{"id":"39543c23-3837-4b87-b415-43dde4562b69","arxiv_id":"2607.20354","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"After in-situ receiver calibration, ultrasonic time-of-flight reconstruction reaches 1.23 cm mean error on the JUNO central axis and a simulated 2.40 cm off-axis positioning uncertainty.","lead":"A team working on the JUNO neutrino experiment shows that an ultrasonic positioning system can locate a radioactive calibration source inside a 35-metre liquid-scintillator sphere with one-to-two-centimeter accuracy. This matters because off-axis calibration needs autonomous, light-obstructing-free positioning, and centimeter-level precision is the benchmark for large neutrino detectors.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Off-axis 2.40 cm claim rests on simulation calibrated by the same central-axis data that produce the 1.23 cm in-sample residual; no independent off-axis ground truth is presented, so a common-mode ACU or effective-geometry bias would shift both numbers together.","rationale":"The reader's weakest assumption identifies essentially the same load-bearing concern: the 1.23 cm central-axis number is an in-sample residual, and the 2.40 cm off-axis number is a simulation whose receiver-coordinate uncertainty is inherited from that same fit. I agree. The engineering content — LS sound-speed measurement, arrival-time extraction, hardware stability, and the qualitative improvement after in-situ calibration — is real and independently supported by data shown in the paper. But the central claim of off-axis capability rests on a simulation that has not been validated against any independent off-axis reference, and the in-situ calibration has no guard against systematic ACU or off-axis effective-geometry errors. Explicitly flagging the Sec. 3.1 horizontal-uniformity assumption adds a concrete physical mechanism by which off-axis errors could exceed the simulation's random smears. The proposed off-axis anchor/cable-metrology check, or failing that a split-half cross-validation of the 33 ACU deployments, would settle whether the concern actually lands. Since the reader already assigned CONDITIONAL with moderate confidence, this stress-test does not move the verdict.","tokens_in":11749,"tokens_out":6664,"duration_ms":60827,"concrete_test":"Run a dedicated off-axis ground-truth check: acquire USS data at the CLS anchor and at 5-10 points where the CLS cable-length/encoder readout provides independent positions at the ~1 cm level, then compute point-by-point 3D residuals between USS reconstruction and these references. If the RMS residual is consistent with 2.40 cm and the residuals show no common-mode bias correlated with position or receiver geometry, the off-axis claim is supported; if a correlated bias appears or the RMS exceeds ~3 cm, the simulated 2.40 cm RMSE is not a valid performance bound and the central claim should be downgraded.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. 4 calibrates six effective receiver coordinates by fixing 33 ACU emitter positions as exact ground truth in the PMT frame. The reported 1.23 cm mean error (Fig. 10) is computed on those same 33 deployments, i.e. it is an in-sample residual of the calibration fit, not an out-of-sample accuracy measurement. The off-axis performance in Sec. 6 is simulation-only: 100,000 points in the CLS plane, with receiver-position uncertainty entered as independent 0.8 cm Gaussian smears (Table 1), a value derived from the statistical covariance of that same central-axis fit. Consequently, any systematic error in the ACU reference (cable-metrology zero offset, coordinate-frame tilt, thermal expansion/deformation) is absorbed by the fitted receiver coordinates and propagates unmitigated into the simulated off-axis reconstruction. The 0.8 cm Gaussian models only random fitting uncertainty and cannot cover such a common-mode bias. In addition, Sec. 3.1 explicitly assumes the temperature field is horizontally uniform, yet the CLS plane extends to 17.7 m radius; an unmeasured horizontal gradient of only ~0.5 C would change the sound speed by ~2 m/s (using Vsound = 1469.69 - 3.916 T), giving path-length errors of order a centimeter, comparable to the claimed 2.40 cm RMSE. No independent off-axis ground truth is used: the CLS anchor is visible in Fig. 11 but is not exploited as a cross-check. Therefore the central claim that off-axis calibration can run without mechanical position control is plausible but not yet established by data.