{"id":"3bc037a9-3454-4c55-87ac-bed403d04063","arxiv_id":"2502.07317","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"PandaX-4T's photon acceptance function reconstruction achieves 1.0 mm bulk and 4.4 mm surface position resolution, yielding surface background estimates of 0.09 (Run0) and 0.17 (Run1) events.","lead":"The PandaX-4T team reports two light-pattern algorithms for locating particle hits in its liquid xenon detector, with the chosen method reaching about 1 mm accuracy for interior events. Using that method, they estimate that surface radioactivity contributes only 0.09 and 0.17 background events in the two science runs analyzed.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The surface background count is not anchored by an mm-level ground truth at the TPC wall, and its quoted uncertainty excludes reconstruction-scale systematics.","rationale":"The reader's weakest assumption identifies the optical simulation calibration as a plausible upstream source of bias. I agree that matching only the RMS of the 83mKr S2 pattern is a weak constraint, and that PAF acceptances inherit any simulation error. However, the more directly load-bearing issue for the paper's headline physics result—the surface background count—is the absence of an mm-level absolute position check at the TPC wall and the omission of reconstruction-scale systematics from the quoted uncertainties. The 210Po radial profile is used both to define the GPC (Sec. 3.3) and to characterize the surface radial distribution (Sec. 5), so it cannot independently validate the radial scale. The final errors are simply the standard deviation across three fitting functions, which captures functional-form choice but not position-reconstruction bias. The fit quality in the validation region (χ2/NDF up to 4.4 in Table 2) further suggests the model shape is imperfect. A direct sensitivity test—shifting reconstructed radii by ±1–2 mm—would settle whether the central background estimate is robust; if the count moves by more than the quoted errors, the claim is conditional on an unverified assumption, consistent with the reader's CONDITIONAL verdict. I therefore do not change the verdict, but I emphasize that the condition should include a radial-shift sensitivity analysis and, ideally, a data-driven resolution check (e.g., 212Bi–212Po coincidences). The paper's individual resolution numbers have some internal consistency via TM/PAF agreement and MC-based checks, so the concern is not that the work is fundamentally wrong, but that the uncertainty budget for the background is incomplete.","tokens_in":12543,"tokens_out":8469,"duration_ms":85316,"concrete_test":"Propagate a systematic radial shift: apply ΔR = -2, -1, +1, +2 mm to every reconstructed surface-event position used in Sec. 5, rebuild the three surface-background models, and recompute the Run0 and Run1 expected counts inside the FV. If any shifted count falls outside the quoted 0.09±0.06 (Run0) or 0.17±0.11 (Run1), the uncertainty budget is understated and the background claim must be revised; if all shifted counts remain inside those intervals, the central number is robust at the mm scale and the optical-calibration concern is not consequential for this result.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim's most exposed element is the surface-background count, not the resolution itself. The FV boundary lies at R≈510 mm while the PTFE wall is at R=600 mm, so the predicted leakage into the FV is governed by the tail of the reconstructed radial distribution of surface events. That tail is set by PAF positions whose calibration chain has no independent, simulation-free ground truth at the wall: PTFE optical parameters are tuned to match a single RMS number of the 83mKr S2 pattern (Sec. 3.1, Fig. 3); PAF acceptance functions are fit to this simulation (Sec. 3.2); and GPC is derived from the same 210Po and 83mKr samples (Sec. 3.3). The 210Po radial width measures the convolution of resolution and any residual bias, but because GPC actively centers those events, it cannot validate the absolute radial scale. The quoted uncertainties—the spread of three fit functions in Table 2—omit this reconstruction-scale systematic. Moreover, the model's fit quality is marginal (χ2/NDF up to 4.4 for Run1, method 1), indicating the radial shape is not fully described. A 1–2 mm inward bias or an unmodeled low-energy resolution tail would shift the leaked count beyond the quoted ±0.06/±0.11, so the background number is not adequately supported without an explicit sensitivity analysis.