{"id":"52db5629-9b56-4566-8ae1-e37d24c84a28","arxiv_id":"2509.03667","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Using real metro IP latency data and Lindblad decoherence models, the paper shows that BBPSSW and DEJMPS purification have sharp latency limits, with DEJMPS often delivering far higher throughput.","lead":"This paper simulates entanglement purification over real-world IP network delays and maps where it improves or degrades quantum state fidelity. It gives network engineers practical latency budgets and resource overhead estimates for future quantum networks.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"All quantitative break-even thresholds and rate claims are computed at a single arbitrary F0=0.75; without a sensitivity sweep the reported latency budgets are conditional.","rationale":"I read the paper in good faith and the core modeling is sound: the Lindblad treatment with T1/T2 parameters is standard, the BBPSSW analytic formula in Eq. (20) is correct, the expected-pair-consumption expression in Eq. (24) is a correct telescoping product for the stated symmetric scheduling, and the appendix provides convergence checks for the single-step integrator. The qualitative finding that DEJMPS outperforms BBPSSW is consistent with the known marginal advantages in fidelity gain and success probability and is not in doubt. The load-bearing weakness is exactly the one identified by the reader: the entire quantitative map is computed for a single F0=0.75, with no sensitivity analysis, while the break-even curves are defined relative to F0 and all latency thresholds inherit this dependence. This does not make the paper incorrect, but it makes the specific design rules (e.g., DEJMPS/QKD viable up to 20-25 ms) conditional on an unvalidated parameter choice. A F0 sweep would settle whether the qualitative structure is robust and would let the authors state their thresholds as ranges rather than point values. I do not see a reason to move the verdict from CONDITIONAL; the appropriate response is to keep the conditional verdict pending the requested sensitivity check.","tokens_in":18737,"tokens_out":9200,"duration_ms":102422,"concrete_test":"Recompute the full pipeline for the 40Ca+ platform (Figs. 4-6) for F0 in {0.55, 0.60, 0.65, 0.70, 0.75, 0.80, 0.85}, keeping all other parameters fixed, and record (a) the DEJMPS latency band for which F >= 0.81 after any purification round, (b) the BBPSSW/DEJMPS break-even latency at F=F0, and (c) R(Fth=0.81) at 5 ms and 20 ms. If the 20-25 ms DEJMPS QKD viability band shifts by more than about 5 ms, or the 'orders of magnitude' DEJMPS advantage over BBPSSW shrinks to less than a factor of 10 at low latency, then the reported break-even map is F0-specific and the paper's unconditional design rules are not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central deliverable is a quantitative map of sharp break-even iso-fidelity contours and latency thresholds (e.g., DEJMPS viable for QKD up to 20-25 ms on 40Ca+ in Section IV-A). Every panel in Figs. 4-6 is generated for one fixed initial fidelity, F0=0.75, justified only as 'a highly conservative worst-case scenario' (Section IV). The break-even contour in Fig. 5 is literally F=F0, so F0 is not a nuisance parameter: it defines the reference level against which purification success is measured. If actual post-verification fidelities are lower (say 0.60, still above the 1/2 purification threshold), the no-gain region expands and the maximum tolerable latency for QKD/DQC shrinks; if higher, the map is too pessimistic. The paper provides no sensitivity analysis, no error bars, and no alternative F0 values, so the headline thresholds and rate curves cannot be assessed for robustness. The rate model in Eq. (25) also assumes immediate initiation of purification once pairs are available, excluding storage decoherence during finite-rate pair generation; this is a further idealization, but the dominant unquantified parameter for the central break-even map is F0.