{"id":"41298ef3-84bb-453c-bcc8-7e463c5af75a","arxiv_id":"2411.15444","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Two silicon photonic chips connected by optical fiber teleported a remote CNOT gate with 94 to 95 percent process fidelity.","lead":"Two silicon photonic chips, linked by an optical fiber, teleported a quantum controlled-NOT gate between remote qubits, with state fidelities near 95 percent. The result is a step toward modular quantum networks made from integrated photonic circuits.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The implemented branch-selection measurement destroys the output control qubit: since qubits 1 and 2 are encoded in the same photon, detecting qubit 2 to herald the |0+> branch absorbs the photon that should carry qubit 1, so no accessible teleported CNOT gate is actually produced.","rationale":"The reader's weakest assumption focuses on the unmeasured teleportation branches and absent feedforward, which is a valid incompleteness in verifying the full gate-teleportation identity. My concern is different and more fundamental: even the single branch that is measured does not produce an output qubit that survives the branch-selection measurement, because qubits 1 and 2 share one photon and the measurement of qubit 2 is destructive. This makes the abstract's claim that 'the CNOT gate is teleported between two remote quantum nodes' an overstatement unless 'teleported' is explicitly qualified as a postselected, destructive verification of a conditional correlation. The experimental work is still valuable and likely correct as a demonstration of the postselected gate-teleportation identity on one branch, so a conditional acceptance with required reframing remains appropriate. I therefore keep the reader's CONDITIONAL verdict, but for a reason that is not fully captured by the missing-branches critique. The proposed simulation test would settle whether the output qubit is truly unavailable, and if so, the paper must be revised to state that no usable remote CNOT gate is delivered, only a conditional joint probability distribution reconstructed from the same detections that select the branch.","tokens_in":8794,"tokens_out":19507,"duration_ms":187729,"concrete_test":"Simulate the exact two-photon, four-path circuit in Fig. 1(b) with a linear-optical network simulator using the reported scattering parameters. Compute the Choi matrix of the map from the input state of qubits 1 and 4 to the surviving optical modes after postselecting on the |0+>23 branch. If the Choi matrix has zero support on the qubit-1 subsystem because that photon is absorbed by the branch-selecting detector, then the experiment implements only a destructive conditional measurement, not a teleported gate on accessible remote qubits. A complementary check: trace the physical path of the photon carrying qubit 1 in a single successful event and confirm whether any output mode leaves the chip after the qubit-2 detection.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that a CNOT gate is teleported onto remote qubits 1 and 4. Equation (1) presupposes that qubits 1 and 2 are distinct physical systems: the protocol measures qubit 2, then applies the correction R1 to qubit 1, and the output state on qubits 1 and 4 remains available. The implementation instead encodes qubits 1 and 2 in the same photon: 'the path dimension of each photon ... is expanded to four' (Sec. II), and 'the measurement of qubit 2 ... is easily done by detecting output ports of the chip with a single-photon detector.' That detection absorbs the photon carrying qubit 1. The same section then states that 'before each single-photon detection, MZI and PSs are positioned ... to analyze the final state |Φ>14,' confirming that the branch-selecting detection and the output-state analysis are the same destructive event. Therefore, even in the only tested branch (|0+>23, R1=R4=I), no surviving output qubit 1 exists after the heralding measurement. The reconstructed 16×16 process matrix characterizes a conditional joint-detection correlation, not a channel that produces an accessible remote qubit for further distributed quantum information processing. This is a structural gap in the claimed gate teleportation, independent of the unmeasured branches and missing feedforward: implementing Eq. (1) physically would require a separate photon for qubit 1, or a non-destructive readout of qubit 2, neither of which is present.