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents the position reconstruction methods and surface background model used for the PandaX-4T dark matter search data. Two horizontal-position algorithms are described: a template matching (TM) method based on optical simulation, and a photon acceptance function (PAF) method with fitted per-PMT acceptance functions; both are supplemented by partial waveform reconstruction (PWR) and geometric position correction (GPC). The authors report a bulk event resolution of about 1.0 mm and a surface event resolution of about 4.4 mm for a typical S2 signal with QS2b = 1500 PE (about 14 keV), a uniformity of roughly 20% RSD, and average deviations of 5.1 mm and 8.8 mm for Run0 and Run1 due to disabled PMTs. A data-driven surface background model, based on the PAF reconstruction of 210Po surface events, predicts 0.09 ± 0.06 surface background events in Run0 (0.54 tonne·year) and 0.17 ± 0.11 in Run1 (1.00 tonne·year). The paper is written as a technical methods paper supporting the previously published PandaX-4T dark matter results.","tokens_in":12745,"tokens_out":3346,"duration_ms":32692,"significance":"If the stated performance is reliable, this paper provides an important technical foundation for PandaX-4T and for future liquid-xenon TPC analyses: it details a data-driven method for estimating the surface background leaking into the fiducial volume, and it compares two reconstruction algorithms with documented resolution, uniformity, and robustness metrics. The use of 210Po alpha events to calibrate the surface resolution in situ, the explicit comparison of three fitting functions for the radial tail, and the reported χ2/NDF values in the non-blind region are useful checks. However, several load-bearing points need strengthening before the central claims can be accepted as stated: the optical simulation is calibrated to a single scalar (the RMS of the 83mKr S2 pattern) without a systematic uncertainty, the bulk resolution is demonstrated only in simulation, and the surface background count is an extrapolation of a radial distribution fitted to the same 210Po events whose absolute radial scale is not independently constrained.","major_comments":[{"comment":"The optical simulation is tuned by matching only the mean RMS of the 83mKr S2 pattern (mean 217±21 mm data vs. 215±18 mm simulation), with PTFE reflectivity set to 99% and refractive index to 1.61. The PAF acceptance functions are then fitted to this simulation and used for all subsequent position reconstructions and for the surface background model. A single RMS number cannot validate the angular or positional details of the simulated light patterns. The authors should provide a sensitivity analysis that varies the optical parameters within plausible ranges (reflectivity, refractive indices, absorption and Rayleigh lengths in Table 1) and quantifies the resulting shifts in reconstructed radial positions, especially for events near the PTFE wall. Without such a study, the quoted resolutions and the surface background counts do not include a dominant systematic contribution.","section":null},{"comment":"The bulk event resolution of about 1.0 mm is derived entirely from simulation: 'Bulk event resolution is estimated based on simulations' and the reconstructed positions are compared with primary positions in MC. No data-based validation of the bulk resolution is provided, for example using the width of a known uniform source such as 83mKr after accounting for its intrinsic spread. As stated, the 1.0 mm figure is a simulation self-consistency check rather than a measured detector resolution. The paper should either present a data-anchored cross-check or explicitly state that the quoted bulk resolution is a MC-based estimate, with the associated caveat applied to the summary.","section":null},{"comment":"The surface background prediction inside the fiducial volume is an integral of a radial distribution fitted to 210Po edge events in the same dataset (the R−Φ median curve is set as the zero point and δR distributions are fitted slice-by-slice). Because GPC actively corrects the 210Po events to be symmetric