{"id":"b179e4c1-9364-47e0-8958-9431f8e2a9f9","arxiv_id":"2411.14757","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"A multiplexed continuous-variable quantum repeater using cat codes, with quantum memories or graph states, increases the secret key rate by up to 11 orders of magnitude compared to the single-channel version.","lead":"This paper proposes adding multiple parallel channels, plus either quantum memories or graph states, to a continuous-variable quantum repeater that uses cat codes, boosting the secret key rate by many orders of magnitude. The work suggests a practical path to long-distance quantum key distribution without frequency converters.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Graph-state SKR formula Eq. (11) is under-supported: the explicit m=2 example in Appendix B implies at least 2mn logical-Z measurements, not mn, and the fidelity equivalence is only shown for m=2.","rationale":"The reader's conditional acceptance is well calibrated. Figure 4 and Eq. (10) for the QM protocol are the source of the headline improvement, and the paper gives a plausible multiplexing mechanism. However, the graph-state protocol — whose advertised advantage is removing quantum memories — is the weakest link. Equation (11) inherits a measurement-counting assumption (km=mn) that the paper's own m=2 example in Appendix B appears to contradict: it performs logical-Z measurements on all four modes, suggesting km=2mn if n=1. Because P_USD<1, this introduces an exponential reduction in R_QKD,3 that is far larger than the claimed orders-of-magnitude gain. Independently, Appendix B's density-matrix construction uses amplitude products where squared amplitudes are required, so even the m=2 fidelity result should be rederived. A focused re-derivation of the m=2 example — counting measurements and normalizing the postselected state correctly — followed by the same calculation for m=3 would settle whether the graph-state variant retains any advantage. If the count is indeed 2mn, the graph-state SKR curves in Figs. 5 and 7 need revisiting; the QM-based results may survive, but the 'no quantum memories' variant would not be supported in its current form. This concern is about the argument, not the authors; it is a standard check of a central formula.","tokens_in":18933,"tokens_out":28199,"duration_ms":275708,"concrete_test":"Take the explicit m=2, n=1 case in Appendix B and write the full sequence of logical-Z USD measurements: removal of modes 2 and 3 plus conversion of modes 1 and 4. Count them and compare with the exponent in Eq. (6). Separately, recompute Eq. (B18) with density-matrix weights |ClCi|^2 instead of ClCi, then repeat the derivation for m=3 and n=1. If the count is 4 or the fidelity changes, update Eq. (11) and regenerate the graph-state curves in Figs. 5 and 7.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The graph-state claim rests on Eq. (11), where Ptz = (P_USD pm)^(mn). In the explicit m=2 case of Appendix B, a four-mode graph is used; after syndrome selection the two undesired nodes (modes 2 and 3) are removed by logical-Z measurements, and the two surviving nodes (modes 1 and 4) are each measured in logical Z in the subsequent entanglement-creation step. That is four logical-Z USD measurements. If this represents one elementary link, the exponent in Eq. (6) should be 2mn, not mn, and R_QKD,3 is overestimated by (P_USD pm)^(mn) — about 10^(-16) for the Fig. 4 parameters. If the four-mode graph is meant to represent two links, the paper must say so, but then the fidelity equivalence of Appendix B is still shown only for m=2 and not for the general m used in Eq. (11). Either way, the graph-state SKR and the statement that it eliminates quantum memories without sacrificing fidelity are not established by the text.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes two multiplexed continuous-variable quantum repeater protocols based on cat codes: one using a few quantum memories at the elementary stations, and one replacing those memories with graph states whose nodes are logical cat-code states. For each protocol, the authors derive a secret-key-rate formula (Eqs. 10 and 11), including the probability of desired syndrome outcomes, the success probability of the logical-Z unambiguous state discrimination, and, for the memory-based protocol, the effect of finite coherence time modeled as dephasing or depolarizing noise. The central claimed results are that a small number of channels per elementary link (m=2,3,4) raises the SKR by many orders of magnitude relative to the single-channel cat-code repeater, that the graph-state version can