{"id":"74ebf135-36f6-4b3e-94a9-deac2b7eba10","arxiv_id":"2505.14519","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The authors introduce oblivious quantum teleportation and control protocols that let distributed quantum computers execute and combine unknown subroutines as black boxes.","lead":"This paper proposes a way to run quantum computations distributed across multiple chips even when the parties do not know what subroutine the other party is running. The method teleports unknown quantum gates and controls them without learning them, which could help scale up quantum computers and enable secure remote computing.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (14) is inconsistent with Eq. (13): iterating OQT for n=2, s=1 gives a minus sign on the U-term, so the parity in Eqs. (14)-(15) should be the number of nontrivial outcomes s, not the step count n; the manuscript never states the needed sign correction.","rationale":"The paper builds a plausible framework from standard tools (Choi states, teleportation, ISI). The OQC flag identity (U tensor U*) |omega> = |omega> is correct, and the exponential suppression of the signal with s is explicitly acknowledged in Sec. III, so it is not a hidden flaw. However, the derivation of the multi-step OQT final state in Sec. III is internally inconsistent: the displayed parity depends on n while direct iteration gives (-1)^s. The reader's weakest assumption identifies exactly this point. The issue is concrete and testable, and it directly supports the CONDITIONAL verdict: the manuscript must correct the parity and specify the sign/offset data processing before the universality claim can be accepted as stated. I see no reason to move to reject or accept.","tokens_in":19302,"tokens_out":15184,"duration_ms":134944,"concrete_test":"Re-derive the composition recurrence: with A0=0, B0=1, after a 0-outcome (A,B) -> (A,B), and after a 1-outcome (A,B) -> ((d-A)/(d^2-1), -B/(d^2-1)). Then compute the two-step case d=2, U1=U2=I, |psi>=|0>, outcome (P1,P0): the resulting density matrix is diag(1/3,2/3). Compare with Eq. (14) and Eq. (15): it matches the minus-sign form. This settles that the sign follows s parity, not n parity, and identifies the required correction.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central composition formula (14) is not a consequence of the single-step rule (13). Applying (13) twice for a sequence U2U1 with one nontrivial outcome (s=1) gives, whether the nontrivial outcome is first or second, (d 1 - U psi U^dag)/(d^2 - 1), i.e., the coefficient of U psi U^dag is -1/(d^2 - 1). Eq. (14) for even n would require +1/(d^2 - 1). As a concrete counterexample, take d=2, U1=U2=I, |psi>=|0>: the two-step state for outcome (P1,P0) is diag(1/3,2/3), which matches Eq. (15) (odd-parity minus), not Eq. (14). More generally, the recurrence for the U-coefficient gives (-1)^s/(d^2 - 1)^s, independent of n. Thus the parity label in Eqs. (14)-(15) appears to be s, not n. If so, the protocol can be repaired by recording s and flipping the sign of the U-dependent term for odd s, but the paper does not say this; as written, any observable estimate based on Eq. (14) is wrong for runs with odd s. This is exactly the load-bearing assumption for the claimed universality of OQT.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a framework for distributed quantum computing in which subroutines are black-box and supplied as Choi states. Three primitives are introduced: oblivious program execution (OPE), oblivious quantum teleportation (OQT), and oblivious quantum control (OQC). OQT is claimed to compose an arbitrary sequence of unknown unitaries while leaving only a depolarizing offset whose parity determines the sign of the recovered U|ψ⟩⟨ψ|U† term; Eqs. (14)-(15) give the composed form. The paper also constructs oblivious versions of DQC1, amplitude amplification, and state superposition, discusses physical realization on matter-qubit and photonic platforms, and sketches a multi-party distributed protocol. The central technical claim is that these primitives allow observable statistics of compositions of unknown subroutines to be evaluated without learning or correcting byproducts.","tokens_in":19594,"tokens_out":9167,"duration_ms":94143,"significance":"If the composition formula is repaired, the framework is a conceptually useful unification of