about the wall position, the 210Po measurements cannot validate the absolute radial scale: an inward bias of 1–2 mm in the reconstructed R would change the leakage into the FV significantly, given that the FV boundary is at R≈510 mm while the wall is at R≈600 mm. The quoted uncertainties (0.09 ± 0.06 and 0.17 ± 0.11) are the standard deviations of the three fitting functions in Table 2 and exclude this reconstruction-scale systematic. The authors should add an explicit sensitivity analysis, e.g., shifting the reconstructed radial distribution by ±1 mm and ±2 mm, and propagating the resulting change to the predicted surface background count.","section":null},{"comment":"The fit quality of the surface background model in the non-blind region is marginal, with χ2/NDF values up to 4.4 (Run1, method 1) and 2.3–2.9 for other entries. These values indicate that the radial shape of the surface background is not fully described by the chosen functions, which weakens the reliability of the extrapolation into the blind region. The paper should examine the residual structure of these fits, consider whether an additional component (e.g., a wider Gaussian or a second exponential) is needed, and discuss how the residual model-data disagreement affects the uncertainty on the integrated surface background count.","section":null}],"minor_comments":[{"comment":"The symbol r⃗ is used for both the event position and the PMT position in the RMS definition; please use distinct symbols (e.g., r⃗_event and r⃗_i) to avoid ambiguity.","section":null},{"comment":"The role of the parameters ω and ρ is not defined precisely. In particular, the sentence 'The term ρ serves as a correction factor for the global reflection effects' is vague, and the 'cut-off correction' is not explained. A short formal definition of ρ and its allowed range, together with how ω is determined for edge versus inner PMTs, would make the PAF model reproducible.","section":null},{"comment":"The CDF endpoint choice is described only qualitatively: 'the segment's endpoint is set where CDF ≤ 50%'. The figure shows several tested ranges, but the optimum condition is not stated as a numerical criterion. Also, the GPC correction method is described only by example; please specify the functional form of the correction (e.g., a polynomial in z and Φ) and how 83mKr events are used to align the absolute geometric scale.","section":null},{"comment":"The three fitting functions are shown in the table, but the integration limits for the 'count' column are not stated. The text says 'integrating radially from 0 to the R2 boundary', but it is not clear whether the count includes events in the gap between the blind region and the wall, and how the normalization to 'low energy events reconstructed outside the PTFE wall' is performed. Please clarify the exact integration range and the normalization procedure.","section":null},{"comment":"The RSD definition uses ncrit in a way that is easy to misread; please specify that ncrit is a fixed bin index (set to 90 in this analysis) and clarify the formula by writing P(n) as a function of the bin index with explicit limits.","section":null}],"recommendation":"major_revision","confidential_remarks":"The paper is a technical detector paper squarely within the scope of physics.ins-det, and its release is timely given the published PandaX-4T dark matter results. The largest risk is that the surface background prediction, which is central to the dark matter analysis, may be biased by a reconstruction-scale systematic that is not currently propagated into the quoted uncertainty. I would encourage the editor to require the sensitivity analysis described in the major comments before acceptance; if the authors can demonstrate that the predicted surface background count is robust at the level of a few tenths of an event to plausible shifts in the reconstructed radial scale, the paper would be publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a competent, clearly written instrumentation paper. The genuinely new pieces are the mirror-image PMT extension to PAF, the PWR and GPC corrections, and the resulting performance numbers for PandaX-4T (1.0 mm bulk, 4.4 mm surface resolution at QS2b=1500 PE, ~20% uniformity, and the Run0/Run1 robustness deviations). These are not in the earlier PandaX-II or XENON papers. The paper also does something useful: it does a real three-way comparison (COG vs TM vs PAF), checks uniformity with two independent calibration sources, tests robustness against dead PMTs, and validates the surface model against non-blind data.