{"id":"5d78dbe5-a3a5-4dd6-91ce-c8a51645f4d0","arxiv_id":"2505.11195","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A two-way teleportation strategy, where both source and destination qubits move toward an intermediate core, reduces end-to-end communication latency by 30-40% and circuit depth by up to 24% in simulations of multi-core quantum computers.","lead":"This paper proposes a decentralized framework for teleporting qubits between cores of a multi-core quantum computer, with a two-way variant where both the source and destination qubits move toward a middle core. The authors report roughly 40% lower communication latency on synthetic benchmarks, 30% lower on real benchmarks, and up to 24% lower circuit depth compared to hop-by-hop teleportation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Two-way teleportation relocates both operands, but the depth model counts only the initial approach; ignoring the need to relocate qubits for later gates likely overstates the 24% depth reduction.","rationale":"The reader correctly identified the unvalidated cost model in Section V.C as the weakest point. My concern is a more specific failure of that model: it does not track logical-qubit locations across the circuit. In the two-way protocol, both source and destination qubits are moved to an intermediate core, so the destination qubit is no longer in its home core. Any subsequent gate involving that qubit and a qubit that stayed behind requires additional teleportation, which the paper's per-request depth accounting omits. This is especially relevant for the real benchmarks, where qubits are reused (e.g., QFT and Cuccaro's adder), yet the 24% depth figure is reported as a circuit-level result. The conceptual idea of decentralized, simultaneous movement is plausible and may reduce latency in favorable mappings, so I would not reject the paper outright. However, the quantitative headline claims are not currently supported without a full circuit-level placement simulation. Keeping the CONDITIONAL verdict from the reader is appropriate, with the concrete location-tracking experiment as the condition that should be met before the quantitative claims are accepted.","tokens_in":8307,"tokens_out":4699,"duration_ms":47994,"concrete_test":"Re-run the experiments in Section V while maintaining a logical-qubit core map: after each inter-core gate, update the core of each operand; for every subsequent gate, compute the minimum teleport hop count needed to bring its two operands together (allowing either operand or both to move), and accumulate this into the circuit depth. Compare hop-by-hop and two-way teleportation on the same benchmarks (QFT, Cuccaro's adder, MCMTV, Quantum Volume, and the random depth-5 and depth-10 circuits). If the two-way depth reduction falls below the claimed 24% or the ordering reverses, the central claim is unsupported. Ideally, the modified SeQUeNCe simulator should be released so the location-tracking accounting can be checked directly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section V.C states that 'each time a qubit is moved closer to the destination core, the circuit depth increases by one.' This treats each inter-core two-qubit gate as an isolated event. However, both proposed protocols physically relocate logical qubits: hop-by-hop moves the source qubit to the destination core, and two-way teleportation moves both operands to an intermediate core (Section IV.B, Algorithm 1). Under the paper's own 'direct 1:1 mapping of logical qubits to physical qubits ... without considering any compiler optimization' (Section V.A), a logical qubit's core is not fixed. If a qubit teleported in an earlier gate later participates in another gate with a qubit that remained in its home core, at least one additional teleport chain is required. The paper never maintains a location map across the circuit; it appears to evaluate each communication request with source and destination cores fixed to their original locations. In two-way teleportation, the destination qubit is the one that leaves home, so the residual relocation cost is asymmetric and likely larger than in hop-by-hop, which at least leaves the destination qubit in place. Since the claimed 24% depth reduction and part of the 40% latency reduction are computed from per-request hop counts, ignoring this residual cost can systematically favor two-way teleportation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a decentralized teleportation framework for multi-core quantum computing, in which entanglement is generated locally at BSM nodes between neighboring cores rather than by a centralized EPR generator. Two protocols are compared: hop-by-hop teleportation, where the source qubit moves one hop at a time to the destination core using deterministic XY