{"id":"ec85cd5c-80cb-40c6-9d56-1226cdfd862c","arxiv_id":"2412.12680","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A switchless photonic quantum computer architecture is simulated to reach fault tolerance at 12 to 13 dB cluster squeezing, with over 90% of generated GKP qubits above threshold and a 4.8% magic state success rate.","lead":"This paper presents an end-to-end, switchless architecture for fault-tolerant photonic quantum computing based on continuous-variable cluster states, with simulations claiming fault tolerance at 12 to 13 dB of Gaussian squeezing. A general reader might care because the design avoids optical switches, needs only low photon-number resolution, and lays out a concrete path toward a manufacturable photonic quantum computer.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The >90% yield and 12–13 dB threshold are computed after an unreported >95%-fidelity filter and zero-homodyne post-selection (App. 11.3), so the headline numbers are conditional, not end-to-end.","rationale":"The reader's weakest-assumption analysis identifies the same structural weakness I find: the simulation pipeline's unquantified filtering and zero-outcome post-selection. This is the load-bearing issue because the paper's most prominent quantitative claims are a >90% generation probability and a 12–13 dB Gaussian cluster squeezing threshold. Both are computed from samples that pass the unpublished 95%-fidelity filter and exact-zero homodyne conditioning. If the discarded fraction is large, the >90% figure is a conditional rather than unconditional success probability, and the threshold numbers would shift if representative samples including nonzero outcomes and their displacement corrections were used. I do not see a stronger objection in the text. The loss simulations in App. 11.9 and the claim that active displacements are needed are acknowledged by the authors and affect the 'passive/end-to-end' framing, but the reader already notes those; the filter/post-selection issue is more damaging to the headline numbers because it directly changes the denominator of the claimed probability. The architecture is coherent and the balanced 10.2 dB threshold reproduces the literature, which is genuine independent support for the simulation methodology. The concern is addressable with additional simulation output, so it does not warrant rejection. Since the reader already returned CONDITIONAL and my analysis points to the same qualification, no verdict change is needed.","tokens_in":31473,"tokens_out":5975,"duration_ms":54868,"concrete_test":"Re-run the PhANTM Monte Carlo (Sec. 4 and App. 11.3) without the 95%-fidelity filter and without zero-homodyne post-selection, applying the actual feedforward displacements for nonzero outcomes; then recompute the adaptive-breeding GKP squeezing distributions, the Table 1 success rates, and the Sec. 8.2 thresholds. Report the fraction of trials discarded by the fidelity filter and the probability mass of the zero-homodyne conditioning event. If the unconditional success probability at 13 dB and 20 PhANTM steps is materially below the claimed 0.91 (e.g., below 0.5), the abstract's >90% yield claim is unsupported and the 12–13 dB thresholds must be restated as conditional on post-selection.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract claims the architecture \"can produce photonic qubits above the fault tolerance threshold with probabilities above 90%\" at 12–13 dB cluster squeezing. For that claim to hold, the quoted probabilities must be unconditional per-attempt probabilities, and the GKP squeezing distributions fed into the threshold simulation must be representative of real PhANTM outputs. Appendix 11.3 states that in the PhANTM Monte Carlo \"only states with a fidelity greater than 95% are retained\" before forming the cat-state inputs, and that \"we post-select homodyne measurements at 0.\" The discarded fraction is not reported, and the probability of the zero-outcome conditioning event is not reported; interpreted literally, exact-zero homodyne outcomes form a measure-zero set for a continuous measurement. The same filtered samples feed adaptive breeding (Sec. 5) and the end-to-end threshold pipeline (Sec. 8.2), so the 0.91 success rate in Table 1 and the 12.1/12.7/13.0 dB thresholds are conditional on these post-selections. If even 20–30% of PhANTM outputs fail the fidelity filter, or if nonzero homodyne outcomes require displacement corrections that degrade cat quality, the unconditional yield and the effective thresholds shift. The central architecture may be salvageable—the 10.2 dB balanced-GKP threshold reproduces earlier work and the structural design is coherent—but the headline probability and threshold are not yet established as stated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes an end-to-end continuous-variable photonic fault-tolerant architecture built from passive components: PhANTM generates squeezed cat states on dual-rail quantum wires; an adaptive breeding protocol converts them to GKP sensor states; a CV-QEC protocol acting on cat states produces magic states; and a macronode RHG lattice