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experiment using silicon photonic integrated circuits to implement a claimed chip-to-chip CNOT gate teleportation. The authors generate path-entangled photon pairs, encode four qubits (two per photon, one photon at each node), and use the shared entangled pair between two chips to implement the protocol of Eq. (1). They characterize the operation by quantum state tomography of the output states and by quantum process tomography, reporting average entangled-state fidelities of 95.69% (5 m fiber) and 94.07% (1 km fiber), and process fidelities of 94.81% and 93.04%, respectively. The central claim is that the CNOT gate is teleported from local qubits 1&2 to remote qubits 1&4 across the fiber link.","tokens_in":9072,"tokens_out":4915,"duration_ms":42964,"significance":"If the result held as stated, it would be a significant step toward distributed quantum computation with integrated photonic nodes. The engineering achievements—on-chip generation of path-entangled photon pairs, chip-to-chip quantum interconnects with path-polarization conversion, and high-visibility interference across a 1 km fiber—are substantial, and the reported fidelities are high. The paper provides enough experimental detail to be reproducible in principle. However, the central scientific claim of CNOT gate teleportation is not supported by the implemented encoding, as detailed in the major comments. The strengths lie in the interconnect and entanglement distribution, not in the demonstration of a teleported gate that produces an accessible remote output.","major_comments":[{"comment":"The protocol identity in Eq. (1) requires qubits 1 and 2 to be distinct physical systems, because the measurement of qubit 2 is supposed to leave qubit 1 available as the output after the appropriate rotation R1. In the experiment, qubits 1 and 2 are encoded in the same photon: the text states, 'the path dimension of each photon in the entangled photon-pair source is expanded to four.' The measurement of qubit 2 is performed 'by detecting output ports of the chip with a single-photon detector,' which absorbs the photon that also carries qubit 1. The subsequent sentence, 'Before each single-photon detection, MZI and PSs are positioned in the designated \"State Measurement\" area to analyze the final state |Φ>14,' confirms that the branch-selecting detection and the output-state analysis are the same destructive event. Therefore, even in the only tested branch (|0+>23, R1=R4=I), no surviving output qubit 1 exists after the heralding measurement. The reconstructed 16×16 process matrix characterizes a conditional joint-detection correlation, not a channel that produces an accessible remote qubit for further distributed quantum information processing. This is a structural gap in the claimed gate teleportation, independent of the unmeasured branches and missing feedforward.","section":"Section II, sentence about collecting only the |0+>23 branch"},{"comment":"The paper states, 'For simplicity, we just collect photons in the term |0+>23 and corresponding single-qubit rotations on qubits 1&4 are R1 = R4 = I.' This means only one of the four branches of Eq. (1) is experimentally realized, and the single-qubit corrections R1 and R4 for the other branches are never implemented. The claim that a CNOT gate has been 'teleported' therefore relies on the unverified assumption that Eq. (1) holds for all four branches and that the corrections would work with comparable fidelity. As presented, the experiment demonstrates a postselected, single-branch conditional correlation, not the full teleportation protocol described in the introduction and in Fig. 1(a).","section":"Section II, paragraph on branch selection"},{"comment":"The process fidelity F = Tr(χ_exp χ_ideal) is computed for a process that includes the destructive measurement of qubit 2. Because qubit 1 is not independently accessible after the heralding event, the reconstructed process matrix does not represent a quantum channel that could be concatenated with subsequent operations in a distributed quantum computation. The claim that the result is 'sufficient for further chip-to-chip quantum information processing' is therefore not supported by the data; the process matrix is more appropriately interpreted as a characterization of a conditional two-photon joint measurement outcome.","section":"Section III, quantum process tomography discussion"}],"minor_comments":[{"comment":"The phrase 'Equip with 5 m (1 km)-long interconnecting fiber' should read 'Equipped with 5 m (1 km)-long interconnecting fiber.'","section":"Abstract"},{"comment":"The word 'teleportated' should be 'teleported.'","section":"Section III, first sentence"},{"comment":"The identity in Eq. (1) would benefit from an explicit enumeration of the four branches and