routing, and two-way teleportation, where both source and destination qubits move simultaneously toward an intermediate core where the two-qubit gate is executed. The authors simulate 4x4 mesh networks by modifying the SeQUeNCe simulator, using synthetic random circuits and four real benchmark circuits (QFT, Cuccaro adder, MCMTV, Quantum Volume). They report that two-way teleportation reduces end-to-end communication latency by 40% on synthetic benchmarks and 30% on real applications, and reduces circuit depth by 24%, compared to hop-by-hop teleportation. The paper also discusses a congestion limitation arising from deterministic XY routing and suggests adaptive routing as future work.","tokens_in":8616,"tokens_out":2759,"duration_ms":27949,"significance":"If the reported reductions hold, the two-way teleportation scheme would be a simple and interesting architectural improvement: moving both operands toward each other roughly halves the communication distance for inter-core two-qubit gates in small mesh networks, without requiring compiler-level optimization. The conceptual contribution is clear and the paper is honest about at least one limitation (middle-core congestion). The use of an established simulator (SeQUeNCe) and a direct comparison against a well-defined baseline are strengths. However, the quantitative claims are currently not adequately supported: the depth model in Section V.C is asserted rather than validated, the simulation methodology is under-specified, and the headline numbers in the abstract conflict with per-benchmark numbers in Section V.B. The central idea is plausible, but the evidence as presented is not yet sufficient to verify the claimed magnitudes.","major_comments":[{"comment":"The cost model 'each time a qubit is moved closer to the destination core, the circuit depth increases by one' is a load-bearing assumption, but it is neither derived nor validated. Under the paper's own setup of a direct 1:1 mapping of logical to physical qubits without compiler optimization (Section V.A), teleportation physically relocates logical qubits. In two-way teleportation the destination operand leaves its home core, so if that qubit is later needed for another two-qubit gate with a qubit that has remained at (or returned to) the original home core, additional teleportation hops are required. The paper appears to evaluate each inter-core communication request with source and destination cores fixed to their original locations and never maintains a location map across the circuit. Ignoring these residual relocation costs can systematically favor two-way teleportation and undermines both the 24% depth reduction and, to a lesser extent, the latency reduction. Please either track qubit locations throughout the circuit and re-evaluate the depth, or explicitly justify why later gates are unaffected by the relocation.","section":"Section V.C, Fig. 8"},{"comment":"The headline numbers are inconsistent. The abstract and conclusion report a 30% reduction in end-to-end latency for real benchmark applications, but Section V.B states that Cuccaro's Adder and MCMTV show 30%, while QFT and Quantum Volume show 42%. Similarly, the abstract reports a 24% depth reduction, while Section V.B reports 24% for Cuccaro's Adder and MCMTV but 45% for QFT and Quantum Volume. The paper should either report a single aggregate number with a clear averaging procedure or consistently report per-benchmark ranges.","section":"Section V.B, Abstract, Conclusion"},{"comment":"The simulation methodology is not described in enough detail to reproduce or verify the results. The text says 'we modify SeQUeNCe' but does not specify the modifications, the parameters for entanglement generation (success probability, decoherence times, classical communication latency), or the number of simulation runs. No error bars are shown in Figures 6-8, and no code or data repository is provided. Given that the quantitative claims are the main contribution, the experimental setup must be documented at a level that allows independent verification.","section":"Section V.A"}],"minor_comments":[{"comment":"There are grammatical errors, e.g., 'We proposes a decentralized framework' and 'the issue of transferring quantum information across the cores remains unresolved' should be 'remains'.","section":"Abstract"},{"comment":"The variable names are inconsistent: the pseudocode uses 'tele src dir' and 'tele dest dir' while the text and algorithm description use 'tel src dir' and 'tel dest dir'. Please unify.","section":"Section IV.B, Algorithm 1"},{"comment":"The figure caption mentions blue arrows for hop-by-hop and green arrows for two-way, but the printed figure appears in grayscale; consider using distinct line styles or labels so that the two paths are distinguishable.","section":"Section IV.B, Fig. 5"},{"comment":"The acronym 'MCMTV' is used without expansion; define it (presumably multi-controlled multi-target gate or similar).","section":"Section V.A, Benchmarks"},{"comment":"The congestion limitation in middle cores is mentioned only qualitatively. Since this is a known drawback of two-way teleportation with XY routing, a quantitative evaluation (e.g., qubit density or hot-spot metrics) would strengthen the discussion, even if only as a limitation.","section":"Section V.C"},{"comment":"The text says 'Barret-Kok generation principle' but the reference [17] is by Barrett and Kok; fix the spelling.