\n\nThe soft spot is the surface-background estimate, and the stress-test note is right: the number is controlled by the inner tail of reconstructed radii for wall events, and that tail has no independent, simulation-free ground truth at the wall. The optical simulation is tuned to match a single RMS distribution of 83mKr (Sec. 3.1, Fig. 3); PAF acceptance functions are calibrated from that simulation (Sec. 3.2); GPC is derived from the same 210Po and 83mKr samples (Sec. 3.3). Since GPC deliberately centers the 210Po radial distribution, the measured 210Po width cannot validate the absolute radial scale. A 1–2 mm inward bias or an unmodeled low-energy tail would move the leaked count by more than the quoted ±0.06/±0.11, which are only the spread of three fitting functions in Table 2. The fit quality in the non-blind region is also marginal for Run1 method 1 (χ2/NDF = 4.4). This doesn't kill the paper, but the systematic uncertainty on the surface background should be expanded, and an explicit sensitivity scan over reconstruction bias and optical parameters would make the central claim much safer.\n\nThe circularity is moderate, as the reader says: the final count is an extrapolation of a data-driven fit to the same 210Po sample. The resolution numbers, by contrast, have independent grounding (MC truth for bulk, peak width for surface). No code or data are released, which is a shame but not disqualifying for a detector paper. Overall the paper is honest, cites the relevant PAF literature, and is written for an expert audience. It should not be desk-rejected. Send it to a referee who knows LXe TPC reconstruction, and ask for clarification of the GPC and re-weighting procedures plus an expanded systematic budget.","headline":"A technically solid PandaX-4T reconstruction paper whose surface-background number is credible but under-uncertaintied; it deserves refereeing, not desk rejection.","tokens_in":13873,"tokens_out":3638,"would_cite":true,"duration_ms":32347,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["95.35.+d","29.40.Mc"],"model":"deepseek-v4-flash","headline":"PandaX-4T's photon-acceptance-function reconstruction resolves bulk events to 1.0 mm and surface events to 4.4 mm at about 14 keV, and the data-driven surface model built on it predicts fewer than one surface background event in each…","keywords":["liquid xenon TPC","position reconstruction","photon acceptance function","surface background model","dark matter direct detection","PandaX-4T","S2 signal","fiducial volume"],"falsifier":"Place a collimated $^{83\\mathrm{m}}\\mathrm{Kr}$ source or a patterned mask at known $(x,y)$ positions across the TPC and compare PAF-reconstructed coordinates with those known positions; a systematic deviation larger than the quoted 1.0 mm bulk resolution would falsify the simulation-based acceptance functions and GPC. As a direct surface check, the fitted radial peak of reconstructed $^{210}\\mathrm{Po}$ events should sit at the PTFE wall radius (600 mm) within the 4.4 mm resolution, and a large residual offset would invalidate the surface model.","tokens_in":12212,"feed_emoji":"🎯","tokens_out":11235,"duration_ms":95644,"temperature":0.7,"pith_summary":"This paper reports how the PandaX-4T liquid-xenon detector converts the S2 light pattern on its top photomultiplier array into a precise horizontal vertex for each event. Two algorithms are developed, template matching and a photon acceptance function method; the PAF method is chosen as primary and the TM method as a verification cross-check. With partial waveform reconstruction and geometric position correction, the PAF method reaches 1.0 mm bulk resolution and 4.4 mm surface resolution for a typical S2 signal with $Q_\\mathrm{S2b}=1500$ PE (about 14 keV), with around 20% uniformity. The same reconstruction feeds a data-driven surface background model built from $^{210}\\mathrm{Po}$ surface events, which predicts $0.09 \\pm 0.06$ surface background events in Run0 (0.54 tonne·year) and $0.17 \\pm 0.11$ in Run1 (1.00 tonne·year). If these estimates hold, wall-origin surface leakage is no longer a dominant background in the current