the corresponding rotations R1 and R4; the current compact form obscures the fact that only one branch is measured.","section":"Equation (1) and surrounding text"},{"comment":"The 'isolation degree' (stated to exceed 200) is not defined precisely; please specify how it is computed from the measurement outcomes and what statistical uncertainty it carries.","section":"Section II, interconnect characterization"}],"recommendation":"reject","confidential_remarks":"The paper describes a well-executed integrated-photonics experiment with high reported fidelities for chip-to-chip entanglement distribution and for a conditional two-photon correlation that resembles a remote CNOT operation. However, the central claim of CNOT gate teleportation is not supported because the encoding of qubits 1 and 2 in the same photon makes the output qubit inaccessible after the heralding detection. This is a load-bearing structural issue that cannot be fixed by rewriting the claims or adding feedforward; it requires a different physical encoding with separate photons for qubits 1 and 2, or a nondestructive measurement scheme. The experiment is better described as a demonstration of postselected remote CNOT correlations in an integrated photonic network. The journal should weigh whether such a revised claim, if made by the authors, would be of sufficient interest. As submitted, the overstatement of the central result is grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a nice demonstration of chip-to-chip entangled-pair distribution and a postselected two-qubit gate operation, but the headline claim of 'CNOT gate teleportation' is not supported. The stress-test note is right: even in the one branch they measure, the output control qubit is destroyed by the heralding detection.\n\nWhat's new and good: they build a two-node silicon photonic network with an on-chip SFWM source on one chip, a 5 m or 1 km fiber link, and PBRC-based path-polarization conversion. The measured Bell-state fidelity for the distributed pair is 95.76% ± 0.83%, and the postselected gate statistics give average state fidelity around 95.7% and process fidelity around 94.8%. The engineering is credible: the isolation of the interconnect exceeds 200:1, and the methods are described in enough detail to be reproducible.\n\nThe soft spot is structural, not just a matter of missing feedforward. Equation (1) assumes qubits 1 and 2 are distinct physical systems. In the implementation, both are path-encoded on the same photon. Detecting qubit 2 (to herald the |0+> branch) absorbs that photon, and qubit 1 goes with it. The paper says the branch-selecting detection and the final-state analysis happen in the same detection event. So there is no surviving output control qubit. The experiment measures joint detection statistics that match the ideal gate for the postselected branch, but it does not produce an accessible remote CNOT gate for further processing. The authors do disclose the branch limitation but not this consequence.\n\nThis is a significant overstatement of the result, but not a fraudulent one. The hardware and the correlations are real. The right framing would be 'postselected gate operation across two chips' with the explicit caveat that the control qubit does not survive. To support the teleportation claim they would need separate photons for qubits 1 and 2, or some non-destructive measurement. Also, no raw data or supplementary material is in the arXiv version.\n\nWho should read it: people working on integrated photonic networks will find the interconnect and source engineering useful. The paper deserves a serious referee, because the flaw is subtle and the engineering is high quality. But I would not cite it for 'teleportation' until the framing is fixed.","headline":"Impressive silicon photonics, but the 'teleported CNOT gate' is actually a postselected correlation that destroys the output control qubit.","tokens_in":9670,"tokens_out":6580,"would_cite":false,"duration_ms":54722,"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":"A two-qubit controlled-NOT gate can be teleported between two remote silicon photonic chips, with measured fidelities above 94 percent.","keywords":["quantum gate teleportation","controlled-NOT gate","silicon photonics","quantum networks","photonic integrated circuits","path entanglement","chip-to-chip interconnect","quantum process tomography"],"falsifier":"Implement the three other measurement branches of Eq. (1) and apply the required single-qubit corrections R1 and R4, then perform quantum process tomography over all branches together. If the branch-averaged process fidelity falls