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper's core idea is reasonable and the comparison to a baseline is appropriate for this venue. The main concern is that the central quantitative claims rest on a cost model that is not validated and appears to ignore the residual relocation of logical qubits across gates. If the authors can re-evaluate circuit depth with a consistent location-tracking model and provide complete simulation details, the paper could become acceptable. Given the current evidence, I cannot recommend acceptance. The abstract/body inconsistencies also need to be resolved before the paper can be considered."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuinely new, plausible optimization for inter-core qubit transfer, and the two-way teleportation variant is worth knowing about. But the headline reductions are not backed by the evidence as written. I would not take the 30–40% latency or 24% depth figures at face value without code and a better cost model.\n\nWhat is actually new: existing teleportation interconnects use a centralized EPR generator; the paper replaces it with per-link BSM nodes, and proposes moving both operands toward a middle core rather than moving the source all the way to the destination. The XY-routing protocol is simple, but the idea is not in the cited prior work. The paper also evaluates on small real circuits (QFT, Cuccaro adder, MCMT, Quantum Volume) in addition to synthetic traffic, which is the right instinct.\n\nSoft spots, in rough order of severity. First, there is no code, no data, and no description of the simulator parameters or SeQUeNCe modifications. Nobody can reproduce the latency numbers. Second, the abstract and conclusion say 30% real-benchmark latency reduction and 24% depth reduction, while Section V.B reports 30% for two benchmarks and 42% for two others, with the same split for depth (24% vs 45%). The headline understates the reported range, which is misleading. Third, the depth model in Section V.C is load-bearing: one depth unit per hop, per request, as if each inter-core gate is isolated. But both protocols physically relocate logical qubits. After a two-way teleport, both operands sit in a middle core; a later gate involving one of those qubits and a qubit that stayed home requires extra movement. The paper never tracks qubit locations across the circuit. That omission systematically favors two-way teleportation and likely overstates the depth reduction. The stress-test note is right. The latency comparison may survive a more careful accounting, but the current 24% depth figure is not supported.\n\nMinor concerns: congestion in middle cores is admitted but not modeled, and there are no error bars. The citation pattern looks fair; the self-citations to earlier multicore architecture work are relevant.\n\nWho this is for: researchers working on multi-core quantum architectures and circuit mapping. The paper deserves a serious referee, not a desk rejection. Ask for open artifacts, corrected headline numbers, and a location-aware cost model before trusting the quantitative claims.","headline":"Two-way teleportation is a plausible new optimization, but the headline numbers outrun the evidence: no artifacts, inconsistent reporting, and a depth model that ignores qubit relocation.","tokens_in":9106,"tokens_out":2515,"would_cite":false,"duration_ms":26354,"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":"Decentralized two-way teleportation cuts multi-core quantum latency by 40%.","keywords":["multicore quantum computing","quantum teleportation","decentralized interconnect","network-on-chip","two-way teleportation","circuit depth","communication latency","quantum entanglement"],"falsifier":"Run the same two-way protocol in a simulator or experiment that tracks actual circuit depth and wall-clock latency while varying the number of hops, including classical message delays, entanglement-generation retry rates, and the cost of later retrieving a qubit that left its home core; if one hop does not cost one depth unit, the reported 24% depth reduction and the part of the latency gain based on it are not