channel-state duality, teleportation, and twirling, and the paper correctly identifies that the signal-to-offset ratio decays as (d²-1)^{-s}. The derivations have no fitted parameters and rely on standard results, which is a strength; the physical-implementation discussion and the comparison table provide useful context. However, the load-bearing composition formula is currently wrong as stated, and OQC and the general protocol are only asserted schematically, so the claimed universality is not established in the present version.","major_comments":[{"comment":"Equations (14)-(15) are not consequences of Eq. (13). Iterating the single-step rule for U = U2U1 with exactly one nontrivial outcome (s = 1) gives (d I - U ψ U†)/(d² - 1), whether the nontrivial measurement occurs on the first or the second teleportation; the coefficient of U ψ U† is -1/(d² - 1), whereas Eq. (14) for even n requires +1/(d² - 1). The general recurrence for the signal coefficient is (-1)^s/(d² - 1)^s, so the two forms should be labelled by the parity of s, not by the parity of n. As a check, for d = 2, U1 = U2 = I, |ψ⟩ = |0⟩, the two-step outcome (P1, P0) gives diag(1/3, 2/3), which has the odd-s minus sign and contradicts Eq. (14). Eq. (15) is also wrong for s = 0 and odd n, where the final state is U ψ U†, not β I - U ψ U†. Since the claimed universality of OQT and every observable estimate extracted from the final state depend on the exact sign and prefactor of the signal term, this must be corrected and explicitly derived in the revision.","section":"Sec. III, Eqs. (14)-(15)"},{"comment":"The oblivious quantum control primitive is asserted rather than constructed. The text states that (U ⊗ U*)|ω⟩ = |ω⟩ and that this 'naturally satisfies' the flag condition, and then says 'it is clear to see U ↦ ∧U is realized in the proper subspace'. No circuit or derivation is given for how the eigenstate is used to produce the controlled operation P0 ⊗ I + P1 ⊗ U on the data register. Because OQC is the basis for ODQC1 and OQS, this missing construction is load-bearing; the revision should provide an explicit circuit or a formal proof, or clearly reproduce the construction it relies on.","section":"Sec. IV, Eq. (17) and Fig. 3"},{"comment":"The manuscript promises a universal distributed quantum computing protocol, but Sec. VII B is only a schematic description: it refers to a quantum superchannel, a classical optimizer, and a space-time tradeoff without stating a theorem about which tasks are realized, how OQT composition behaves for heterogeneous local dimensions, or what the communication and resource costs are. The two-party and tri-party examples in Sec. VII A do not constitute a universality proof. Please add a precise statement of the protocol's input/output, correctness condition, and overhead, or explicitly restrict the claimed scope.","section":"Sec. VII B (general protocol)"}],"minor_comments":[{"comment":"The notation 'd1' in Eq. (13) should be written as d I (or d 1 with the identity explicitly indicated); as printed it looks like a scalar d1.","section":"Sec. III, Eq. (13)"},{"comment":"The statement that 'the minimal number of samples is two' is unexplained; please specify what is being counted and why two is minimal in the space-time tradeoff.","section":"Sec. VI A, around Fig. 10"},{"comment":"There is a typo in the list of protocol features: 'Forth' should be 'Fourth'.","section":"Sec. I"},{"comment":"The claim that 'for two unknown states |ψ⟩ and |ϕ⟩, there is no algorithm to compute ⟨ψ|ϕ⟩' needs a qualifier, since the SWAP test can estimate |⟨ψ|ϕ⟩|² from copies; the intended statement is presumably about deterministic exact computation without additional assumptions.","section":"Sec. V A"}],"recommendation":"major_revision","confidential_remarks":"The sign error in Eqs. (14)-(15) is the main technical obstacle; it is local and repairable, but it sits in the core composition lemma and must be fixed with an explicit derivation. I would also ask the authors to clarify the novelty and formal status of the general protocol relative to the prior work by the corresponding author cited as Refs. [8, 37, 39], since several primitives are presented as 'schemes' without full derivations. If those two points are addressed, I would be willing to reconsider."