reach the same fidelity as the no-loss case while removing the need for quantum memories, and that the 3-loss cat-code implementation becomes competitive with a third-generation discrete-variable repeater while using fewer optical resources.","tokens_in":19202,"tokens_out":22025,"duration_ms":217754,"significance":"If the claims are correct, the paper is significant: it would show that cat-code repeaters, which are experimentally accessible at optical frequencies without frequency conversion, can reach performance comparable to leading DV repeater proposals with far fewer resources. The paper contains explicit analytical derivations of graph-state generation and of the m=2 fidelity equivalence, and the SKR formulas are derived from the protocol assumptions and standard QKD secret-fraction expressions rather than fitted to the claimed improvement. The main quantitative claims are, however, sensitive to two points that are not fully established: the fidelity equivalence for the graph-state protocol is proven only for a special m=2 case, and the counting of logical-Z measurements that enters Eq. (11) is stated in a way that is easy to misread. Both issues are fixable, but they are load-bearing for the graph-state protocol's performance claims.","major_comments":[{"comment":"The fidelity equivalence that underlies F_tot,3 in Eq. (11) is demonstrated only for the special case m=2, and only for the post-selection pattern in which exactly one node on each side of the graph has the desired even-loss syndrome (modes 1 and 4) and exactly one node on each side has odd loss (modes 2 and 3). The protocol as described in Section II.B accepts any outcome with at least one desired syndrome per side, and Eq. (4) sums over all such events; cases with multiple desired nodes per side or with m>2 are not analyzed, and the graph state for general m is not specified. Because Eq. (11) and Figures 5 and 7 use F_tot,3 for optimized values of m, the graph-state secret-key rates and the comparison with Ref. [40] rest on an unproven generalization. Please provide a derivation for general m or state and justify this as an assumption.","section":"Appendix B / Eq. (11)"},{"comment":"The counting km=nm for the graph-state protocol is internally consistent only if n is the number of directed ES1-to-ES2 elementary links, so that the four-mode graph in Appendix B represents one ES1 with two incident elementary links (left and right), each carrying m=2 channels, and hence 2m=4 logical-Z measurements in total. The text defines n=Ltot/L0 as \"the number of elementary links\" but does not specify whether an elementary link is a directed ES1-ES2 segment or an ES2-ES2 connection; under the latter reading, each ES2 receives m nodes from each adjacent ES1 and the exponent in Eq. (6) would be 2mn, not mn. Because Eq. (11) is exponentially sensitive to this exponent, the definition must be made explicit.","section":"Section II.C, Eq. (6), and Appendix B"},{"comment":"The manuscript never states the explicit expression for F_tot,3 used in the numerical evaluation of Eq. (11). Appendix B proves only a single-link fidelity equivalence for m=2; the multi-link total fidelity after n entanglement swaps, the values of F0, Pdsm, eta, t0, and the optimized m and alpha used in Figures 5 and 7 are not given. Without these, the graph-state curves and the claimed comparison with the third-generation repeater of Ref. [40] are not reproducible. Please include the full parameter table and the formula for F_tot,3.","section":"Eq. (11) and Figures 5/7"}],"minor_comments":[{"comment":"In Eq. (B6), the action of A_{2m+1} on |bar{1}> is listed twice; the second line should act on |bar{0}> and yield |tilde{0}_->, matching the relations stated immediately after Eq. (B9).","section":"Appendix B, Eq. (B6)"},{"comment":"There are numerous typos, including \"Eqation\" near Eqs. (B15)-(B16), \"normaliza=on constant\" after Eq. (B11), \"to to\" in Section IV, and \"sold\" for \"solid\" in the Figure 7 caption; these should be corrected.","section":"Throughout"},{"comment":"The captions should state which protocol (QMs or graph states) is plotted, whether pm=1 in Figure 4, and the optimized values of m and alpha used in Figure 5; currently the text refers to these only indirectly.","section":"Figures 4 and 5"},{"comment":"The phrase \"at least one syndrome measurement will give the desired outcome\" in the derivation of Eq. (4) should be \"at least one channel in each elementary link\", since the probability is for a per-link event.","section":"Section II.C, Eq. (4)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the central idea is