{"id":"8f6390b5-1c4d-45bd-b263-5769a598a0a3","arxiv_id":"1908.05351","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A 12-photon experiment demonstrates a 2x2 parallel all-photonic quantum repeater with a 1.89x entanglement-rate enhancement over standard parallel entanglement swapping, and verifies the output is genuinely entangled.","lead":"This paper reports the first experiment that realizes an all-photonic quantum repeater, using 12 photons and a 2-by-2 parallel setup. It observes an 89% increase in entanglement generation rate over standard parallel entanglement swapping, a proof of principle for an approach that avoids quantum memories at intermediate repeater nodes.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Charlie-node GHZ loss does not cancel in the rate comparison; the 89% enhancement assumes lossless local GHZ preparation and is therefore a proof-of-principle, not a deployed rate.","rationale":"The single most load-bearing issue is not the fidelity measurement, the PCM characterization, or the internal consistency of the eight-photon postselection; those parts of the paper are credible. The load-bearing issue is the interpretation of the measured count ratio as an entanglement-generation-rate enhancement. The paper's own theoretical rate advantage over conventional parallel swapping is derived under the assumption that Charlie's local GHZ state is lossless and can be prepared just before the distant photons arrive. The experiment does not implement this delayed preparation: all 12 photons come from simultaneously pumped SPDC sources, and the eight-photon coincidence condition includes Charlie's GHZ photons. A loss model with a realistic Charlie-node efficiency shows that the all-photonic scheme is more sensitive to local photon loss than the conventional parallel baseline, because the GHZ switch requires all four local photons while the conventional scheme can still succeed if only one channel's local pair survives. With the paper's stated 38% average system efficiency, the predicted ratio drops well below unity, so the observed 1.89 does not establish a practical rate gain. This does not invalidate the proof-of-principle demonstration of passive switching or the measured two-qubit entanglement, so the reader's CONDITIONAL verdict remains appropriate: the paper should be read as a conditional demonstration whose headline rate advantage relies on an unverified losslessness assumption for the local GHZ state.","tokens_in":12951,"tokens_out":19157,"duration_ms":204880,"concrete_test":"Recompute the rate ratio for the M=2, N=1 protocols from the count of required photon detections with a Charlie-node per-photon efficiency η_C and distant-photon efficiency η_L, including the probability that the four-photon GHZ is fully present. Then evaluate r(η_C)=2η_C² at η_C=0.38 (the paper's stated average system efficiency) and compare with the measured r=1.89. If the predicted value remains far below 1.89 unless η_C=1, the enhancement is contingent on the un-implemented delayed-lossless-GHZ assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central experimental claim — the measured r=1.89±0.10 rate enhancement (Fig. 4a) and its theoretical curve r=2−4p+2p² — treats Charlie's four-photon GHZ as present with unit probability at the moment the distant photons arrive. The main text explicitly relies on 'delayed preparation' so that 'the GHZ state is lossless compared with the photons sent from distant nodes.' In the actual apparatus all 12 photons are generated in the same pulse and the SPDC sources have 38% average system efficiency; the 8-photon coincidence postselection includes Charlie's GHZ photons in the all-photonic count. The loss terms do not cancel in a real comparison: for M=2, N=1, the conventional parallel scheme can succeed if either channel's two Charlie-side qubits survive (probability ~2η_C²), while the all-photonic GHZ switch requires all four GHZ photons to survive (probability η_C^4). Including long-distance link efficiencies η_L², the ratio is r≈2η_C², not 2. With η_C≈0.38 this gives r≈0.29, far below the measured 1.89. Thus the observed count ratio is a conditional, postselected rate advantage, not an unconditional entanglement-generation-rate gain for a practical repeater; the paper's own caveat about delayed GHZ preparation is the load-bearing assumption, and it is not demonstrated experimentally.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental implementation of a 2x2 parallel all-photonic quantum repeater using 12 photons generated by six SPDC sources. Charlie's repeater node uses a four-photon GHZ state and four passive-choice measurement (PCM) devices to emulate the switching function of a repeater graph state without quantum memory. The authors measure a rate ratio r = 1.89 +/- 0.10 at a down-conversion probability p = 0.0344 and r = 1.74 +/- 0.07 at p = 0.0483, and compare these with a claimed theoretical formula r = 2 - 4p + 2p^2. They also reconstruct the GHZ state, characterize the PCM devices via detector tomography, and measure final-state fidelities for four possible output pairings, finding an overall fidelity of 0.606 +/- 0.010 with individual values all above 0.5.","tokens_in":13206,"tokens_out":16667,"duration_ms":166669,"significance":"If the claims hold, this is a noteworthy proof-of-principle