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version: this paper proposes a distributed quantum computing scheme based on two primitives—oblivious quantum teleportation (OQT) and oblivious quantum control (OQC)—that let parties apply unknown subroutines stored as Choi states without learning or correcting the teleportation byproducts. The single-shot OQT formula (Eq. 13) is a straightforward Pauli-twirling calculation and checks out, and the OQC trick of using the ebit as a known eigenstate is clean. What's genuinely new is the claim that you can compose multiple such teleportations obliviously, with the final state depending only on the number of nontrivial outcomes s, not their order. If that held up, it would indeed give a communication-efficient way to execute black-box programs and a space-time tradeoff for circuit depth.\n\nBut the composition formula doesn't hold up as written. Eqs. (14)-(15) put the sign of the signal term U(ψ) as even/odd in n, the number of steps. Working the recurrence from Eq. (13) gives the sign as (-1)^s, independent of n. The concrete counterexample with d=2, U1=U2=I, |ψ>=|0> and one nontrivial outcome gives diag(1/3, 2/3), which matches the odd-s sign, not the even-n sign. So any run with an odd number of nontrivial outcomes produces the wrong sign in the observable estimate unless someone adds a parity correction that the paper never mentions. That's load-bearing: the claimed universality of OQT rests on being able to extract observable statistics from these final states.\n\nThere's also a lesser issue: the general multi-party protocol in Sec. VII is schematic. It's plausible, but there's no formal correctness statement, and the security discussion is appropriately hedged to passive adversaries, which is good. The heavy reliance on the corresponding author's prior work is not itself a flaw, since those results are cited properly, but it does mean the novelty is mostly in the composition step—which is exactly the step that's wrong.\n\nIf the sign error is fixable, this could be a solid paper. As it stands, the central claim isn't supported. I'd send it to review because it deserves a referee to work through the recurrence, but I wouldn't cite it in its current form.","headline":"The core idea—composing unknown unitaries via teleportation with discarded byproducts—is plausible and worth engaging, but the central composition formula has a sign error that undercuts the stated universality claim.","tokens_in":20143,"tokens_out":4729,"would_cite":false,"duration_ms":40735,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68","81P45"],"pacs":["03.67.Lx","03.67.Hk"],"model":"deepseek-v4-flash","headline":"The paper proposes that distributed quantum computers can execute programs made of arbitrary unknown subroutines, using only Choi-state programs and parity measurements, without decoding or correcting the subroutines.","keywords":["distributed quantum computing","oblivious quantum teleportation","oblivious quantum control","black-box subroutines","Choi-state programs","quantum superchannels","circuit-depth reduction"],"falsifier":"A direct two-step calculation for $d=2$: apply Eq. (13) first for $U_1$ and then for $U_2$ with one nontrivial outcome, and compare the coefficient of $U_2U_1\\psi U_1^\\dagger U_2^\\dagger$ in the resulting density operator with Eq. (14). If the coefficient is $-1/(d^2-1)^2$ rather than $+1/(d^2-1)^2$, the recurrence as stated fails and a parity correction is required; equivalently, a numerical simulation of two sequential OQT runs measuring an observable on $U_2U_1|\\psi\\rangle$ would reveal the discrepancy.","tokens_in":19108,"feed_emoji":"🔗","tokens_out":9242,"duration_ms":81524,"temperature":0.7,"pith_summary":"Distributed quantum computers, the paper argues, do not need to know what a subroutine does in order to run it. The authors propose a universal protocol in which arbitrary unknown subroutines are stored as Choi states and composed by oblivious quantum teleportation and oblivious quantum control, with only parity measurements and no correction of teleportation byproducts. If the central recurrence is right, this lets multi-chip machines execute black-box programs with less communication, allows a space-time tradeoff that lowers circuit depth, and extends to unknown channels via quantum superchannels. The authors also argue the required measurements and gates are within reach of current