promising. The factor-of-two issue raised in the stress test appears to be resolvable by clarifying that n counts directed elementary links; the deeper problem is the unproven generalization of the graph-state fidelity to general m and to post-selection patterns with more than one desired node per side. I believe this is fixable with an extended appendix or an explicit assumption statement, and I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the result that actually carries the paper is the memory-based multiplexing. That part looks right to me, and it is enough to justify the headline order-of-magnitude improvement (Fig. 4). The graph-state variant is the fragile piece: its central claim—that after syndrome selection and removal of undesired nodes the final state is a Bell pair with no-loss fidelity—is proven only for m=2 in Appendix B, then used for arbitrary m in Eq. (11). That's a gap, not a contradiction, so I'd send it to a serious referee but ask for a general proof or a worked example for m=3.\n\nOn the counting worry: I think that's a misread. In the m=2 example, the four-mode graph belongs to one ES1 and spans two elementary links—two nodes travel left, two travel right. Each of the four nodes is measured once in logical Z (two removals, two entanglement-creation measurements), so four measurements per two links, i.e. km = nm. The exponent in Eq. (6) is correct as written.\n\nWhat's new: applying multiplexing to CV cat-code repeaters, using QMs or graph states to pick a favorable syndrome outcome; the cavity-QED construction of photonic graph states with RSBC nodes (Appendix A); and the resource comparison with third-generation DV repeaters. The memory analysis with dephasing and depolarizing noise is standard but applied cleanly, and the trade-off between memory coherence time and measurement error in Fig. 5 is a useful plot.\n\nSoft spots, in proportion: the unproved m-to-general fidelity equivalence is the main one. The comparison with [40] is narrow—one error probability, one code family—so the claim of \"comparable performance\" rests on a single point. There are also typos (Eq. B6 repeats the |1bar> line; \"to to\" in the conclusions) and the graph-state SKR in Eq. (11) should be treated as provisional until the fidelity argument is filled in. None of this undermines the QM result, which stands on its own.\n\nWho this is for: anyone working on CV repeaters or memory-based long-distance QKD. The graph-state construction in Appendix A is the most original part, even if its performance claim needs work. I'd referee it, with the fidelity generalization as a mandatory revision.","headline":"Memory-based multiplexing is the real result; graph-state claims need a general-m proof before they're cited.","tokens_in":19743,"tokens_out":13364,"would_cite":false,"duration_ms":131272,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P45","81P94"],"pacs":["03.67.Hk","03.67.Dd"],"model":"deepseek-v4-flash","headline":"Multiplexing cat-code repeaters with a few extra channels lifts the secret key rate from 10^-14 to 10^-3 bits per channel.","keywords":["Quantum repeaters","Cat codes","Continuous-variable quantum key distribution","Secret key rate","Multiplexing","Quantum memories","Graph states","Rotation-symmetric bosonic codes"],"falsifier":"Directly compute the graph-state protocol for $m=3$ channels: construct the three-channel graph state, apply the amplitude-damping loss channel to every mode, project onto the desired syndrome outcomes, discard the failed nodes, and check whether the surviving two-mode state has fidelity equal to the no-loss case; at the same time, re-derive Eq. (6) by counting all logical $Z$ measurements in the network, since each ES2 connects nodes from two ES1s and the count could be $2mn$ rather than $nm$.","tokens_in":2159,"feed_emoji":"🔐","tokens_out":2458,"duration_ms":88801,"temperature":0.7,"pith_summary":"Continuous-variable quantum repeaters based on cat codes—bosonic codes whose codewords are superpositions of coherent states—have been experimentally feasible but suffered from secret key rates too low for practical use. This paper claims that the low rate comes from using a single channel and averaging over all syndrome outcomes, and that multiplexing with a few parallel copies of the encoded state fixes the problem. With $m=3$ channels, the secret key rate for 1-loss cat codes rises from roughly $10^{-14}$ to $10^{-3}$ bits per channel, and the protocol becomes comparable to third-generation discrete-variable repeaters while using far fewer optical resources. The paper also presents a graph-state version that avoids quantum memories