demonstration that an all-photonic repeater node can be built from linear optics and a GHZ state, and that it yields a measurable rate advantage over a conventional entanglement-swapping baseline under postselection. The experiment is technically impressive: it involves 12-photon interference, full tomographic characterization of a four-photon GHZ state and of the PCM devices, and rate measurements at two values of p that support the predicted p-dependence. The central rate formula is not fitted; its only input, the down-conversion probability p, is measured independently from two-photon rates. The main caveats are that the reported rate is a conditional, postselected eight-fold coincidence rate and that the practical rate advantage relies on the assumption of lossless, delayed GHZ preparation, which is not demonstrated experimentally.","major_comments":[{"comment":"The formula r = 2 - 4p + 2p^2 is asserted in the main text with a pointer to the Supplementary Information, but the Supplementary as provided does not contain a derivation of this formula. This formula is the quantitative basis for the central claim of an 89% rate enhancement, so the derivation should be given explicitly, including a clear definition of the success events counted in each protocol and the treatment of higher-order SPDC noise that produces the p-dependent corrections. Without this derivation, the reader cannot verify that the measured ratios r = 1.89 and r = 1.74 are correctly predicted by the model.","section":"Main text, section 'In our experiments, we define the ratio r' and Supplementary Information"},{"comment":"The claim of an '89% enhancement of entanglement-generation rate over the standard parallel entanglement swapping' is overstated as stated. The measured ratio is a postselected eight-fold coincidence count ratio, and the comparison baseline is a single conventional entanglement-swapping channel, not the full M = 2 parallel conventional scheme. Moreover, the rate advantage assumes that Charlie's four-photon GHZ state is prepared losslessly and is present exactly when the distant photons arrive; in the actual experiment all 12 photons are generated in the same pulse and the all-photonic count requires all four GHZ photons to be present and detected. With the stated 38% average system efficiency, an unconditional rate comparison would include an additional loss penalty for the four GHZ photons, so the practical regime of the advantage is not demonstrated. The manuscript should explicitly distinguish the conditional, postselected rate from an unconditional end-to-end rate and should state that the delayed-preparation/lossless-GHZ assumption is an assumption, not an experimentally demonstrated feature.","section":"Abstract and main text, rate comparison and Fig. 4a"}],"minor_comments":[{"comment":"The horizontal-axis labels 'TT ZZ', 'TR YY', 'RT XX', and 'RR' are not explained in the text or caption; the notation T and R should be defined.","section":"Fig. 4b-e"},{"comment":"The sentence 'The overall fidelity is 0.606 +/- 0.010, which clearly indicates that the final shared state is genuinely entangled' should be clarified: the value 0.606 is the average of four outcome-dependent fidelities, and the entanglement conclusion follows because each of the four individual fidelities (0.587, 0.598, 0.597, 0.628) exceeds the 0.5 threshold. As written, the sentence could be read as referring to a single mixed state whose average fidelity exceeds 0.5.","section":"Main text, paragraph on final-state fidelity"},{"comment":"There is a typo 'One the other hand' that should read 'On the other hand'.","section":"Supplementary Information, section 'Passive implementation of the selective BSM'"},{"comment":"The phrase 'photons 2, 3, 6 and 7 (5, 8, 9 and 12) are send to the node C1 (C2)' contains a subject-verb agreement error ('are send' should be 'are sent').","section":"Main text, experimental setup"},{"comment":"The caption contains the typo '2ed leaf' instead of '2nd leaf'.","section":"Supplementary Information, Fig. 5 caption"}],"recommendation":"major_revision","confidential_remarks":"The experimental work is substantial and likely of interest to the quantum-information community, but the manuscript's central quantitative claim needs careful re-framing. The missing derivation of r = 2 - 4p + 2p^2 in the Supplementary is a concrete gap that must be filled. I would also ask the authors to tighten the language so that 'rate enhancement' is consistently described as a conditional, postselected rate, and to explicitly acknowledge the lossless-delayed-GHZ assumption in the abstract and conclusion."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know two things. This is a genuinely new experiment, not a rehash: nobody had implemented an all-photonic repeater before. They use a four-photon GHZ as a simplified repeater graph state and a passive choice measurement to realize the selective BSM. The headline ratio r=1.89±0.10 at p=0.034 is measured directly from eightfold coincidences, with two data points that track the theory curve. Fidelities of the four possible output pairs are all above 0.5, so the final shared states are genuinely entangled. That is solid work, and the 12-photon manipulation is a real technical achievement.