{"id":"afd7e814-bf3f-4bc4-8272-193ca62b809c","arxiv_id":"2607.27906","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Two C-band entangled-photon sources stayed temporally synchronized for 8 hours over a real 7-km metropolitan fiber loop, with ~120 ps worst-case drift.","lead":"This paper reports an 8-hour experiment in Naples in which two entangled-photon sources stayed synchronized over lab and 7-km metropolitan fiber links, with the largest drift about 120 ps. It matters because quantum-network nodes need stable clocks to make photons indistinguishable for entanglement swapping.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Picosecond drift claims lack uncertainty quantification: without per-fit errors, the reported 12–120 ps drifts may be below statistical resolution.","rationale":"The reader's weakest assumption and my load-bearing concern are the same: the quantitative drift and width claims are extracted from Gaussian fits with no uncertainty quantification. This is not a manufactured issue—the paper itself reports values to 0.01 ps and 0.01 ps variation, which is impossible to assess without knowing the coincidence counts and fit covariance. I considered other potential concerns (e.g., the overclaim of 'entanglement distribution' without HOM/BSM, the manual shifting of peaks, the absence of raw data/code), but those are secondary: even if the experiment is qualitatively correct, the specific quantitative claim of sub-150 ps stability is only as strong as the fit precision. The proposed bootstrap test is concrete and would settle whether the reported 120 ps drift exceeds statistical noise. Since the reader's verdict is already CONDITIONAL and my concern reinforces that condition, the verdict should remain UNCHANGED; the paper should be accepted only after the authors supply uncertainty intervals or the data needed to compute them.","tokens_in":10133,"tokens_out":2151,"duration_ms":22030,"concrete_test":"For the CH1-CH4 (5 km spool) and CH1-CH5 (metropolitan loop) data, take each 5-minute coincidence histogram, treat each bin as Poisson, and generate 1000 bootstrap resamples. Refit the Gaussian (or use a robust centroid) to each resample to obtain a distribution of fitted means and widths, and compute 95% confidence intervals. Then fit a weighted linear slope to the time series of means using these per-fit uncertainties, and report the slope confidence interval. If the 95% CI for the CH1-CH4 slope includes 0 ps/h, the headline '120 ps drift' is not statistically supported; if the CI on width variation for CH1-CH2 includes 0, the claim of minimal broadening is likewise unsupported. This test requires only the raw time-tag data or per-bin counts, which the paper should make available.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—two independent SPDC sources remain synchronized within ~120 ps over 8 h across deployed metropolitan fiber—rests on Gaussian fits to 5-minute coincidence histograms. The paper reports fitted means and widths to two decimal places (Table I) but provides no count statistics, no per-fit uncertainties, no goodness-of-fit metrics, and no confidence intervals on the linear slopes in Fig. 7a. This matters because the measured correlation width is ~110 ps, dominated by SNSPD jitter. The statistical error on a fitted mean is approximately sigma/sqrt(N_eff); resolving a 12 ps drift from a 110 ps peak requires N_eff > ~85 (for a 95% CI excluding zero), and resolving a 4 ps width change requires far more. The paper does not state coincidence counts per 5-min bin, so the reader cannot verify that the reported drifts and broadenings exceed fit noise. The 120 ps maximum drift for the 5 km spool (CH1-CH4) is especially vulnerable: if the per-fit mean uncertainty is comparable to 30 ps, a single 8-hour linear slope has a confidence interval that likely includes zero. The asymmetry noted in Fig. 9 and the lack of HOM/BSM evidence are secondary; the load-bearing assumption is that the fitting procedure has picosecond-level statistical precision, which is never demonstrated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental study of clock-synchronization stability for two rack-mounted SPDC entangled-photon sources sharing a common electrical clock. One photon from each source is detected after propagation over short laboratory links, 500-m and 5-km fiber spools, and a deployed 7-km metropolitan loop. Using 5-minute coincidence histograms, the authors fit Gaussian profiles and track the peak mean and interquartile width over an 8-hour acquisition. They claim that once locked, the temporal correlation peak drifts by at most ~120 ps (mainly on the 5-km spool) and that the correlation width remains consistent with SNSPD jitter, implying negligible additional source or link broadening. The paper also includes a spectral noise characterization of the metropolitan loop to justify the choice of DWDM channels 25 and 