significantly below the single-branch values reported here (94.81% and 93.04%), the full teleportation identity is not supported.","tokens_in":8574,"feed_emoji":"🔗","tokens_out":4862,"duration_ms":41009,"temperature":0.7,"pith_summary":"Quantum networks need operations that act jointly on qubits in different nodes, not just distribution of entanglement. This paper claims to demonstrate the key missing primitive: teleporting a two-qubit controlled-NOT (CNOT) gate between two separate silicon photonic chips connected by an optical fiber. A path-entangled photon pair is generated on one chip, one photon is sent to the other chip, and local linear-optical manipulations make the remote qubits behave as if a CNOT had acted between them. The authors verify the operation by entangling remote qubits and by quantum process tomography, reporting average fidelities above 94% for both 5 m and 1 km fiber links. If correct, the result provides a practical building block for distributed quantum computation with integrated photonics.","feed_headline":"CNOT gate teleported between two silicon chips at 94% fidelity","feed_subtitle":"A nonlocal two-qubit gate now works across separate chips linked by optical fiber, enabling modular quantum networks.","key_machinery":"The central object is Eq. (1), the gate-teleportation identity that rewrites the action of a local CNOT C12 on qubits 1 and 2 (when qubits 2 and 3 are entangled) into the action of a remote CNOT C14 on qubits 1 and 4, followed by single-qubit rotations R1 and R4 that depend on measurement outcomes on qubits 2 and 3. The experiment realizes this identity with a path-entangled photon-pair source based on spontaneous four-wave mixing in silicon waveguides, Mach-Zehnder interferometers for state preparation, a polarization beam rotator and combiner (PBRC) for chip-to-chip path-polarization interconversion, and a network of beam splitters for the {|+>, |->} basis measurement on qubit 3. The identity is what converts the local entangling operation C12, together with the shared entanglement, into a nonlocal entangling gate on remote qubits 1 and 4.","core_discovery":"On its own terms, the paper demonstrates chip-to-chip CNOT gate teleportation using a silicon-photonics platform. An entangled photon pair is generated on one chip; one idler photon is sent to the second chip through single-mode fiber. Local CNOT gates and a measurement on one branch of the teleportation identity make the remote qubits 1 and 4 become entangled exactly as if a CNOT had acted on them. The teleported gate is characterized with quantum state and process tomography: the average entangled-state fidelity is 95.69% (5 m) and 94.07% (1 km), and the process fidelity is 94.81% and 93.04% for the same two distances. These numbers are close to the values reported for photonic CNOT gates implemented on a single chip, indicating that the chip-to-chip interconnect does not introduce a major fidelity penalty.","pith_inferences":["The full gate-teleportation identity includes four measurement branches with associated rotations R1 and R4; the paper only implements the |0+>23 branch where both rotations are the identity, so the demonstration is postselected rather than a full feedforward teleportation.","A natural extension would be to implement the other three branches with active feedforward (fast polarization or path switching) and measure the branch-averaged process fidelity, which would be a stronger test of the teleportation claim.","The same silicon-photonics architecture could be reprogrammed (via the Mach-Zehnder interferometers) to teleport other two-qubit gates, such as controlled-phase gates, by changing the local rotations and measurement bases.","If the identity branch is representative of all branches, this approach could be combined with heralded entanglement swapping to build a repeater-like network where nonlocal gates are performed between distant nodes without direct qubit routing."],"forward_implications":["A remote CNOT gate between two silicon chips provides a primitive for distributed quantum computation, where processing is spread across multiple nodes rather than confined to one physical chip.","The demonstrated operation works over a 1 km standard single-mode fiber, so chip-to-chip entangling gates are compatible with practical distances for quantum networks.","Silicon photonics is CMOS-compatible, so the same fabrication technology could scale to more qubits per node and to networks with many interconnected nodes.","The process fidelity of the teleported gate (about 94% at 1 km) is comparable to that of single-chip CNOT gates, suggesting that remote gate