superconducting, trapped-ion, cold-atom, and photonic platforms.","feed_headline":"Oblivious teleportation lets chips run unknown quantum subroutines","feed_subtitle":"A small set of oblivious primitives composes black-box quantum programs with less communication and shallower circuits.","key_machinery":"The central machinery is the Choi-state program $|U\\rangle = (U\\otimes \\mathbb{1})|\\omega\\rangle$, which stores an unknown operation in a bipartite state, together with the indirect binary Bell measurement $\\{P_0=|\\omega\\rangle\\langle\\omega|,\\ P_1=\\mathbb{1}-|\\omega\\rangle\\langle\\omega|\\}$ that underlies oblivious program execution and OQT. Grouping Pauli byproducts into trivial and nontrivial parity converts teleportation into an oblivious operation with the two-branch output of Eq. (13), and iterating yields the outcome-count recurrence of Eqs. (14) and (15). The same ebit $|\\omega\\rangle$, as an eigenstate of $U\\otimes U^*$, serves as the flag that makes OQC possible, supplying the controlled version of an unknown gate without a separately known eigenstate.","core_discovery":"The central claim is that unknown unitary subroutines can be composed and evaluated without ever being decoded. The paper introduces oblivious quantum teleportation (OQT), which replaces the full Bell measurement of ordinary teleportation with an indirect binary Bell measurement that only records whether the Pauli byproduct is trivial or nontrivial; the trivial branch delivers $U|\\psi\\rangle$, and the nontrivial branch delivers a depolarized state $\\frac{1}{d^2-1}(d\\mathbb{1}-U\\psi U^\\dagger)$, from which the same observable statistics can be extracted. Iterated over a sequence $U_n\\cdots U_1$, the paper claims the output depends only on the number of nontrivial outcomes, not their positions or identities, giving $\\alpha\\mathbb{1} + U(\\psi)/(d^2-1)^s$ for even $n$ and $\\beta\\mathbb{1} - U(\\psi)/(d^2-1)^s$ for odd $n$. The companion oblivious quantum control (OQC) scheme realizes $\\wedge U$ for unknown $U$ by using the ebit $|\\omega\\rangle$, an eigenstate of $U\\otimes U^*$, as a natural flag, and combining these gives a distributed black-box quantum computing protocol (DBQC) that evaluates observables of $U|\\psi\\rangle$ across physically separated chips.","pith_inferences":["If the sign in Eq. (14) is correct, OQT gives a conceptually simple, parameter-free way to concatenate unknown channels; this suggests a direct experimental test on two-qubit platforms where the parity measurement can be implemented without the full three-qubit Toffoli gate.","One can read the protocol as a resource identity: each OQT step costs one ebit and one parity measurement, so communication cost scales with the number of black-box subroutines rather than with the classical length of their gate decompositions.","Combining OQT with randomized measurements would turn the scheme into a black-box subroutine version of the Hadamard test, estimating $\\langle\\psi|U_2U_1|\\psi\\rangle$ for two unknown gates with only a few ebits and no decoding.","The gap between honest-but-curious and malicious security suggests that adding lightweight verification, such as test programs with known outputs, could upgrade DBQC toward secure multiparty quantum computation, an extension the paper leaves open."],"forward_implications":["A subroutine held by one party can be executed inside a computation running on another party's chip without uploading the subroutine's classical description or correcting its teleportation byproducts, reducing communication to shared ebits and reported measurement parities.","Because the OQT output depends only on the number of nontrivial parity outcomes, a sequence of unknown programs can be composed in parallel, and teleporting the computation to fresh qubits trades space for reduced circuit depth.","OQC turns unknown unitaries into controlled operations, making DQC1, amplitude amplification, and LCU-based superpositions oblivious, so they can estimate quantities such as $\\operatorname{tr}(U\\rho)\\operatorname{tr}(U^*\\eta)$ even when $U$ is unknown.","Via channel-state duality and the superchannel formalism, the protocol extends from unitaries to non-unitary channels, so unknown-channel estimation and sequential channel discrimination can be broken into parallel transversal steps.","In the multiparty setting, the