\n\nNow the soft spots, in proportion. The formula r=2-4p+2p^2 is just asserted in the main text; the derivation is deferred to the supplement. No raw data are deposited. And the \"without quantum memory\" title is a bit strong: the paper itself acknowledges that Alice and Bob still need end-node memories if they demand a quantum output. Those are minor or easily fixed.\n\nThe larger caveat is the lossless-GHZ assumption. The theoretical rate gain assumes Charlie can prepare his GHZ state locally just before the photons from Alice and Bob arrive, so his loss is negligible. In this experiment, all photons come from the same pulsed source, and the eightfold coincidences include the GHZ photons. If you include realistic Charlie-node loss (the average system efficiency is 38%), the unconditional rate gain largely disappears — the stress-test estimate gives r≈0.29, far below the measured 1.89. This does not invalidate the proof-of-principle; the authors state the delayed-preparation assumption, and postselection is standard in these demonstrations. But the abstract's \"89% enhancement\" should be read as a conditional, postselected rate, not a field-tested gain. A referee should ask for a sentence that makes that explicit in the abstract or conclusion.\n\nWho's this for? Anyone working on quantum repeaters, all-photonic schemes, or photonic graph states. It's a proof-of-concept that the RGS switching idea can be realized with linear optics and a small GHZ state. The experiment deserves serious peer review. I'd accept it with requests for clarity about the idealization, the derivation of r, and data availability. My view: conditionally accept, not desk reject.","headline":"First demonstration of an all-photonic repeater with a 12-photon interferometer; the measured rate gain is real but conditional on an idealization about Charlie's GHZ loss.","tokens_in":13821,"tokens_out":4546,"would_cite":true,"duration_ms":42867,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["03.67.Hk","03.67.Bg","42.50.Ex"],"model":"deepseek-v4-flash","headline":"A 12-photon interferometer demonstrates an all-photonic quantum repeater that beats parallel entanglement swapping by 89 percent.","keywords":["all-photonic quantum repeater","quantum memory","graph state","GHZ state","entanglement swapping","linear optics","passive choice measurement","12-photon interferometer"],"falsifier":"Repeat the 2x2 one-node configuration while recording how often Charlie successfully produces the GHZ state instead of postselecting on it; if the eightfold-coincidence ratio over parallel swapping falls to 1 or below once those preparation failures are counted, the enhancement reported here is an artefact of postselection.","tokens_in":12724,"feed_emoji":"🔗","tokens_out":7457,"duration_ms":67808,"temperature":0.7,"pith_summary":"The paper reports a proof-of-principle all-photonic quantum repeater that does away with quantum memory inside the repeater node. Using a four-photon GHZ state at the middle node and four passive-choice measurements in a 12-photon interferometer, it implements a 2x2 parallel repeater and measures an entanglement-generation rate 1.89 +/- 0.10 times that of conventional parallel entanglement swapping at a down-conversion probability p = 0.0344. The final states shared by Alice and Bob retain a combined fidelity of 0.606 +/- 0.010, which the paper takes as evidence of genuine entanglement. The point of the demonstration is that a suitably prepared photonic graph state can play the role a quantum memory plays in the standard repeater paradigm, by connecting successful channels and disconnecting failed ones without feed-forward.","feed_headline":"All-photonic repeater lifts entanglement rate by 89%","feed_subtitle":"No quantum memory needed: a GHZ state at the repeater node swaps EPR pairs across two parallel channels.","key_machinery":"The mechanism that carries the argument is the passive-choice measurement (PCM) used with a local GHZ state. The GHZ state is local-unitary equivalent to the complete graph state at the heart of the all-photonic proposal, so at Charlie it serves as a switch: if a Bell-state measurement between a GHZ photon and an incoming EPR photon succeeds, the entanglement is extended into the network; if the photon arrives alone, an X-basis projection removes that qubit without destroying the remaining entanglement. The PCM implements this switch passively: a circular polarising beam splitter routes two coincident photons to a Bell analyser and one photon to an X-basis projector, so no active feed-forward is required. The repeater node also delays preparation of the GHZ state until the distant photons are about to arrive, which is what lets the paper treat the local state as lossless relative to the transmitted photons.","core_discovery":"The central claim is that the all-photonic repeater idea can be realised with linear optics at the few-photon scale, and that even this small instance already shows the predicted rate advantage. At the repeater node Charlie, the paper replaces the large repeater graph state of the original proposal with a four-photon GHZ state, and replaces active feed-forward with a passive-choice measurement that performs a Bell-state measurement when two photons arrive together and an X-basis projection when only one arrives. Registering eight-photon coincidences, the experiment