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies whether entanglement purification (BBPSSW and DEJMPS) is practically beneficial in quantum networks when the classical coordination channel has realistic, finite latency. The authors model qubit storage decoherence with a Lindblad master equation parameterized by T1 and T2 times of several quantum memory platforms, and use measured metropolitan IP network latency statistics. They derive an expected pair-consumption formula E(Fth) = ∏ 2/p(F_i) and a steady-state distillable rate R(Fth) = R_pair/E(Fth). They then numerically map fidelity evolution over purification rounds, break-even iso-fidelity contours (F = F0), and rate curves versus latency for QKD and DQC thresholds. The central quantitative results are all computed for a single initial fidelity F0 = 0.75.","tokens_in":19039,"tokens_out":10509,"duration_ms":113618,"significance":"If the numerical results are correct and robust, the paper provides a useful engineering methodology for assessing whether purification can run over real control-plane latencies, and for setting latency and memory-coherence budgets. Its strengths are that the forward model is standard, the memory parameters and latency data come from external sources, and the main formulas are derived rather than fitted. The paper also clearly identifies that DEJMPS can outperform BBPSSW by orders of magnitude in steady-state rate, and that sharp break-even boundaries exist. The contribution is significant for quantum network systems engineering, though the lack of sensitivity analysis and a likely sign error in Eq. (20) currently limit confidence in the specific quantitative thresholds.","major_comments":[{"comment":"Eq. (20) appears to have a sign error: the numerator should be F^2 + (1-F)^2/9, not F^2 − (1-F)^2/9. With the printed minus sign, F'(0.6) = 0.562 < 0.6, contradicting the claim in Section III-C3 and Fig. 3 that all protocols give F' > F for F > 1/2. The correct plus-sign formula crosses break-even at F = 1/2 as stated. This needs to be fixed and the numerical results checked to ensure they use the correct expression.","section":"Section III-C1, Eq. (20)"},{"comment":"All quantitative maps, break-even contours, latency budgets, and rate curves are computed for a single initial fidelity F0 = 0.75. The break-even contour in Fig. 5 is explicitly F = F0, so F0 is not a nuisance parameter but the reference level of the central output. Since the paper advertises concrete design rules (e.g., DEJMPS viable up to 20–25 ms for 40Ca+), a sensitivity analysis over a plausible F0 range (e.g., 0.60–0.85) is necessary. Without it, the reported latency thresholds and rate comparisons are conditional on an arbitrary input.","section":"Section IV, Figs. 4–6"},{"comment":"The rate expression R(Fth) = R_pair/E(Fth) assumes that purification is initiated immediately once two base pairs are available, so no storage decoherence is incurred while waiting for the second pair during finite-rate generation. This idealization is not stated as a limitation and can affect the quantitative rate curves, especially at low R_pair. The authors should either incorporate a waiting-time degradation term or explicitly bound the error introduced by this assumption.","section":"Section III-D, Eq. (25)"}],"minor_comments":[{"comment":"Typo: 'invesitgates' should be 'investigates'.","section":"Section II-A"},{"comment":"Typo: 'environmengtal' should be 'environmental'.","section":"Section III-C"},{"comment":"Panel (b) caption says 'Fth = 0.81 required for DQC' but the DQC threshold used throughout the paper is 0.98. Please correct.","section":"Section IV-C, Fig. 6 caption"},{"comment":"The text says latencies are 'sampled from the empirical distribution' and the Introduction claims capture of the 'full stochastic impact' of the control plane, but the results are deterministic per fixed T_C. Please clarify that the latency PDF is used as a source of representative fixed values, not propagated through the model.","section":"Section IV-A"},{"comment":"The notation ΔF' and Δp'_{succ}(F) is inconsistent; in Eq. (22) ΔF' is the fidelity difference, while Eq. (23) defines Δp'_{succ} with a prime. Consider using distinct symbols, e.g., ΔF and Δp_succ.","section":"Section III-C3, Eqs. (22)–(23)"},{"comment":"Figures 5 and 6 report averages over 1024 random initial states but no error bars. Please add error bars or state explicitly that they are smaller than the marker size.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the scope of cs.NI and quantum network systems. The likely sign error in Eq. (20) should be caught before publication; it is probably a typo but it is in a central formula. The lack of F0 sensitivity is the main reason I cannot recommend acceptance now; the authors' claim that F0=0.75 is a 'highly conservative worst-case scenario' needs quantitative support because the break-even contours are defined relative to F0."