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an ultrasonic positioning system (USS) for JUNO calibration sources. The system determines source coordinates by fitting time-of-flight measurements to a sound-speed model based on laboratory LS measurements and in-detector temperature profiles, with arrival times extracted by a fixed-peak method. An in-situ calibration of the six active receiver coordinates is performed using 33 central-axis ACU deployments, after which the central-axis reconstruction yields a mean 3D error of 1.23 cm (Fig. 10). For off-axis CLS operation, a detector-realistic simulation with timing jitter, sound-speed variation, and receiver smearing predicts an RMSE of 2.40 cm across the CLS plane (Fig. 15). The paper argues that these results demonstrate centimetre-level source positioning and support off-axis calibration in JUNO-like detectors.","tokens_in":12165,"tokens_out":2157,"duration_ms":20403,"significance":"If the claimed accuracy holds, the USS would enable off-axis calibration in JUNO without mechanical position control, which is an important capability for the experiment's physics program. The paper's strengths include the integration of a custom sound-speed measurement, careful waveform-based arrival-time extraction, and a realistic simulation chain with parameters tied to measured data. The in-situ calibration of receiver geometry under filled-detector conditions is a useful contribution, and the central-axis improvement from 14.75 cm to 1.23 cm clearly shows that the calibration corrects large survey/deformation offsets. However, the central-axis result is an in-sample residual of the same fit that determines the receiver coordinates, and the off-axis result rests on a simulation with no independent ground truth. These two limitations are load-bearing for the central claim and need to be addressed before the paper can support its conclusion.","major_comments":[{"comment":"The 1.23 cm central-axis mean error is computed on the same 33 ACU deployments used to fit the six receiver coordinates. This is an in-sample residual of the calibration fit, not an out-of-sample accuracy measurement. Any systematic error in the ACU reference frame (cable zero offset, tilt, thermal expansion) is absorbed by the fitted receiver positions and does not appear in the residual. Please provide a leave-one-out or independent validation (e.g., a subset of ACU positions reserved from the calibration, or a comparison with the CLS anchor) to substantiate the absolute accuracy claim.","section":"§4, Fig. 10"},{"comment":"The off-axis 2.40 cm RMSE is entirely simulation-based. The 0.8 cm receiver smearing is taken from the statistical covariance of the central-axis calibration and is added as independent Gaussians. This models only random fitting uncertainty; a common-mode bias in the ACU reference or an off-axis drift of effective receiver positions would shift both the central-axis and off-axis results in the same direction and is not covered by the smearing. The CLS anchor visible in Fig. 11 is not used as an independent cross-check. Please include a measurement-based off-axis validation, or at least quantify the sensitivity to a global receiver-geometry bias (e.g., by offsetting all fitted R_i coherently) and to the ACU reference uncertainty.","section":"§6, Table 1"},{"comment":"The temperature field is assumed horizontally uniform, with variation only along z. The CLS plane extends to r = 17.7 m, so an unmeasured horizontal gradient of only ~0.5 °C would change the sound speed by ~2 m/s (using Vsound = 1469.69 − 3.916 T), producing path-length errors of order 1 cm—comparable to the claimed 2.40 cm RMSE. The simulation uses a single sound-speed value with ±0.9 m/s smearing (Table 1), which is not equivalent to a spatially varying field. Please justify the horizontal-uniformity assumption with data or include a horizontal gradient in the simulation to estimate its impact.","section":"§3.1, §6"},{"comment":"The reconstruction treats all six receivers with equal weights and uses t0 as a free parameter per event. The calibration in §4 also fits receiver coordinates while assuming emitter positions are exact ground truth. It would be helpful to explicitly state whether the 0.8 cm receiver-parameter uncertainty is fully propagated into the central-axis residual (Fig. 10) or whether the residual is computed with receiver coordinates fixed at the fitted values. If the latter, the reported 1.23 cm does not include the calibration's own covariance contribution and is optimistic.","section":"§6.1, Eq. (1)"}],"minor_comments":[{"comment":"The text refers to a resonant frequency of 142 Hz for the piezoelectric element; this must be a typo (likely 142 kHz). Please correct.","section":"§2 / Fig. 3 caption"},{"comment":"The fitting formula Vsound,LS = 1469.69 − 3.916×T is given without units for the coefficients and without an uncertainty on the slope. Adding the fit uncertainty and noting the valid temperature range (already stated) would help the reader propagate the sound-speed error.","section":"§3.1, Eq. after Fig. 6"},{"comment":"Figure 9 shows the resolution for four methods with and without a 5% adjacent-peak contamination. It would be useful to state explicitly how the 5% fraction was obtained from manual counting and whether the same fraction is applied in the simulation consistently.","section":"§3.2, Fig. 9"},{"comment":"The CLS anchor is marked but never used as a positioning cross-check. Since it is a known reference point, a sentence explaining why it is not used (or using it) would strengthen the paper.","section":"§5.2, Fig. 11"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses an important practical problem and the engineering work is solid, but the headline numbers (1.23 cm and 2.40 cm) are both anchored to the same calibration dataset without an independent out-of-sample check. The off-axis claim depends on simulation assumptions that are not validated against the one available reference (the CLS anchor). I would support acceptance after the authors provide at least one independent cross-check (e.g., leave-one-out on the ACU data, a CLS-anchor measurement, or a bias-sensitivity scan) and address the horizontal temperature-gradient concern. The lack of such validation is the main reason for major revision rather than minor."