entirely, at the price of sensitivity to measurement errors, and it identifies a concrete trade-off between memory coherence time and measurement error.","feed_headline":"Three channels turn cat-code repeaters from 10^-14 to 10^-3","feed_subtitle":"A few parallel encoded copies and a graph-state option match third-generation repeaters with far fewer resources.","key_machinery":"The load-bearing construction is the multiplexed elementary link: $m$ identical cat-coded light modes travel in parallel between ES1 and ES2 stations, and at each ES2 a syndrome measurement on ancillary qubits selects the desired photon-loss outcome. In the memory-based variant, matter qubits at ES1 wait in quantum memories until classical communication identifies successful channels, giving repetition time $t_{r,2}=\\max\\{t_0,2L_0/c\\}$. In the graph-state variant, each ES1 holds a continuous-variable graph state whose nodes are cat-coded modes; measuring failed nodes in the logical $Z$ basis removes them and leaves a Bell pair among the survivors, so no memory and no waiting time are needed and $t_{r,3}=t_0$. The mathematical core is the Appendix B fidelity analysis: after syndrome selection and node removal, the surviving state is equivalent to the no-loss transmitted state, giving the graph-state secret key rate $R_{\\rm QKD,3}=[1-(1-P_{\\rm dsm})^m]^n(P_{\\rm USD}p_m)^{mn}[1-h(1-F_{\\rm tot,3})]/t_0$, with $k_m=mn$ logical $Z$ measurements counted for graph states and $k_m=n$ for the quantum-memory version.","core_discovery":"The central claim is that the poor performance of cat-code repeaters is not intrinsic to the cat code but a consequence of single-channel operation, which forces a compromise value of the mean photon number $\\alpha$ across different loss outcomes. By sending $m$ identical cat-coded light modes in parallel between elementary stations and selecting only channels whose syndrome measurement gives the desired even (or odd) photon-loss number, the paper obtains a large success-probability boost and can fix $\\alpha$ at its optimal value. For the memory-based version, quantum memories at the ES1 stations hold matter qubits until classical communication announces which channels succeeded; decoherence is modeled by depolarizing or dephasing channels. For the memory-free version, a continuous-variable graph state whose nodes are cat-coded modes is created through cavity-QED light-matter interactions; after nodes with undesired syndrome outcomes are removed by logical $Z$ measurements, the remaining state is claimed to be a Bell pair with fidelity equal to the no-loss case. With 3-loss cat codes and $0.1\\%$ measurement error, the paper reports secret key rates around $2\\times10^5$ bits/s at 100 km and $7\\times10^3$ bits/s at 1000 km, comparable to a third-generation discrete-variable repeater that encodes in hundreds of photons.","pith_inferences":["If the Appendix B fidelity equality holds for all $m$, the graph-state protocol becomes a memory-free multiplexed repeater whose rate benefits directly from more parallel channels; a comparison with other multiplexed continuous-variable schemes at fixed total photon number would show where the advantage saturates.","The paper's counting of logical $Z$ measurements in the graph-state rate, $k_m=nm$, may undercount: each ES2 receives nodes from two ES1s, so the number could be $2mn$; if so the plotted graph-state secret key rate would fall by a factor $(P_{\\rm USD}p_m)^{mn}$, which would shrink but not necessarily erase the multiplexing gain for small $m$.","The assumption that at least one channel per elementary link succeeds, $[1-(1-P_{\\rm dsm})^m]^n$, implicitly assumes independent channels and no correlated losses; a natural extension is to test whether realistic fiber correlations or detector dark counts change the optimal $m$ and $\\alpha$.","Since the authors note that local losses accumulate at every elementary station, adding a local-loss term to $F_{\\rm tot}$ for both variants and recomputing the Figure 5 crossing points would show how the graph-state advantage degrades when local errors dominate."],"forward_implications":["For 1-loss cat codes, multiplexing with $m=3$ channels raises the secret key rate from about $10^{-14}$ to $10^{-3}$ bits per channel at $\\alpha\\approx1.268$, more than ten orders of magnitude, using only a few extra channels.","The graph-state version eliminates quantum memories and the associated waiting time, setting the repetition time to $t_0$ rather than $\\max\\{t_0,2L_0/c\\}$, and, under the Appendix B argument, still yields a Bell pair with the no-loss fidelity.","With 3-loss cat codes