teleportation is not a limiting step for near-term integrated quantum processors."],"supporting_citations":[{"why":"Establishes the theoretical framework for teleportation-based universal quantum computation using CNOT gates and single-qubit operations.","marker":"[20]"},{"why":"Supplies the minimal-resource gate-teleportation identity (Eq. 1) that the experiment realizes.","marker":"[26]"},{"why":"Reports the first experimental teleportation of a quantum controlled-NOT gate in bulk optics, the prior art this work extends to integrated chips.","marker":"[21]"},{"why":"Demonstrates a teleportation-based realization of an optical two-qubit entangling gate, providing a benchmark for gate teleportation fidelity.","marker":"[22]"},{"why":"Describes the chip-to-chip quantum photonic interconnect via path-polarization interconversion, the method used here to transfer one photon between chips.","marker":"[11]"},{"why":"Reports chip-to-chip quantum teleportation in silicon, the direct precursor that transferred quantum states (not gates) between chips.","marker":"[18]"},{"why":"Provides a single-chip silicon CNOT gate result used as a fidelity comparison for the teleported gate.","marker":"[34]"},{"why":"Supplies the quantum state tomography method used to reconstruct output density matrices and fidelities.","marker":"[30]"}],"fun_headline_variants":["CNOT teleported between chips at 95% fidelity","Quantum gate teleportation across silicon chips at 95% fidelity","Chip-to-chip CNOT gate teleportation at 95% fidelity","Teleporting a quantum gate between separate chip nodes","Gate teleportation links remote silicon chips at 95% fidelity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The experiment verifies only one of the four measurement branches of the gate-teleportation identity (the |0+>23 term with R1 = R4 = I), so the full claim that a CNOT gate has been teleported assumes that the other branches and their feedforward rotations would work with comparable fidelity.","fun_headline_variants_meta":{"raw":{"variants":["CNOT teleported between chips at 95% fidelity","Quantum gate teleportation across silicon chips at 95% fidelity","Chip-to-chip CNOT gate teleportation at 95% fidelity","Teleporting a quantum gate between separate chip nodes","Gate teleportation links remote silicon chips at 95% fidelity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001391,"raw_usage":{"total_tokens":5867,"prompt_tokens":923,"completion_tokens":4944,"prompt_tokens_details":{"cached_tokens":896},"prompt_cache_hit_tokens":896,"prompt_cache_miss_tokens":27,"completion_tokens_details":{"reasoning_tokens":4856}},"tokens_in":27,"tokens_out":4944,"duration_ms":46750,"temperature":1.0,"reasoning_tokens":4856,"cache_read_input_tokens":896,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:17:38.208517+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Implement the three other measurement branches of Eq. (1) and apply the required single-qubit corrections R1 and R4, then perform quantum process tomography over all branches together. If the branch-averaged process fidelity falls significantly below the single-branch values reported here (94.81% and 93.04%), the full teleportation identity is not supported.","supporting_citations":[{"cited_title":"Zheng, C","cited_arxiv_id":null,"evidence_quote":"Establishes the theoretical framework for teleportation-based universal quantum computation using CNOT gates and single-qubit operations."},{"cited_title":"Rudolph, Why I am optimistic about the silicon-photonic route to quantum computing","cited_arxiv_id":null,"evidence_quote":"Supplies the minimal-resource gate-teleportation identity (Eq. 1) that the experiment realizes."},{"cited_title":"Gottesman, and I","cited_arxiv_id":null,"evidence_quote":"Reports the first experimental teleportation of a quantum controlled-NOT gate in bulk optics, the prior art this work extends to integrated chips."},{"cited_title":"Huang, X.-F","cited_arxiv_id":null,"evidence_quote":"Demonstrates a teleportation-based realization of an optical two-qubit entangling gate, providing a benchmark for gate teleportation fidelity."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the chip-to-chip quantum photonic interconnect via path-polarization interconversion, the method used here to transfer one photon between chips."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides a single-chip silicon CNOT gate result used as a fidelity comparison for the teleported gate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the quantum state tomography method used to reconstruct output density matrices and fidelities."}],"review_version":1}