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Jon,\n\nThis is a real engineering paper from the JUNO group. They have a working ultrasonic positioning system in a large liquid-scintillator detector, they measured the LS sound speed as a function of temperature with a custom device, and they developed an in-situ calibration that takes the central-axis reconstruction error from ~15 cm to ~1.23 cm. That calibration procedure is the genuinely new piece here, and the sound-speed relation is a useful measurement. The hardware and TOF concept were in their earlier papers.\n\nBut treat the headline numbers with caution. The 1.23 cm is a residual computed on the same 33 ACU deployments that were used to fit the six receiver coordinates. That is in-sample; it measures how well the fit reproduces its own input, not how well the system locates a source the receiver geometry wasn't calibrated on. The 2.40 cm off-axis figure is simulation-only, and the receiver smearing in that simulation is taken from the statistical covariance of the same fit. Any systematic error in the ACU reference frame, or a real horizontal temperature gradient in the LS, would feed into both numbers in the same direction. The paper explicitly assumes the temperature is horizontally uniform; a 0.5°C gradient would change sound speed by ~2 m/s and give path-length errors of order a centimeter, which is the same scale as the claimed 2.40 cm.\n\nThere are also two presentation issues. Table 2 lists the JUNO off-axis figure alongside measured numbers from Daya Bay, Super-K, etc., without making it clear the JUNO entry is a simulation prediction. And the CLS anchor, which is sitting right there in Fig. 11, is never used as an independent cross-check. No code or data is released.\n\nNone of this kills the paper. The in-situ calibration showing a reduction from ~15 cm to ~1 cm is a legitimate result, and the simulation framework is reasonable given measured timing and sound-speed inputs. But the central claim—that off-axis calibration can run without mechanical position control at cm level—is not yet demonstrated by data. A referee should ask for leave-one-out or held-out-axis validation, a systematic error budget for the ACU reference, and at least one off-axis point with independent ground truth.\n\nI'd send it to peer review, not desk reject. It's a solid instrumentation paper with a fixable validation gap.","headline":"Useful JUNO calibration engineering, but the headline 1.23 cm and 2.40 cm numbers are in-sample and simulation-only; the hardware work is solid, the validation is not yet.","tokens_in":12739,"tokens_out":3221,"would_cite":true,"duration_ms":28612,"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":"An ultrasonic time-of-flight system can position calibration sources inside a large liquid-scintillator neutrino detector to within 1.23 cm on the central axis and 2.40 cm in an off-axis plane, according to this paper.","keywords":["ultrasonic positioning","detector calibration","liquid scintillator","time-of-flight","JUNO","receiver geometry calibration","sound speed","neutrino detector"],"falsifier":"Deploy the CLS source at a set of points whose true positions are determined independently (e.g., by the calibrated cable-loop metrology or a CCD camera system), reconstruct the positions with the ultrasonic system, and compare the three-dimensional residuals across the full plane. If the RMS deviation exceeds the claimed 2.40 cm, or if the residual pattern correlates with the central-axis calibration positions, the off-axis extrapolation is not supported.","tokens_in":11636,"feed_emoji":"📍","tokens_out":4405,"duration_ms":42848,"temperature":0.7,"pith_summary":"The paper argues that ultrasonic positioning can supply centimetre-level three-dimensional coordinates for radioactive calibration sources inside JUNO's 35.4 m-diameter liquid-scintillator sphere, even in off-axis regions where mechanical control is unavailable. On the central axis, where the deployment machinery gives a known reference, the system's mean reconstruction error is 1.23 cm after an in-situ calibration of the receiver geometry. For the off-axis Cable Loop System plane, a detector-realistic simulation that includes timing jitter, sound-speed variation, peak misidentification, and receiver-coordinate smearing predicts a positioning RMSE of 2.40 cm. If these numbers hold, JUNO can trust acoustic positions for calibration in regions that older detectors could only reach with ropes and poles, and other large liquid-scintillator detectors could adopt the same approach.","feed_headline":"Acoustic tracking locates JUNO calibration sources to 1.23 cm","feed_subtitle":"Centimetre-level off-axis positions would let JUNO calibrate