43.","tokens_in":10372,"tokens_out":7188,"duration_ms":70008,"significance":"If the quantitative claims survive scrutiny, the result is a useful practical baseline for synchronizing independent entangled-photon sources for future entanglement-swapping nodes on deployed metropolitan fiber. The experiment's strengths are the use of a real operational fiber loop with ~14 dB loss and background noise, a continuous 8-hour acquisition, visible coincidence peaks in the locked condition, and a preliminary spectral noise characterization. The authors do not demonstrate two-photon interference, polarization entanglement, or an entanglement witness; the contribution is a timing-stability characterization rather than a complete interconnection demonstration. The significance is therefore moderate but real, contingent on the reliability of the reported picosecond-level drift and width numbers.","major_comments":[{"comment":"The central quantitative claims—maximum drift ~120 ps, drift rates to two decimal places, width variations of 4–14 ps—are presented with no statistical uncertainties. The manuscript never states the total coincidence counts per 5-minute histogram, count rates, per-fit standard errors, or confidence intervals on the linear slopes in Fig. 7a. With a correlation width of ~110 ps, dominated by SNSPD jitter, resolving a 12 ps mean shift at 95% confidence requires of order a few hundred peak coincidences; resolving a 4 ps width change requires substantially more. Without these numbers, the reader cannot determine whether the reported drifts (e.g., +12.56 ps for CH1-CH2) are statistically distinguishable from zero, or whether the '~120 ps maximum drift' is within fit noise. This is load-bearing for the abstract's stability claim and must be addressed with per-fit errors, count statistics, or ra","section":"Sec. VI / Table I / Fig. 7"},{"comment":"Gaussianity is asserted rather than tested. The manuscript repeatedly states that the correlation histograms are 'well approximated' by Gaussians and uses Gaussian fits to extract mean and IQR, but no goodness-of-fit metric (e.g., reduced chi-square, residual analysis) is reported. Figure 6 shows the metropolitan-loop profile as noticeably jagged, and Figure 9 shows a visibly asymmetric correlation. If the fitted distributions are not Gaussian, the table values and drift slopes may be biased. At minimum, the authors should report fit residuals or compare against nonparametric peak estimators.","section":"Sec. V / Sec. VI / Figs. 3, 6, 8, 9"},{"comment":"Table I lists the initial time shift for CH1-CH2 as +4500 ps, but Sec. IV and Fig. 3 describe a residual delay of approximately 6.9 ns that is subsequently compensated. The table does not state whether this entry is the offset before or after compensation, and the two passages appear inconsistent. This needs clarification, since the 'initial time shift' values enter the interpretation of the channel-specific delays.","section":"Table I vs. Sec. IV"},{"comment":"The abstract and Sec. V state that 'entanglement distribution is performed over existing telecommunication infrastructure' and describe the synchronization of two sources as a 'central result' in that context. However, the experiment measures temporal cross-correlations between one signal photon from EPS1 and one signal photon from EPS2; these two photons are not entangled with each other, and no polarization/energy-time entanglement witness, HOM interference, or BSM is performed. The authors should rephrase to avoid implying that entanglement was distributed between the two sources or that photon indistinguishability was demonstrated.","section":"Abstract / Sec. V"}],"minor_comments":[{"comment":"Footnote 7 states 'as will be clarified in Section V', but Section V precedes this passage in the manuscript. The footnote also seems to contradict the main text: the text says CH1-CH2 was chosen to keep optical path lengths similar, while the footnote says the narrower Gaussian-fit in Fig. 9 arises from different operating conditions. Please revise the cross-reference and clarify the comparison.","section":"Footnote 7"},{"comment":"The sentence 'a variation of approximately 100 ps is observed, estimated as the interquartile range (between P25 and P75) of the distribution' is confusing; it likely means the IQR width is approximately 100 ps, not that the variation is 100 ps. Please rephrase to avoid ambiguity.","section":"Sec. VI, dispersion paragraph"},{"comment":"The histogram figures lack axis labels and legends. The reader cannot tell whether vertical axes are counts or normalized counts, and the color/line styles for 'no lock', 'lock', and 'phase-shifted' conditions are not defined in the captions. Please add clear axis labels and legends.","section":"Fig. 3 / Fig. 6"},{"comment":"The text refers to 'channels 25-43' and 'channels 25 and 43' inconsistently. If the sources are configured on the two specific DWDM channels 25 and 43, the range notation should be avoided to