finds a rate ratio r = 2 - 4p + $2p^{2}$ as a function of down-conversion probability p, measured as r = 1.89 +/- 0.10 at p = 0.0344 and r = 1.74 +/- 0.07 at p = 0.0483; as p tends to 0 the ratio tends to 2. The reconstructed final states for photon pairs 1&11, 4&11, 1&10 and 4&10 give an overall fidelity of 0.606 +/- 0.010, which the paper cites as clear evidence that the output is genuinely entangled rather than a classical mixture.","pith_inferences":["The 89% advantage is a proof of principle: Charlie's GHZ photons are postselected and their preparation is treated as lossless, so the real-world gain will only be this large if a deterministic, delayed source of graph states is available.","If a deterministic single-photon source replaced the SPDC crystals, the same passive-choice design should reproduce the r = 2 - 4p + 2p^2 curve at much lower effective p, which would be a clean test of the scaling.","The passive-choice idea may transfer to larger repeater graph states, not only the four-photon GHZ case, allowing the same no-feed-forward switching to be tested in multi-node repeaters."],"forward_implications":["A repeater node can in principle be built without matter memories, removing coherence-time limits and long-distance heralding at intermediate nodes.","The measured rate ratio r = 2 - 4p + 2p^2 implies the advantage grows as the source's multi-pair emission probability p goes down, reaching a factor 2 for a perfect single-pair source in the 2x2 case.","In the full scheme with more channels and nodes, the rate scaling becomes M^{N+1} eta^{N+1} instead of M eta^{N+1}, so the benefit is exponential in the number of parallel channels.","If Alice and Bob need a quantum output state, memories remain at the end nodes, but the required memory time scales only linearly with distance rather than polynomially or subexponentially.","For tasks that only need shared classical information, such as quantum key distribution, the end-node memories can also be removed by delay-choice entanglement swapping."],"supporting_citations":[{"why":"Proposes the all-photonic repeater graph-state protocol and its M^{N+1} eta^{N+1} rate law, the theoretical target this experiment scales down to 2x2.","marker":"[23]"},{"why":"Defines entanglement swapping, the conventional mechanism whose parallel version serves as the comparison baseline.","marker":"[9, 10]"},{"why":"Introduces the memory-based repeater paradigm and the requirement of quantum memory that the all-photonic scheme removes at intermediate nodes.","marker":"[7, 8]"},{"why":"Supplies the memory-time scaling comparison (polynomial/subexponential versus linear) used to argue the all-photonic advantage at end nodes.","marker":"[6, 8]"},{"why":"Provides the quantum detector tomography method used to characterise the passive-choice measurement operators.","marker":"[32]"},{"why":"Provides the maximum-likelihood reconstruction used to certify the GHZ state and the final shared entangled states.","marker":"[31]"}],"fun_headline_variants":["No-memory quantum repeater demoed: 89% rate boost","All-photonic repeater without memory lifts rate 89%","12-photon graph-state repeater swaps without memory","89% faster entanglement: memory-free repeater works","Photonic GHZ state replaces memory in repeater test"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The premise that carries the experiment is that Charlie can prepare the local GHZ state deterministically and without loss just before the long-distance photons arrive; if that preparation is actually probabilistic or lossy, the measured rate ratio overstates the gain a real repeater would deliver.","fun_headline_variants_meta":{"raw":{"variants":["No-memory quantum repeater demoed: 89% rate boost","All-photonic repeater without memory lifts rate 89%","12-photon graph-state repeater swaps without memory","89% faster entanglement: memory-free repeater works","Photonic GHZ state replaces memory in repeater test"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000638,"raw_usage":{"total_tokens":2949,"prompt_tokens":965,"completion_tokens":1984,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":581,"completion_tokens_details":{"reasoning_tokens":1910}},"tokens_in":581,"tokens_out":1984,"duration_ms":16599,"temperature":1.0,"reasoning_tokens":1910,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:17:30.470885+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the 2x2 one-node configuration while recording how often Charlie successfully produces the GHZ state instead of postselecting on it; if the eightfold-coincidence ratio over parallel swapping falls to 1 or below once those preparation failures are counted, the enhancement reported here is an artefact of postselection.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proposes the all-photonic repeater graph-state protocol and its M^{N+1} eta^{N+1} rate law, the theoretical target this experiment scales down to 2x2."},{"cited_title":"Hasegawa, R","cited_arxiv_id":null,"evidence_quote":"Provides the quantum detector tomography method used to characterise the passive-choice measurement operators."},{"cited_title":"Ewert and P","cited_arxiv_id":null,"evidence_quote":"Provides the maximum-likelihood reconstruction used to certify the GHZ state and the final shared entangled states."}],"review_version":1}