with GKP qubits is decoded with minimum-weight perfect matching. The authors report cluster-squeezing thresholds of 12.1, 12.7, and 13 dB for 20, 15, and 10 PhANTM steps, GKP generation probabilities above 90% (Table 1), and a magic-state success rate of 4.8% at 13 dB. They also benchmark a balanced-GKP squeezing threshold of 10.2 dB, consistent with prior macronode work [27].","tokens_in":31705,"tokens_out":6922,"duration_ms":61310,"significance":"If the headline yields and thresholds hold, the architecture would be significant: it removes active switches from the qubit-generation path, requires only low photon-number resolution (up to about 10 photons), and operates at squeezing levels below the 15 dB demonstrated in free space while exceeding the 8.3 dB demonstrated on-chip. The end-to-end simulation structure, with realistic GKP squeezing distributions fed into QEC decoding, is a useful step beyond idealized fixed-squeezing analyses. The paper's external anchor (10.2 dB balanced threshold) and its loss analysis lend credibility. The main caveat is that the headline probabilities and thresholds are computed conditioned on unquantified post-selections in the PhANTM pipeline (Appendix 11.3), so the unconditional claims are not yet established.","major_comments":[{"comment":"The end-to-end yields and thresholds rest on two post-selections in the PhANTM Monte Carlo: only output states with fitted fidelity greater than 95% are retained, and every homodyne outcome is post-selected to zero. The discarded fraction and the probability of the zero-outcome conditioning event are not reported. Since these filtered cat states feed adaptive breeding (Sec. 5) and the threshold pipeline (Sec. 8.2), the 0.91 success rate in Table 1 and the 12.1/12.7/13 dB thresholds are conditional success probabilities, not unconditional end-to-end yields. Exact-zero homodyne outcomes form a measure-zero set for a continuous variable; the authors defer nonzero-outcome effects to ref. [55], but the abstract's claim of qubit production 'with probabilities above 90%' requires the unconditional per-attempt probability. Please report the discarded fraction and the zero-outcome probability, or repeat the simulation with the full homodyne distribution and displacement corrections.","section":"Appendix 11.3 / Sec. 8.2 / Table 1"},{"comment":"The magic-state success rate of 0.048 and the factor-of-10 improvement over vacuum input inherit the same conditioning: the deterministic CV-QEC curves in Fig. 8 are computed with post-selection on zero quadrature outcomes, and the Monte Carlo uses PhANTM cats after the greater-than-95% fidelity filter. The text states that the zero-outcome post-selection 'slightly' underestimates output squeezing [54], but the size of the effect on the success rate is not quantified for the cat-input protocol. Please report the unconditional success probability, or at least the probabilities of the conditioning events, so the 4.8% claim can be compared fairly with other magic-state protocols.","section":"Sec. 6 / Fig. 10"},{"comment":"The QEC simulation samples noise from only the diagonal elements of Σout after symplectic propagation, ignoring correlations between modes introduced by the macronode beamsplitter network and the dictionary corrections. The manuscript does not quantify the impact of this approximation on the threshold. Because this approximation is used in the end-to-end threshold simulations, please either justify that the off-diagonal terms are negligible for the threshold, or sample from the full covariance matrix.","section":"Appendix 11.8 / Sec. 8.2"}],"minor_comments":[{"comment":"The text states that 1000 trials are run for the PhANTM Monte Carlo, while the Fig. 14 caption says 500 iterations for each dataset; please reconcile the sample sizes.","section":"Appendix 11.3 / Fig. 14"},{"comment":"The architecture is described as 'fully-passive' and 'switchless', but Sec. 4 requires active displacements between repeated PhANTM steps. Please clarify that 'switchless' refers to the absence of optical switches on the quantum modes, not the absence of all active feedforward elements.","section":"Sec. 3 / Sec. 4"},{"comment":"The model assumes that GKP effective squeezing values are Gaussian-distributed across modes with standard deviation σd. Since the yield-above-threshold calculation is sensitive to the tails of this distribution, please include a goodness-of-fit comparison or overlay the simulated histograms from adaptive breeding.","section":"Sec. 8.1 / Fig. 12"},{"comment":"In ref. [55], the author list contains 'Pfister Pfister'; this should be corrected to 'Olivier Pfister'.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The architecture is coherent and the external threshold anchor to prior macronode work is reassuring, but the conditional-yield issue is central to the abstract's main quantitative claim. I view this as fixable with additional reporting or simulation rather than a fundamental flaw. I also note that the manuscript header indicates an acceptance in Quantum; my review is based solely on the scientific content as submitted."