prevent confusion.","section":"Sec. V, channel notation"}],"recommendation":"major_revision","confidential_remarks":"The central experimental concept is reasonable and the qualitative conclusion of stable temporal correlation under an 8-hour locked acquisition appears credible. However, the quantitative drift and width claims are currently unsupported because no count statistics or fit uncertainties are given. I would request the raw coincidence counts and a supplementary uncertainty analysis; without that, the reported picosecond-level numbers are not verifiable. I also suggest the authors align the abstract and title phrasing with the actual measurement (temporal cross-correlation of photons from two synchronized sources) rather than implying full entanglement distribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is an honest engineering report: two commercial SPDC sources locked to one electrical clock maintain visible coincidence peaks across lab fibers, spools, and a 7-km deployed metro loop for eight hours. The noise spectral scan of the loop is a useful public dataset in itself, and the qualitative stability claim is credible from the displayed histograms.\n\nWhat is actually new is the long-duration stability measurement: drift and width evolution over 8 h on deployed fiber, with maximum fitted drift about 120 ps on the 5-km spool. The authors do not introduce new protocols or theory, but they quantify something network designers need. The paper is clearly written and the experimental description is largely reproducible in principle.\n\nThe soft spot is exactly where the reader put it: the quantitative drift and width values are reported to two decimal places (Table I) with no per-fit uncertainties, no coincidence counts, no goodness-of-fit statistics, and no confidence intervals on the linear slopes. Given the measured peak width is ~110 ps and is dominated by SNSPD jitter, a 12 ps drift or a 4 ps width change may be below statistical resolution. The paper does not show that the fits have picosecond-level precision, so the strongest quantitative claims are under-evident. This is a real weakness, not a manufactured one, and it applies to the central '120 ps max drift' number as much as to the smaller ones.\n\nA second issue is framing: the paper says 'entanglement distribution' is performed, but the measurement is a classical temporal cross-correlation between photons from two independent sources. There is no HOM dip, no Bell-state measurement, no state tomography. The temporal correlation is necessary for entanglement swapping, but it is not evidence of distributed entanglement. That claim should be softened.\n\nThe citation pattern is fine. Refs [9] and [10] (which include one of the present authors) already reported synchronized entanglement distribution over deployed fiber; this work is a distinct extension focused on long-term stability. No circular reasoning found.\n\nBottom line: the qualitative result — locked sources stay synchronized well enough that coincidence peaks remain detector-limited over eight hours on metro fiber — is likely right and worth publishing. The quantitative drift claims need uncertainty quantification and, ideally, raw count data. This paper deserves peer review, but it should not be accepted in its current form.\n\nMy recommendation: send to a competent experimental referee; insist on error bars and a re-framing of the entanglement claim.","headline":"Useful 8-hour stability measurement on deployed metro fiber, but the picosecond drift claims lack uncertainty quantification and the 'entanglement distribution' framing overreaches.","tokens_in":10902,"tokens_out":2651,"would_cite":true,"duration_ms":27049,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P45","81V80"],"pacs":["03.67.Hk","42.50.Ex"],"model":"deepseek-v4-flash","headline":"Two entangled-photon sources stay synchronized within 120 picoseconds for eight hours across a deployed city fiber loop.","keywords":["quantum networks","entanglement distribution","clock synchronization","temporal stability","SPDC sources","metropolitan fiber","Hong-Ou-Mandel interference","time-tagged coincidence"],"falsifier":"Re-analyze the raw time-tagged data with Poisson error propagation: if the scatter of successive 5-minute peak positions is comparable to or larger than the claimed drift slopes (12–120 ps), then the sub-150 ps stability claim is not supported by the data.","tokens_in":9969,"feed_emoji":"⏱️","tokens_out":4016,"duration_ms":35366,"temperature":0.7,"pith_summary":"This paper tries to establish that clock synchronization between independent entangled-photon sources can survive the transition from a controlled lab to a working metropolitan fiber network, staying stable enough for entanglement-swapping operations. It reports an experimental demonstration over eight hours in which two SPDC sources, locked by an electrical clock, keep their correlation peak within roughly 120 ps