WIMP search region.","feed_headline":"Light-pattern method pins xenon events to 1 mm","feed_subtitle":"A photon-acceptance algorithm also drives the surface background model to 0.09 and 0.17 events in the two runs.","key_machinery":"The load-bearing object is the photon acceptance function (PAF), an analytic function $\\eta_i(\\iota_i)$ that gives, for each top PMT, the fraction of S2 photons collected by that PMT as a function of the distance $\\iota_i$ from the scattering point to the PMT center. Boundary reflection is modeled by adding 'mirror image' PMTs, and the horizontal position is obtained by maximum likelihood over the measured S2 charge pattern, scanning $(x,y)$ at 0.01 mm steps. Two corrections matter: partial waveform reconstruction (PWR), which uses only the CDF $\\le 50\\%$ portion of the S2 waveform to avoid afterpulse and photoionization tails, and geometric position correction (GPC), which uses $^{210}\\mathrm{Po}$ surface events and $^{83\\mathrm{m}}\\mathrm{Kr}$ events to remove azimuthal and radial biases. The same PAF-based coordinates generate the radial PDFs of the surface background model from $^{210}\\mathrm{Po}$ events sliced in $z$ and $Q_\\mathrm{S2b}$.","core_discovery":"The central claim is that the photon acceptance function (PAF) method, together with partial waveform reconstruction (PWR) and geometric position correction (GPC), gives vertex reconstruction accurate enough to support a data-driven surface background model in a multi-tonne liquid-xenon TPC. For a typical S2 signal with $Q_\\mathrm{S2b}=1500$ PE (about 14 keV), the authors report 1.0 mm bulk-event resolution and 4.4 mm surface-event resolution, an RSD uniformity of about 20%, and robustness against dead PMTs at the 5.1 mm (Run0) and 8.8 mm (Run1) level in off-PMT regions. Using $^{210}\\mathrm{Po}$ $\\alpha$ events from the PTFE wall as a template, reweighted to the low-energy ROI in $Q_\\mathrm{S2b}$, the model estimates $0.09 \\pm 0.06$ surface events for Run0 and $0.17 \\pm 0.11$ for Run1 inside the optimized fiducial volume. The paper's position is that these improvements are what make the surface background small enough to proceed with the Run0+Run1 WIMP analysis.","pith_inferences":["By extension, the PAF method should transfer to any dual-phase xenon TPC with a different PMT layout; the dominant cost is the one-time optical simulation calibration, not the reconstruction step.","The model implicitly assumes that $^{210}\\mathrm{Po}$ surface alphas, after reweighting in $Q_\\mathrm{S2b}$, represent the radial distribution of $^{210}\\mathrm{Pb}$ events at WIMP-recoil energies; a dedicated low-energy surface-source calibration could test that proxy assumption.","The remaining ~20% uniformity suggests the light-response model still has position-dependent structure; using the bottom PMT array or the S1 pattern as an additional cross-check could push surface resolution below 4.4 mm."],"forward_implications":["PAF positions allow the fiducial volume boundary to be set with confidence, since surface events are resolved at the 4.4 mm level.","The expected surface background inside the WIMP ROI is $0.09 \\pm 0.06$ events in Run0 and $0.17 \\pm 0.11$ in Run1, small enough that this source does not dominate the background budget.","The reconstruction remains usable when top PMTs fail: mean deviations in off-PMT regions are 5.1 mm for Run0 and 8.8 mm for Run1 at $Q_\\mathrm{S2t}=6000$ PE.","Better horizontal vertices propagate to improved energy reconstruction and detector non-uniformity corrections, which feed the dark matter sensitivity of the combined run."],"supporting_citations":[{"why":"Defines the combined Run0+Run1 exposure, the optimized fiducial volume, and the energy ROI in which the surface background count is integrated.","marker":"[4]"},{"why":"Supplies the Run0 dark matter result and the commissioning-run exposure that anchor the first surface background estimate.","marker":"[3]"},{"why":"Provides the analytic single-PMT acceptance form (Eq. 4) that the PAF method adapts with image PMTs.","marker":"[12]"},{"why":"Establishes the maximum-likelihood position estimation formalism used to infer $(x,y)$ from the measured S2 pattern.","marker":"[13]"},{"why":"The optical Monte Carlo simulation supplies the light patterns and templates used to calibrate PAF and TM.","marker":"[10]"},{"why":"The