and $0.1\\%$ measurement error, the protocol reaches about $2\\times10^5$ bits/s at 100 km and $7\\times10^3$ bits/s at 1000 km, comparable to a third-generation discrete-variable repeater that requires hundreds of photons.","For a fixed secret key rate, the protocol tolerates quantum-memory coherence times around one minute under depolarizing noise and roughly three orders of magnitude smaller under dephasing noise, trading off against gate measurement error.","At a fixed cost coefficient $C'=100$, the multi-channel cat-code repeater permits longer elementary distances than the binomial-code single-channel repeater, reducing the number of stations and auxiliary devices along a 1000 km link."],"supporting_citations":[{"why":"Defines the memoryless cavity-QED repeater protocol and the rotation-symmetric bosonic code encoding that this paper modifies by adding multiplexing.","marker":"[46]"},{"why":"Showed that cat codes underperform other rotation-symmetric bosonic codes in that base protocol; this paper starts from that low-rate result.","marker":"[47]"},{"why":"Provides the experimentally demonstrated deterministic creation of atom-light Schrödinger-cat states used as the cat-code nodes.","marker":"[52]"},{"why":"The third-generation discrete-variable one-way repeater used as the comparison baseline in the rate and resource figures.","marker":"[40]"},{"why":"The BBM92 quantum key distribution protocol whose secret-key-rate formula the paper evaluates.","marker":"[8]"},{"why":"Bounds the unambiguous state discrimination success probability used in the logical $Z$ measurement success factor $P_{\\rm USD}$.","marker":"[61, 62]"},{"why":"Supplies the cost-coefficient definition $C=N_{\\rm tot}/R_{\\rm QKD}$ used in the resource comparison of the cat-code and binomial-code schemes.","marker":"[3]"}],"fun_headline_variants":["Multiplexed cat codes lift QKD rates from 10^-14 to 10^-3","Parallel cat-coded modes boost secret key rates by 10^11","Graph-state cat codes deliver high-rate QKD with minimal resources","Three parallel channels make cat-code repeaters viable at 1000 km"],"cache_read_input_tokens":21888,"weakest_assumption_plain":"The whole graph-state result rests on the unproven-in-general assumption that the same 'remove bad nodes and get a perfect Bell pair' trick works for any number of parallel channels; the paper demonstrates this explicitly only for two channels, and its counting of the number of repeated measurements may be off by a factor of two.","fun_headline_variants_meta":{"raw":{"variants":["Multiplexed cat codes lift QKD rates from 10^-14 to 10^-3","Parallel cat-coded modes boost secret key rates by 10^11","Graph-state cat codes deliver high-rate QKD with minimal resources","Three parallel channels make cat-code repeaters viable at 1000 km"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000882,"raw_usage":{"total_tokens":3805,"prompt_tokens":932,"completion_tokens":2873,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":548,"completion_tokens_details":{"reasoning_tokens":2790}},"tokens_in":548,"tokens_out":2873,"duration_ms":21650,"temperature":1.0,"reasoning_tokens":2790,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:56:11.813131+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Directly compute the graph-state protocol for $m=3$ channels: construct the three-channel graph state, apply the amplitude-damping loss channel to every mode, project onto the desired syndrome outcomes, discard the failed nodes, and check whether the surviving two-mode state has fidelity equal to the no-loss case; at the same time, re-derive Eq. (6) by counting all logical $Z$ measurements in the network, since each ES2 connects nodes from two ES1s and the count could be $2mn$ rather than $nm$.","supporting_citations":[{"cited_title":"Continuous-Variable Multiplexed Quantum Repeater Networks","cited_arxiv_id":"2411.14757","evidence_quote":"Showed that cat codes underperform other rotation-symmetric bosonic codes in that base protocol; this paper starts from that low-rate result."},{"cited_title":"Lo Franco, G","cited_arxiv_id":null,"evidence_quote":"Provides the experimentally demonstrated deterministic creation of atom-light Schrödinger-cat states used as the cat-code nodes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The BBM92 quantum key distribution protocol whose secret-key-rate formula the paper evaluates."},{"cited_title":"Muralidharan, J","cited_arxiv_id":null,"evidence_quote":"Supplies the cost-coefficient definition $C=N_{\\rm tot}/R_{\\rm QKD}$ used in the resource comparison of the cat-code and binomial-code schemes."}],"review_version":1}