supported.","tokens_in":1738,"feed_emoji":"⚛️","tokens_out":2802,"duration_ms":74922,"temperature":0.7,"pith_summary":"This paper tries to establish that inter-core qubit transfer in a multi-core quantum processor need not be a one-way journey from source to destination. It proposes a decentralized teleportation framework in which each adjacent pair of cores generates its own entanglement, and it introduces two variants: hop-by-hop teleportation, where only the source qubit moves, and two-way teleportation, where both the source and destination qubits move simultaneously toward a middle core, where the two-qubit gate is executed. The central claim is that two-way teleportation cuts end-to-end communication latency by about 40% on synthetic benchmarks, by 30% to 42% on real benchmark subroutines, and reduces circuit depth by 24% to 45% compared with hop-by-hop teleportation. A sympathetic reader would care because communication latency and circuit depth directly limit how long qubits must stay coherent, so reducing both could make modular, multi-core quantum architectures more practical.","feed_headline":"Two-way teleportation cuts multi-core quantum latency by 40%","feed_subtitle":"Source and destination qubits move toward a shared middle core, so gates fire sooner and circuits stay shallower.","key_machinery":"The mechanism is decentralized entanglement generation through Bell State Measurement nodes placed between adjacent quantum cores, combined with deterministic XY routing for movement. Each adjacent pair of cores entangles its communication qubits by emitting photons to the shared BSM node, and a successful measurement establishes an EPR pair that enables one teleportation hop. In two-way teleportation, the source and destination cores each route their relevant qubit along the XY path at the same time, so the qubits meet at an intermediate core, where the two-qubit gate is executed. This halves the number of sequential hops and, under the paper's cost model, reduces both end-to-end latency and compiled circuit depth.","core_discovery":"The paper's central claim is that two-way teleportation outperforms hop-by-hop teleportation in a decentralized multi-core quantum architecture by moving both operands of a distant two-qubit gate toward each other rather than moving one operand all the way to the other's core. In a simulated 4 by 4 mesh of quantum cores, with entanglement generated locally between adjacent cores through Bell State Measurement nodes, two-way teleportation reduces end-to-end communication delay by an average of 40% on random circuits of depth 5 and 10, and by 30% on the Cuccaro adder and MCMTV benchmarks and 42% on QFT and Quantum Volume. The paper also reports that compiled circuit depth decreases by 24% for the adder and MCMTV and by 45% for QFT and Quantum Volume. The explanation offered is that simultaneous movement halves the number of sequential teleportation hops before the gate can execute, and the gate fires at an intermediate core instead of the destination.","pith_inferences":["If the one-depth-unit-per-hop cost model is replaced by a more realistic accounting that includes classical handshakes and entanglement-generation retries, the optimal meeting point may shift; the same two-way protocol could be tuned rather than discarded.","The underlying idea, moving both operands toward a rendezvous point, is not specific to 2D meshes; it could plausibly extend to higher-dimensional or irregular topologies and even to multi-operand gates, though the paper only evaluates pairwise gates on a 4 by 4 mesh.","A direct testable extension is to vary mesh size and entanglement success probability: the paper's fixed 4 by 4 ideal-BSM setting leaves open whether the 40% figure persists under practical noise and retry overhead.","The acknowledged congestion at middle cores under deterministic XY routing suggests that adaptive routing, which chooses meeting points based on current load, could make two-way teleportation even more attractive in larger networks."],"forward_implications":["If the reported gains hold, inter-core communication latency for distant operand pairs scales with roughly half the source-to-destination distance, so applications like QFT and Quantum Volume, which involve many far-apart two-qubit gates, benefit the most.","Reduced circuit depth means qubits spend fewer layers waiting for distant gates, lowering their exposure to decoherence and potentially enabling larger computations within coherence limits.","Decentralized entanglement generation removes the single EPR-pair generator as a bottleneck, so multiple teleportations in disjoint regions of the mesh can proceed in parallel.","Executing two-qubit gates at an intermediate meeting core rather than at the destination means qubit traffic is redistributed across the network and