single-variable PAF approach used in an earlier xenon TPC is the direct antecedent of this method.","marker":"[18]"},{"why":"Supplies the optical-simulation template-matching method that runs as the verification algorithm against PAF.","marker":"[9]"},{"why":"Provides the figure-of-merit used to choose the fiducial-volume radius, which sets the integration range for the background estimates.","marker":"[19]"}],"fun_headline_variants":["Photon acceptance function achieves 1.0 mm bulk-event resolution","Surface background model slashes dark matter noise to 0.09 events","PandaX-4T's PAF method pins vertex positions to the millimeter","1 mm vertex resolution and robust surface background estimation","PAF algorithm sharpens xenon detector's eye for dark matter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire reconstruction chain is calibrated by tuning the optical simulation until one summary number, the RMS of the $^{83\\mathrm{m}}\\mathrm{Kr}$ S2 light pattern, matches data; if the simulated per-PMT light pattern is wrong in a way that leaves that RMS unchanged, every reconstructed position, resolution, and surface background count inherits the error.","fun_headline_variants_meta":{"raw":{"variants":["Photon acceptance function achieves 1.0 mm bulk-event resolution","Surface background model slashes dark matter noise to 0.09 events","PandaX-4T's PAF method pins vertex positions to the millimeter","1 mm vertex resolution and robust surface background estimation","PAF algorithm sharpens xenon detector's eye for dark matter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000301,"raw_usage":{"total_tokens":1788,"prompt_tokens":1050,"completion_tokens":738,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":666,"completion_tokens_details":{"reasoning_tokens":647}},"tokens_in":666,"tokens_out":738,"duration_ms":6752,"temperature":1.0,"reasoning_tokens":647,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T13:07:12.129150+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Place a collimated $^{83\\mathrm{m}}\\mathrm{Kr}$ source or a patterned mask at known $(x,y)$ positions across the TPC and compare PAF-reconstructed coordinates with those known positions; a systematic deviation larger than the quoted 1.0 mm bulk resolution would falsify the simulation-based acceptance functions and GPC. As a direct surface check, the fitted radial peak of reconstructed $^{210}\\mathrm{Po}$ events should sit at the PTFE wall radius (600 mm) within the 4.4 mm resolution, and a large residual offset would invalidate the surface model.","supporting_citations":[{"cited_title":"Meng, et al., Dark Matter Search Results from the PandaX-4T Com- missioning Run, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the Run0 dark matter result and the commissioning-run exposure that anchor the first surface background estimate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the analytic single-PMT acceptance form (Eq. 4) that the PAF method adapts with image PMTs."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the maximum-likelihood position estimation formalism used to infer $(x,y)$ from the measured S2 pattern."},{"cited_title":"Chen, et al., BambooMC — A Geant4-based simulation program for the PandaX experiments, Journal of Instrumentation 16 (09) (2021) T09004","cited_arxiv_id":null,"evidence_quote":"The optical Monte Carlo simulation supplies the light patterns and templates used to calibrate PAF and TM."},{"cited_title":"Zhang, et al., Horizontal position reconstruction in PandaX-II, Journal of Instrumentation 16 (11) (2021) P11040","cited_arxiv_id":null,"evidence_quote":"The single-variable PAF approach used in an earlier xenon TPC is the direct antecedent of this method."},{"cited_title":"Langrock, Energy response and position reconstruction in the DEAP- 3600 dark matter experiment, Journal of Physics: Conference Series 1342 (2020) 012071","cited_arxiv_id":null,"evidence_quote":"Supplies the optical-simulation template-matching method that runs as the verification algorithm against PAF."},{"cited_title":"Cowan, Discovery sensitivity for a counting experiment with back- ground uncertainty (2012)","cited_arxiv_id":null,"evidence_quote":"Provides the figure-of-merit used to choose the fiducial-volume radius, which sets the integration range for the background estimates."}],"review_version":1}