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read. The paper has two genuinely new pieces: the adaptive breeding algorithm (Algorithm 1) that accepts variable-amplitude, variable-parity cats from PhANTM and substitutes momentum-squeezed states when cats are out of range, and the magic state protocol that runs CV-QEC on a squeezed cat input rather than vacuum. Both are real, concrete ideas. The chained simulation from PhANTM cat statistics through breeding to RHG lattice decoding is also a step beyond prior work; the fact that the balanced-quadrature threshold comes out at 10.2 dB, matching earlier macronode calculations, is a good external anchor.\n\nThe soft spot is exactly where the reader put it. Appendix 11.3 says only PhANTM outputs with fidelity greater than 95% are retained, and all homodyne outcomes are post-selected to zero. Those two filters sit upstream of everything: the GKP squeezing distributions feeding the threshold simulations, the 0.91 yield in Table 1, and the \"above 90%\" in the abstract. The discarded fraction is never reported, and exact-zero homodyne outcomes are measure zero for a continuous measurement. So the headline numbers are conditional, not end-to-end, until the authors quantify how much probability mass those filters remove and show nonzero outcomes don't degrade the cats.\n\nOne smaller overstatement: the architecture needs active feedforward displacements between PhANTM rounds, so \"fully passive\" and \"switchless\" should be read as \"no optical switches on the quantum modes,\" which is still useful but weaker than the abstract's phrasing. And the loss-inclusive end-to-end simulation is explicitly deferred, with only partial per-stage loss numbers in App. 11.9.\n\nNone of this sinks the architecture. The structure is coherent, the pieces are standard CV building blocks assembled sensibly, and the central claims are falsifiable by better simulation statistics. If the authors report the filter discard rate and unconditional yields, the 12–13 dB threshold claim becomes meaningful. As it stands, the paper deserves a serious referee but needs major revision before the headline numbers can be accepted.\n\nI'd cite the adaptive breeding protocol if I were working on CV GKP state preparation, and I'd bring it to reading group to discuss the post-selection issue.","headline":"Real architecture ideas, but the 90% yield and 12–13 dB thresholds are conditional on an unquantified fidelity filter and zero-outcome post-selection; worth refereeing, not worth accepting as headline numbers yet.","tokens_in":32393,"tokens_out":1972,"would_cite":true,"duration_ms":18923,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper's central claim is that a fully passive, switchless photonic architecture can run fault-tolerant continuous-variable quantum computing with only 12 to 13 dB of Gaussian cluster squeezing.","keywords":["continuous-variable quantum computing","GKP qubits","photonic cluster states","measurement-based quantum computation","fault tolerance","magic states","integrated photonics","quantum error correction"],"falsifier":"Re-run the published Monte Carlo chain from PhANTM through adaptive breeding without the 95%-fidelity filter and without zero-outcome post-selection, then count the fraction of GKP states that land in the correctable region at 13 dB; a true success probability well below 90% or a threshold above 13.5 dB would refute the central claim. A complementary experimental check is to demonstrate 13 dB squeezing from an integrated on-chip source, which has not yet been reported.","tokens_in":31091,"feed_emoji":"💡","tokens_out":10069,"duration_ms":90490,"temperature":0.7,"pith_summary":"The paper argues that fault-tolerant quantum computing can be done on a fully passive photonic chip, with no optical switches and no quantum memories. Its central claim is that a continuous-variable architecture built from time-frequency multiplexed cluster states produces GKP qubits above the fault-tolerance threshold with probability above 90%, at Gaussian cluster squeezing between 12 and 13 dB. It also claims a magic-state generation protocol that, with 13 dB squeezing, yields distillable magic states at 4.8% success probability, about an order of magnitude higher than the vacuum-based alternative. If these numbers hold, the main obstacle shifts from integrating low-loss switches and high-photon-number detectors to pushing on-chip squeezing from the current ~8 dB toward 13 dB.","feed_headline":"Switchless photonic chip design targets fault tolerance at 12-13 dB of squeezing","feed_subtitle":"GKP qubits pass the fault-tolerance threshold over 90% of the time, with magic states at 10x previous yields.","key_machinery":"The central mechanisms are PhANTM (repeated photon subtraction and teleportation along a dual-rail cluster state, which builds a large squeezed cat state without switches), adaptive breeding (an algorithm that squeezes each probabilistic cat to the required amplitudes, replaces unsuitable cats by momentum-squeezed states, and breeds pairs through homodyne measurements to form GKP sensor states), and the teleportation-based squeezing gate whose