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a solid systems-level modeling study of entanglement purification under realistic classical latency. It delivers a clear quantitative map of where purification helps, where it hurts, and how fast usable pairs can be delivered; that map is new. But the headline thresholds—e.g., DEJMPS viable up to 20–25 ms for QKD on 40Ca+—are all computed at one arbitrarily chosen initial fidelity, F0=0.75, and there is no sensitivity analysis anywhere. So treat the numbers as conditional until they add a sweep over F0.\n\nWhat is genuinely good: they couple empirical metropolitan IP latency statistics with a Lindblad treatment of T1/T2 decoherence for several memory platforms, and they map break-even iso-fidelity contours and steady-state distillable rates for BBPSSW and DEJMPS. The forward model is standard and internally consistent, the expected-pair-consumption formula is derived cleanly, and the RK4 convergence check in Appendix A is the right kind of sanity check. The finding that DEJMPS consistently outperforms BBPSSW, by orders of magnitude at low latency, is consistent with the known marginal advantage snowballing through rounds.\n\nThe soft spots are real but not load-bearing. The fixed F0 is the major one: the break-even contour is literally F=F0, so the entire phase map is anchored to that assumed value. They call it a 'highly conservative worst-case,' but give no justification and no alternative values. A lower F0 would shrink the viable latency windows; a higher one would expand them. Second, the rate model in Eq. (25) assumes purification starts immediately when pairs are available, ignoring decoherence during finite-rate pair generation; that biases rates upward at low Rpair. Third, they average over 1024 random initial states but show no error bars or spread, and no code or data is released, so independent verification of the specific numbers is not possible.\n\nNone of this undermines the qualitative story. The central argument—that finite classical latency shifts the purification viability boundary in a predictable way, and that DEJMPS is the better choice under realistic latency—holds up.\n\nFor network engineers deciding latency budgets and memory targets, this is useful. For quantum networking researchers, it is a reasonable baseline to cite. It deserves a serious referee; the requested revision should include a sensitivity sweep over F0 and error bars, and ideally an artifact release.\n\nRecommendation: engage with it, but don't take the specific latency thresholds at face value until the F0 sensitivity is addressed.","headline":"Solid systems-level modeling paper with new break-even and throughput maps for purification under real latency, but all headline thresholds rest on a single arbitrarily chosen F0=0.75, so treat the numbers as conditional.","tokens_in":19480,"tokens_out":3575,"would_cite":true,"duration_ms":34815,"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":"Classical signaling latency decides whether entanglement purification helps or hurts.","keywords":["entanglement purification","quantum networks","classical communication latency","DEJMPS protocol","BBPSSW protocol","quantum memory decoherence","fidelity thresholds","distillable entanglement rate"],"falsifier":"Re-run the heatmaps and rate calculations with F0 = 0.70 and F0 = 0.85 for the 40Ca+ platform and check whether the DEJMPS QKD break-even latency moves outside the 20–25 ms band. Alternatively, a laboratory demonstration of repeated DEJMPS rounds at 20 ms one-way classical latency on a 40Ca+ memory starting from fidelity 0.75 would either reach or miss F >= 0.81 and settle the viability claim directly.","tokens_in":18685,"feed_emoji":"⚛️","tokens_out":6423,"duration_ms":65460,"temperature":0.7,"pith_summary":"Quantum networks need entangled pairs, but those pairs decay while stored in quantum memory. Entanglement purification can restore fidelity, yet the protocol itself requires classical messages between nodes, and the wait exposes the pairs to more decoherence. This paper asks when purification is a net win, a break-even, or an active loss under realistic internet-style latency. Using a Lindblad decoherence model with measured memory lifetimes and a metropolitan IP latency distribution, it maps sharp break-even contours in the plane of latency versus resource cost and shows that the DEJMPS protocol can deliver orders of magnitude more high-fidelity pairs than BBPSSW where purification works. The practical payoff is a set of design