without mechanical control.","key_machinery":"The load-bearing object is the in-situ calibration of the 'effective receiver geometry'. Since the acrylic sphere deforms after filling and the surveyed receiver positions shift by 1–10 cm relative to the photomultiplier-tube reference frame, the paper treats the six receiver coordinates as free parameters in a global fit against time-of-flight measurements to 33 ACU positions that are taken as known truth. This calibrated geometry, combined with a custom laboratory-derived linear sound-speed-versus-temperature relation for the liquid scintillator (1469.69 − 3.916·T m/s over 18–22.5 °C) and a fixed-peak arrival-time algorithm with roughly 0.5 microsecond resolution, is what carries the centi","core_discovery":"The central claim is that the limiting obstacles to acoustic positioning in a large liquid-scintillator detector are not fundamental. By measuring the temperature-dependent sound speed of the scintillator in the laboratory, reconstructing arrival times from a fixed peak in the received waveform, and re-fitting the six effective receiver coordinates in situ using 33 known central-axis emitter positions, the system removes the dominant systematic errors. After calibration, the mean three-dimensional positioning error along the central axis drops from 14.75 cm to 1.23 cm. A simulation of the off-axis CLS plane, smearing receiver coordinates by 0.8 cm, arrival times by 0.5 microseconds, and incl","pith_inferences":["The 1.23 cm central-axis result is measured against the ACU's own mechanical reference, so it validates the full chain but cannot resolve a common-mode error shared by both the ACU reference and the USS reconstruction.","The off-axis 2.40 cm figure rests on the assumption that receiver-coordinate errors remain Gaussian at 0.8 cm outside the calibrated central axis; a future independent off-axis truth reference (such as a CCD camera or cable metrology) could test this directly.","The 5% adjacent-peak misidentification probability, if underestimated, could dominate the tails of the position-error distribution rather than the mean; waveform-level checks on noisy channels would settle it.","The same trigger-aligned time-of-flight approach could be extended to sparse arrays with more than six receivers, where the vertical degeneracy noted by the authors would be further reduced."],"forward_implications":["JUNO's CLS and ROV calibration campaigns can rely on USS-reported positions to build detector response maps over off-axis volumes.","The online variant of the system gives operators real-time position feedback (about 30 s per point), sufficient to steer the source safely and to seed the offline reconstruction.","The 2.40 cm RMSE in the CLS plane corresponds to a relative accuracy of about 0.14% of the acrylic radius, matching the order of magnitude of benchmarks from other large detectors.","Receiver-geometry optimization or adding receivers would reduce the remaining directional weakness, particularly in the vertical direction.","The sound-speed calibration procedure and the in-situ receiver-fitting procedure are transferable to other liquid-scintillator or water detectors with similar geometry."],"fun_headline_variants":["Ultrasonic positioning hits 1.23 cm accuracy in JUNO","Acoustic tracking localizes JUNO calibration sources to 1.23 cm","Sound waves pin JUNO calibration to centimetre level","JUNO's sound-based positioning achieves 1.23 cm error","Ultrasound yields centimetre precision for JUNO calibration"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The 33 central-axis ACU deployments are treated as exactly known positions in the photomultiplier-tube reference frame when fitting the six receiver coordinates, and that fitted geometry is assumed to hold across the off-axis CLS plane with only an independent 0.8 cm Gaussian smearing to represent its uncertainty.","fun_headline_variants_meta":{"raw":{"variants":["Ultrasonic positioning hits 1.23 cm accuracy in JUNO","Acoustic tracking localizes JUNO calibration sources to 1.23 cm","Sound waves pin JUNO calibration to centimetre level","JUNO's sound-based positioning achieves 1.23 cm error","Ultrasound yields centimetre precision for JUNO calibration"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000203,"raw_usage":{"total_tokens":1202,"prompt_tokens":700,"completion_tokens":502,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":444,"completion_tokens_details":{"reasoning_tokens":408}},"tokens_in":444,"tokens_out":502,"duration_ms":6422,"temperature":1.0,"reasoning_tokens":408,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T10:02:49.204853+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deploy the CLS source at a set of points whose true positions are determined independently (e.g., by the calibrated cable-loop metrology or a CCD camera system), reconstruct the positions with the ultrasonic system, and compare the three-dimensional residuals across the full plane. If the RMS deviation exceeds the claimed 2.40 cm, or if the residual pattern correlates with the central-axis calibration positions, the off-axis extrapolation is not supported.","supporting_citations":[],"review_version":1}