homodyne angles set the effective squeezing. The magic-state protocol replaces the vacuum mode in GKP error correction by an optimized cat state, which is what lets weaker GKP states remain distillable. At the logical level, GKP Bell pairs are assembled into a macronode RHG lattice with static linear optics; a dictionary protocol converts physical homodyne outcomes into effective canonical-lattice measurements, and decoded syndromes are processed by a minimum-weight perfect-matching decoder. The target cat amplitude for breeding is set by the spacing formula $\\alpha_b = \\xi 2^{(M-3)/2}$, which the adaptive protocol adjusts for each PhANTM output.","core_discovery":"The central claim is that a switchless, all-passive continuous-variable architecture can cross the fault-tolerance threshold with 12 to 13 dB of Gaussian cluster squeezing. The authors report fault-tolerance thresholds, expressed as cluster squeezing, of 12.1, 12.7, and 13.0 dB for 20, 15, and 10 rounds of their photon-counting-assisted node-teleportation (PhANTM) method; at 13 dB and 20 PhANTM steps, 91% of generated sensor states fall in the correctable region. They further report that replacing the vacuum mode in GKP-based error correction with an optimized cat state makes magic states distillable with GKP resources at roughly 13 dB effective squeezing, with a Monte Carlo success rate of 4.8% rather than 0.3%. The logical layer is constructed as a macronode RHG lattice assembled from GKP Bell pairs with static linear optics, and the logical error rates come from full end-to-end simulations that feed distributions of GKP squeezing from the generation pipeline into a surface-code decoder.","pith_inferences":["Inference: if the 95%-fidelity filter and zero-outcome post-selection are removed, the authors' own deferred analysis suggests nonzero homodyne outcomes mainly unbalance the cat amplitudes; whether feedforward displacement restores the quoted >90% yield is not tested in this paper.","Inference: the p/q squeezing imbalance produced by breeding points toward biased or asymmetric decoders that could lower the required cluster squeezing below 12 dB, an optimization the paper leaves open.","Inference: the paper's loss analysis implies the practical loss budget is dominated by the PhANTM cat-generation stage, since sub-1% propagation loss already erodes Wigner negativity there, whereas breeding and QEC tolerate a few percent loss before thresholds shift notably.","Inference: if integrated sources reach ~12 dB only at higher pump powers or with added loss, the next constraint may be feedforward timing, because active displacements must complete within one temporal-mode spacing; the paper does not quantify clock-rate limits."],"forward_implications":["If the threshold claim is right, fault-tolerant photonic computation no longer needs fast photonic switches or quantum memories; adaptivity reduces to setting local-oscillator phases and tracking corrections in software.","Low photon-number resolution (up to about 10 photons) suffices for cat generation, so high-bandwidth room-temperature detectors become viable instead of cryogenic detectors resolving 40 to 100 photons.","At 13 dB cluster squeezing with 20 PhANTM steps, roughly 91% of generated sensor states land above the fault-tolerance threshold, so qubit factories can operate at near-unity yield.","The 12 to 13 dB cluster-squeezing requirement sits below the 15 dB demonstrated in free space but above the ~8.3 dB shown on chip, making on-chip squeezing rather than detector resolution the pacing requirement.","Because the magic-state protocol works with GKP resource states at roughly 13 dB effective squeezing, universal non-Clifford gates become accessible without the ~20 dB GKP squeezing that vacuum-based magic-state preparation demands."],"supporting_citations":[{"why":"Supplies the PhANTM protocol that generates cat states on a cluster state via repeated photon subtraction and teleportation.","marker":"[55]"},{"why":"Establishes the cat-to-GKP breeding equations, including the target amplitude formula, that the adaptive-breeding algorithm is built around.","marker":"[54]"},{"why":"Shows that GKP error correction on vacuum can produce magic states using Gaussian operations; the paper's magic protocol extends this error-correcting a cat state instead.","marker":"[25]"},{"why":"Provides the macronode-based static-linear-optics construction of the RHG lattice and the dictionary protocol that this architecture adopts.","marker":"[27]"},{"why":"Gives the earlier switch-based photonic FTQC blueprint and the simulation methodology that this paper adapts to its generation pipeline.","marker":"[26]"},{"why":"Reports a fault-tolerant CV MBQC architecture with a 12.7 dB threshold against which this paper compares its end-to-end thresholds.","marker":"[28]"},{"why":"Defines the RHG cluster-state lattice, the foliated surface code used as the logical substrate.","marker":"[56]"},{"why":"Supplies the minimum-weight perfect-matching decoder used to turn sampled syndromes into logical error rates.","marker":"[87]"},{"why":"Derives the teleportation-based squeezing gate