rules: latency budgets, memory coherence targets, and pair-consumption costs for applications such as QKD and distributed quantum computing.","feed_headline":"Entanglement purification fails past a sharp latency cutoff","feed_subtitle":"Map of when purification beats decoherence on real IP delays, and why DEJMPS throughput leads BBPSSW by orders of magnitude.","key_machinery":"The load-bearing object is the break-even iso-fidelity contour in the plane of classical latency versus expected pair consumption: a curve along which the final fidelity after purification equals the starting fidelity F0, separating configurations where purification helps from those where it actively hurts. Inside the model, continuous-time decoherence of idling qubits is captured by a Lindblad master equation with amplitude-damping and dephasing rates derived from measured T1 and T2 times, and the multi-round resource cost is the product E(F_th) = prod_i 2/p(F_i) of inverse success probabilities. The steady-state distillable rate R(F_th) = R_pair/E(F_th) turns that cost into the throughput","core_discovery":"Treating purification as a race between fidelity gained per successful round and fidelity lost while qubits idle during classical signaling, the paper finds sharp operational boundaries. For the two-pair protocols BBPSSW and DEJMPS, starting from a conservative initial fidelity F0 = 0.75 and propagating states under Lindblad dynamics during control-plane waits on a metro IP network, purification either lifts fidelity to high plateaus, flattens below application thresholds, or collapses toward the maximally mixed state F = 1/4. The break-even iso-fidelity contour F = F0 divides the regime where purification pays from a no-gain region where consuming more pairs is worse than doing nothing. The","pith_inferences":["Because F0 = 0.75 is a single fixed worst-case input, the specific latency cutoffs are point estimates; if real initial fidelities are higher, viable latency windows widen and pair-consumption costs drop, and a sensitivity sweep over F0 would quantify that shift.","The model assumes symmetric one-way latency and conservatively takes the higher of the two directions; real asymmetric or bursty IP delays could make purification success heterogeneous across rounds and alter the break-even contours.","The resource metric E(F_th) assumes symmetric scheduling that only pairs states of matching fidelity; a scheduler that pools unequal-fidelity pairs, or exploits DEJMPS's tolerance for non-Werner input states, could lower consumption at a given latency.","The same latency-versus-coherence race applies to other feedback-based quantum network operations, such as entanglement swapping and repeater chains, so the break-even-contour methodology transfers beyond purification."],"forward_implications":["Network architects receive quantitative latency budgets: for 40Ca+ memories with F0 = 0.75, DEJMPS sustains QKD-grade fidelity up to roughly 20–25 ms one-way latency, while DQC-grade fidelity demands the 0–5 ms band.","There is a well-defined no-gain region below the F = F0 break-even contour, where purification degrades fidelity toward 1/4 and operators should use pairs directly rather than purify.","DEJMPS is the more reliable protocol in practice: its small per-round advantages compound, yielding steady-state distillable rates often orders of magnitude above BBPSSW in the latency range where purification works.","Longer-coherence memories shift the positive-rate region to higher latencies; for rare-earth-ion memories, DEJMPS keeps QKD-grade fidelity across the full 0–50 ms range studied.","High-fidelity thresholds are exponentially expensive in pair consumption, so even a theoretically attainable threshold can be operationally infeasible."],"supporting_citations":[{"why":"Defines the BBPSSW purification protocol and supplies the analytic fidelity recursion used in Eq. (20).","marker":"[3]"},{"why":"Defines the DEJMPS purification protocol, the paper's preferred protocol throughout the study.","marker":"[4]"},{"why":"Prior control-plane protocol work that first quantified latency-induced storage decoherence; the present model builds on it.","marker":"[9]"},{"why":"Supplies the empirical metropolitan IP network latency distribution from which one-way signaling delays are sampled.","marker":"[10]"},{"why":"Provides measured T1 and T2 lifetimes for the 40Ca+ ion-trap memory used as the primary benchmark.","marker":"[13]"},{"why":"Provides coherence times for the