protocol realizes a weak, passive-adversary form of oblivious transfer of quantum information, though the paper notes that malicious participants who send wrong measurement outcomes can break it."],"supporting_citations":[{"why":"States the no-programming theorem, which motivates storing a program as a Choi state and demanding only observable access rather than full gate execution.","marker":"[7]"},{"why":"Supplies the stored-program quantum teleportation and the initial-state injection scheme that the oblivious teleportation modifies.","marker":"[8]"},{"why":"Defines standard quantum teleportation whose Bell measurement is replaced by the parity measurement in OQT.","marker":"[47]"},{"why":"Proves the no-control theorem for unknown operations, the limitation OQC circumvents with an ebit flag.","marker":"[33]"},{"why":"Provides the earlier bypass strategy for oblivious program handling that OQT and OQC improve on.","marker":"[37]"},{"why":"Establishes channel-state duality, the foundation of Choi-state programs and of extending the protocol to channels.","marker":"[41, 42]"},{"why":"Introduces quantum supermaps, the framework used to describe composition of unknown programs and the general distributed protocol.","marker":"[38]"},{"why":"Gives the flag condition for controlled unknown operations, here satisfied naturally by the ebit eigenstate.","marker":"[51]"},{"why":"Introduces oblivious amplitude amplification, a primitive the protocol uses to boost success probability and reduce circuit depth.","marker":"[27]"}],"fun_headline_variants":["Oblivious teleportation runs black-box quantum subroutines","Black-box subroutines composed across chips without decoding","Distributed chips run unknown subroutines via oblivious teleportation","Oblivious teleportation and control compose black-box subroutines"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The protocol's universality depends on the exact sign and prefactor of the useful term in the teleported state after an even number of steps: if that term comes out with the opposite sign, every observable estimate would need a correction the paper does not provide.","fun_headline_variants_meta":{"raw":{"variants":["Oblivious teleportation runs black-box quantum subroutines","Black-box subroutines composed across chips without decoding","Distributed chips run unknown subroutines via oblivious teleportation","Oblivious teleportation and control compose black-box subroutines"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001143,"raw_usage":{"total_tokens":4734,"prompt_tokens":925,"completion_tokens":3809,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":3734}},"tokens_in":541,"tokens_out":3809,"duration_ms":24753,"temperature":1.0,"reasoning_tokens":3734,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T15:32:23.080345+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct two-step calculation for $d=2$: apply Eq. (13) first for $U_1$ and then for $U_2$ with one nontrivial outcome, and compare the coefficient of $U_2U_1\\psi U_1^\\dagger U_2^\\dagger$ in the resulting density operator with Eq. (14). If the coefficient is $-1/(d^2-1)^2$ rather than $+1/(d^2-1)^2$, the recurrence as stated fails and a parity correction is required; equivalently, a numerical simulation of two sequential OQT runs measuring an observable on $U_2U_1|\\psi\\rangle$ would reveal the discrepancy.","supporting_citations":[{"cited_title":"Wang, Weak, strong, and uniform quantum simu- lations, Phys","cited_arxiv_id":null,"evidence_quote":"Defines standard quantum teleportation whose Bell measurement is replaced by the parity measurement in OQT."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the no-control theorem for unknown operations, the limitation OQC circumvents with an ebit flag."},{"cited_title":"Araujo, A","cited_arxiv_id":null,"evidence_quote":"Provides the earlier bypass strategy for oblivious program handling that OQT and OQC improve on."},{"cited_title":"Oszmaniec, A","cited_arxiv_id":null,"evidence_quote":"Introduces quantum supermaps, the framework used to describe composition of unknown programs and the general distributed protocol."},{"cited_title":"Luo, Y.-Z","cited_arxiv_id":null,"evidence_quote":"Introduces oblivious amplitude amplification, a primitive the protocol uses to boost success probability and reduce circuit depth."}],"review_version":1}