of drift across a 7 km deployed loop and 5 km spools, with no active feedback. The correlation width stays consistent with detector jitter, meaning the sources themselves do not add temporal broadening. A sympathetic reader would take this as concrete evidence that existing telecom fiber, with its losses and noise, can host the timing backbone of a quantum network.","feed_headline":"Quantum sources locked to 120 ps drift for 8 hours on city fiber","feed_subtitle":"Two independent photon sources stay synchronized across a deployed 7 km loop, a step toward entanglement swapping on existing telecom fiber","key_machinery":"The central object is the temporal correlation function built from time-tagged single-photon detection events: a histogram of arrival-time differences between photons from the two sources, accumulated over one 40 ns clock period. A Gaussian fit to each 5-minute histogram supplies the peak position and width; the peak's evolution gives drift, and the interquartile range (P75-P25) of the fit gives temporal dispersion. A shared electrical clock from one source to the other is the locking mechanism, and the metropolitan loop is the stress test for real-world stability.","core_discovery":"The paper reports an experimental demonstration that two independent SPDC-based entangled-photon sources, once locked to a shared electrical clock, maintain stable temporal correlations over an 8-hour acquisition in a real metropolitan fiber loop with ~14 dB loss and background noise. The correlation peak drifts at most about 120 ps over that period, and the correlation width stays near the detector jitter limit (~100 ps interquartile range), implying no significant system-induced broadening. The authors see the result as evidence that clock sharing suffices for entanglement swapping in existing telecom infrastructure.","pith_inferences":["If this stability holds, a natural next step is to measure HOM interference visibility over the same links; the 120 ps drift would predict a quantitative drop in visibility that could be calibrated against the clock.","The method could extend to more than two sources by fanning out the clock, but the paper leaves open how clock-distribution noise scales with fan-out.","Temperature monitoring along the metropolitan fiber might allow predicting or compensating the drift in real time, turning the observed 8-hour limit into an actively corrected longer-term lock.","The reported picosecond-level drift numbers should be treated as upper bounds rather than exact rates until the fit uncertainties are published."],"forward_implications":["Once clock-locked, two independent sources can be used for Hong-Ou-Mandel interference and entanglement swapping without active recalibration for at least eight hours.","The drift grows with fiber length (up to ~120 ps on 5 km and 7 km paths), so longer links set the recalibration interval for a quantum network.","Since the correlation width is detector-limited, better SNSPD timing resolution would directly tighten the achievable synchronization.","Choosing quieter DWDM channels (25/43) in the C-band allows entanglement distribution even where the fiber carries classical network traffic.","The observed drift is consistent with thermo-optic effects in long fibers, suggesting temperature-induced path-length changes are the dominant stability limit."],"fun_headline_variants":["8-hour sync on city fiber: 120 ps drift max","Two photon sources stay locked for 8 hours on urban fiber","Quantum sources keep 120 ps alignment over 8-hour city run","Entangled sources hold timing through 7 km telecom loop","City fiber sync: 120 ps drift over 8 hours"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the Gaussian fits to the 5-minute coincidence histograms have picosecond-level statistical precision, so the reported 12 ps drifts are real signals rather than fit noise.","fun_headline_variants_meta":{"raw":{"variants":["8-hour sync on city fiber: 120 ps drift max","Two photon sources stay locked for 8 hours on urban fiber","Quantum sources keep 120 ps alignment over 8-hour city run","Entangled sources hold timing through 7 km telecom loop","City fiber sync: 120 ps drift over 8 hours"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000449,"raw_usage":{"total_tokens":2115,"prompt_tokens":770,"completion_tokens":1345,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":514,"completion_tokens_details":{"reasoning_tokens":1269}},"tokens_in":514,"tokens_out":1345,"duration_ms":9802,"temperature":1.0,"reasoning_tokens":1269,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T23:14:52.724148+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-analyze the raw time-tagged data with Poisson error propagation: if the scatter of successive 5-minute peak positions is comparable to or larger than the claimed drift slopes (12–120 ps), then the sub-150 ps stability claim is not supported by the data.","supporting_citations":[],"review_version":1}