rare-earth-ion memory, the longest-coherence platform considered.","marker":"[15]"},{"why":"Supplies the twirling construction and comparison of fidelity yields for BBPSSW and DEJMPS protocols.","marker":"[22]"},{"why":"Defines symmetric purification scheduling, the basis of the resource-consumption formula E(F_th).","marker":"[23]"},{"why":"Gives the entangled-photon generation rate used to set R_pair in the throughput calculation.","marker":"[25]"}],"fun_headline_variants":["Purification loses to decoherence past a latency cutoff","Break-even contour: when entanglement purification stops paying","Latency cutoff determines if purification beats decoherence","DEJMPS throughput leads BBPSSW by orders of magnitude","Quantum purification has a hard latency budget on IP nets"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"Every quantitative boundary in the paper is computed at a single initial fidelity F0 = 0.75; if the real starting fidelity of deployed entangled pairs differs, all latency cutoffs and rate curves shift, and the paper does not quantify that sensitivity.","fun_headline_variants_meta":{"raw":{"variants":["Purification loses to decoherence past a latency cutoff","Break-even contour: when entanglement purification stops paying","Latency cutoff determines if purification beats decoherence","DEJMPS throughput leads BBPSSW by orders of magnitude","Quantum purification has a hard latency budget on IP nets"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000641,"raw_usage":{"total_tokens":2788,"prompt_tokens":744,"completion_tokens":2044,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":488,"completion_tokens_details":{"reasoning_tokens":1979}},"tokens_in":488,"tokens_out":2044,"duration_ms":16377,"temperature":1.0,"reasoning_tokens":1979,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:45:18.625324+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the heatmaps and rate calculations with F0 = 0.70 and F0 = 0.85 for the 40Ca+ platform and check whether the DEJMPS QKD break-even latency moves outside the 20–25 ms band. Alternatively, a laboratory demonstration of repeated DEJMPS rounds at 20 ms one-way classical latency on a 40Ca+ memory starting from fidelity 0.75 would either reach or miss F >= 0.81 and settle the viability claim directly.","supporting_citations":[{"cited_title":"Purification of noisy entanglement and faithful teleportation via noisy channels,","cited_arxiv_id":null,"evidence_quote":"Defines the BBPSSW purification protocol and supplies the analytic fidelity recursion used in Eq. (20)."},{"cited_title":"Quantum privacy amplification and the security of quantum cryptography over noisy channels,","cited_arxiv_id":null,"evidence_quote":"Defines the DEJMPS purification protocol, the paper's preferred protocol throughout the study."},{"cited_title":"Control Protocol for Entangled Pair Verification in Quantum Optical Networks","cited_arxiv_id":"2411.07410","evidence_quote":"Prior control-plane protocol work that first quantified latency-induced storage decoherence; the present model builds on it."},{"cited_title":"Speedtest by ookla global fixed network performance—2024- 01-01 fixed tiles data,","cited_arxiv_id":null,"evidence_quote":"Supplies the empirical metropolitan IP network latency distribution from which one-way signaling delays are sampled."},{"cited_title":"Experimental and theoretical study of the 3d d 2-level lifetimes of 40ca+,","cited_arxiv_id":null,"evidence_quote":"Provides measured T1 and T2 lifetimes for the 40Ca+ ion-trap memory used as the primary benchmark."},{"cited_title":"Coherence time of over a second in a telecom-compatible quantum memory storage material,","cited_arxiv_id":null,"evidence_quote":"Provides coherence times for the rare-earth-ion memory, the longest-coherence platform considered."},{"cited_title":"Mixed-state entanglement and quantum error correction,","cited_arxiv_id":null,"evidence_quote":"Supplies the twirling construction and comparison of fidelity yields for BBPSSW and DEJMPS protocols."},{"cited_title":"System design for a long-line quantum repeater,","cited_arxiv_id":null,"evidence_quote":"Defines symmetric purification scheduling, the basis of the resource-consumption formula E(F_th)."},{"cited_title":"Visible- wavelength polarization-entangled photon source for quantum commu- nication and imaging,","cited_arxiv_id":null,"evidence_quote":"Gives the entangled-photon generation rate used to set R_pair in the throughput calculation."}],"review_version":1}