destination cores no longer serve as the sole convergence point.","Compilers for multi-core quantum architectures could be designed to choose a meeting point for each two-qubit gate based on current qubit locations, rather than always routing one operand to the other's core."],"supporting_citations":[{"why":"Supplies the standard quantum teleportation protocol used for each inter-core hop.","marker":"[7]"},{"why":"Defines the centralized multicore architecture and communication model used as the point of comparison.","marker":"[5]"},{"why":"Supports the use of teleportation instead of swapping for large multi-core quantum architectures.","marker":"[16]"},{"why":"Provides the entanglement-generation principle used to entangle communication qubits through BSM nodes.","marker":"[17]"},{"why":"Provides the open-source simulator that the experiments are built on.","marker":"[18]"},{"why":"Supplies the Quantum Fourier Transform benchmark subroutine used in the real-benchmark evaluation.","marker":"[19]"},{"why":"Supplies the ripple-carry adder benchmark used in the real-benchmark evaluation.","marker":"[20]"},{"why":"Characterizes the inter-core connectivity patterns that distinguish QFT and adder benchmarks.","marker":"[21]"}],"fun_headline_variants":["Two-way teleportation halves hops in quantum core mesh","Decentralized teleportation cuts multi-core quantum latency up to 40%","Move both qubits toward each other for 40% faster teleportation","Quantum mesh teleportation: two-way cuts latency by up to 40%","Circuit depth drops 24% with two-way teleportation"],"cache_read_input_tokens":11264,"weakest_assumption_plain":"The load-bearing premise is the cost model that each teleportation hop that moves a qubit one core closer adds exactly one unit to circuit depth, with no overhead counted for classical coordination, failed entanglement attempts, or the fact that the destination's qubit leaves its home core.","fun_headline_variants_meta":{"raw":{"variants":["Two-way teleportation halves hops in quantum core mesh","Decentralized teleportation cuts multi-core quantum latency up to 40%","Move both qubits toward each other for 40% faster teleportation","Quantum mesh teleportation: two-way cuts latency by up to 40%","Circuit depth drops 24% with two-way teleportation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001444,"raw_usage":{"total_tokens":5814,"prompt_tokens":940,"completion_tokens":4874,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":556,"completion_tokens_details":{"reasoning_tokens":4778}},"tokens_in":556,"tokens_out":4874,"duration_ms":32953,"temperature":1.0,"reasoning_tokens":4778,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:55:29.607588+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same two-way protocol in a simulator or experiment that tracks actual circuit depth and wall-clock latency while varying the number of hops, including classical message delays, entanglement-generation retry rates, and the cost of later retrieving a qubit that left its home core; if one hop does not cost one depth unit, the reported 24% depth reduction and the part of the latency gain based on it are not supported.","supporting_citations":[{"cited_title":"Teleporting an Unknown Quantum State via Dual Classi- cal and Einstein-Podolsky-Rosen Channels,","cited_arxiv_id":null,"evidence_quote":"Supplies the standard quantum teleportation protocol used for each inter-core hop."},{"cited_title":"Assessing the Role of Communication in Scalable Multi-Core Quantum Architectures,","cited_arxiv_id":null,"evidence_quote":"Defines the centralized multicore architecture and communication model used as the point of comparison."},{"cited_title":"Building Quantum Wires: The Long and the Short of it,","cited_arxiv_id":null,"evidence_quote":"Supports the use of teleportation instead of swapping for large multi-core quantum architectures."},{"cited_title":"Efficient high-fidelity quantum computation using matter qubits and linear optics,","cited_arxiv_id":null,"evidence_quote":"Provides the entanglement-generation principle used to entangle communication qubits through BSM nodes."},{"cited_title":"SeQUeNCe: a customizable discrete-event simulator of quantum networks,","cited_arxiv_id":null,"evidence_quote":"Provides the open-source simulator that the experiments are built on."},{"cited_title":"Quantum arithmetic with the Quantum Fourier Transform,","cited_arxiv_id":null,"evidence_quote":"Supplies the Quantum Fourier Transform benchmark subroutine used in the real-benchmark evaluation."},{"cited_title":"Characterizing the Inter-Core Qubit Traffic in Large-Scale Quantum Modular Architectures","cited_arxiv_id":"2310.01921","evidence_quote":"Characterizes the inter-core connectivity patterns that distinguish QFT and adder benchmarks."}],"review_version":1}