","supporting_citations":[],"review_version":1}