whose homodyne angles set the squeezing applied to PhANTM cats before breeding.","marker":"[58]"},{"why":"Demonstrates 15 dB squeezed light in free space, the experimental yardstick showing the 12 to 13 dB requirement is below demonstrated squeezing.","marker":"[90]"}],"fun_headline_variants":["Switchless photonic architecture hits fault tolerance at 12-13 dB","Passive photonic chip reaches quantum error-correction threshold with 12 dB squeezing","All-passive photonic quantum computing breaks fault-tolerance barrier at 13 dB","Switchless photonic design: 91% qubits above threshold at 13 dB squeezing","Magic states 10x easier: switchless photonic design at 13 dB squeezing"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The reported numbers depend on keeping only simulated cat states with fitted fidelity above 95% and post-selecting every homodyne outcome to zero; if the rejected tail is sizeable or nonzero outcomes degrade the cats, the yield and thresholds shift.","fun_headline_variants_meta":{"raw":{"variants":["Switchless photonic architecture hits fault tolerance at 12-13 dB","Passive photonic chip reaches quantum error-correction threshold with 12 dB squeezing","All-passive photonic quantum computing breaks fault-tolerance barrier at 13 dB","Switchless photonic design: 91% qubits above threshold at 13 dB squeezing","Magic states 10x easier: switchless photonic design at 13 dB squeezing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000809,"raw_usage":{"total_tokens":3565,"prompt_tokens":972,"completion_tokens":2593,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":588,"completion_tokens_details":{"reasoning_tokens":2486}},"tokens_in":588,"tokens_out":2593,"duration_ms":16239,"temperature":1.0,"reasoning_tokens":2486,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:51:46.630608+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the published Monte Carlo chain from PhANTM through adaptive breeding without the 95%-fidelity filter and without zero-outcome post-selection, then count the fraction of GKP states that land in the correctable region at 13 dB; a true success probability well below 90% or a threshold above 13.5 dB would refute the central claim. A complementary experimental check is to demonstrate 13 dB squeezing from an integrated on-chip source, which has not yet been reported.","supporting_citations":[{"cited_title":"Measurement-based generation and preservation of cat and grid states within a continuous-variable cluster state","cited_arxiv_id":null,"evidence_quote":"Supplies the PhANTM protocol that generates cat states on a cluster state via repeated photon subtraction and teleportation."},{"cited_title":"Generating grid states from schrödinger-cat states without postselection","cited_arxiv_id":null,"evidence_quote":"Establishes the cat-to-GKP breeding equations, including the target amplitude formula, that the adaptive-breeding algorithm is built around."},{"cited_title":"All-gaussian uni- versality and fault tolerance with the gottesman-kitaev-preskill code","cited_arxiv_id":null,"evidence_quote":"Shows that GKP error correction on vacuum can produce magic states using Gaussian operations; the paper's magic protocol extends this error-correcting a cat state instead."},{"cited_title":"Fault-tolerant quantum computation with static linear op- tics","cited_arxiv_id":null,"evidence_quote":"Provides the macronode-based static-linear-optics construction of the RHG lattice and the dictionary protocol that this architecture adopts."},{"cited_title":"Blueprint for a Scalable Photonic Fault-Tolerant Quantum Computer","cited_arxiv_id":null,"evidence_quote":"Gives the earlier switch-based photonic FTQC blueprint and the simulation methodology that this paper adapts to its generation pipeline."},{"cited_title":"Fault- tolerant continuous-variable measurement- based quantum computation architecture","cited_arxiv_id":null,"evidence_quote":"Reports a fault-tolerant CV MBQC architecture with a 12.7 dB threshold against which this paper compares its end-to-end thresholds."},{"cited_title":"A fault-tolerant one-way quantum com- puter","cited_arxiv_id":null,"evidence_quote":"Defines the RHG cluster-state lattice, the foliated surface code used as the logical substrate."},{"cited_title":"Sparse blossom: correcting a million errors per core second with minimum-weight match- ing","cited_arxiv_id":null,"evidence_quote":"Supplies the minimum-weight perfect-matching decoder used to turn sampled syndromes into logical error rates."},{"cited_title":"Noise analysis of single-mode gaussian op- erations using continuous-variable cluster states","cited_arxiv_id":null,"evidence_quote":"Derives the teleportation-based squeezing gate whose homodyne angles set the squeezing applied to PhANTM cats before breeding."},{"cited_title":"Detection of 15 dB squeezed states of light and their application for the absolute cali- bration of photoelectric quantum efficiency","cited_arxiv_id":null,"evidence_quote":"Demonstrates 15 dB squeezed light in free space, the experimental yardstick showing the 12 to 13 dB requirement is below demonstrated squeezing."}],"review_version":1}