Parallel Repetition in the Two-Player Quantum Cloning Game

Eli Coe Naig; Stephen A. Fenner

REVIEW 2 major objections 4 minor 8 references

Parallel Repetition in the Two-Player Quantum Cloning Game

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The two-copy quantum cloning game beats strong parallel repetition: the value lies strictly between $(3/4)^2$ and $\cos^4(\pi/8)$, with explicit bounds $\frac{5+\sqrt{17}}{16}$ and $\frac{11+\sqrt{65}}{32}$.

desk verdict Settles the n=2 tightness question with an explicit challenge-dependent counterexample to strong parallel repetition and a strictly improved upper bound; the math is careful and checkable, with the main caveat that the 64x64 eigenvector computation is asserted rather than displayed. read the letter →

arxiv 2608.00930 v1 pith:MKS3YAVB submitted 2026-08-02 quant-ph

Eli Coe Naig , Stephen A. Fenner This is my paper · ORCID

classification quant-ph MSC 81P4581P68 PACS 03.67.-a

keywords quantumcloninggameparallelrepetitionmonogamyofentanglementpositionverificationblockGrammatrixcapchallenge-dependentstrategystrong

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper asks whether playing the two-player quantum cloning game twice in parallel multiplies the win probability, as strong parallel repetition would predict. Earlier work put the $n$-copy value between $(3/4)^n$ and $\cos^{2n}(\pi/8)$, leaving both endpoints open. Here the authors prove that for $n=2$ neither endpoint is tight: an explicit challenge-dependent strategy wins with probability at least $(5+\sqrt{17})/16>9/16$, while a block-Gram argument bounds the value by $(11+\sqrt{65})/32<\cos^4(\pi/8)$. They also show that challenge-independent strategies are stuck at exactly $(3/4)^n$ for every $n$, so the gain is inherently coordinated across rounds. The immediate stake is whether copying games behave like single games under repetition, with consequences for quantum position-verification protocols.

What carries the argument

The paper's central object is the block Gram matrix $[M_{x,y}]_{x,y}$ whose entries are $M_{x,y}=C_x S_x S_y^\dagger C_y^\dagger$, with $C_x$ the Bell-test contraction for challenge $x$ and $S_x$ the challenge-dependent response unitary. By cyclicity of trace and the identity of nonzero spectra, the game value equals $2^{-n}$ times the supremum over response unitaries of the operator norm of this block matrix. The key bound is the directional overlap lemma: for $x\ne y$, $\|M_{x,y}\|\le 2^{-\max(t_A,t_B)}$, where $t_A$ counts coordinates where $x$ tests Alice under $y$, and $t_B$ counts the reverse; this improves on the Hamming-distance-only bound $2^{-|x\oplus y|/2}$. These caps form the ca

What would settle it

Numerically optimize general strategies for the two-copy game, not restricted to unitaries; if any success probability exceeds $(11+\sqrt{65})/32$, the cap-matrix upper bound is false. Conversely, recompute the largest eigenvalue of the explicit $64\times64$ integer matrix $K=4H$: it should be exactly $5+\sqrt{17}$, and any deviation falsifies the lower-bound construction.

Watch

Extended reading notes

Core claim

The central discovery is that strong parallel repetition fails for the unrestricted two-player quantum cloning game. Writing $\omega^*(QCG_2^{\times2})$ for the best two-copy success probability, the paper establishes $$(3/4)^2 < \frac{5+\sqrt{17}}{16} \le \omega^*($QCG_2^{{\times2}}$) \le \frac{11+\sqrt{65}}{32} < \$cos^{4}$(\pi/8).$$ The lower bound is constructive: choosing the response unitaries $U^{01}_A=V^{10}_B=G$ (a controlled $-iY$ gate), $U^{10}_A=V^{01}_B=\mathrm{SW}\,G\,\mathrm{SW}$, and the identity elsewhere, the averaged acceptance operator $H$ satisfies $K=4H$, a $64\times64$ integer matrix whose largest eigenvalue $5+\sqrt{17}$ is certified by an explicit eigenvector $w$. The upper

Load-bearing premise

The operator-norm formula depends on representing every possible local response by a challenge-dependent unitary on a fixed output register, via Stinespring dilation; if a non-unitary channel—one that measures or discards qubits—could outperform all unitary strategies, both new bounds would collapse.

Editorial extensions

If this is right

Strong parallel repetition fails for the unrestricted cloning game: two copies are worth strictly more than $(3/4)^2$.
The exact two-copy value lies in the interval $[(5+\sqrt{17})/16,\,(11+\sqrt{65})/32]$, with neither endpoint proved tight.
The cap-matrix upper bound $2^{-n}\|N_n\|$ strictly improves the previous $\cos^{2n}(\pi/8)$ bound for every $n$.
Challenge-independent strategies achieve exactly $(3/4)^n$ for every $n$, so any super-multiplicative advantage requires challenge-dependent responses.
The lower-bound strategy does not by itself break the routed quantum position-verification protocol of the original model, because that model restricts the initial state and the referee's marginal.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: the same directional-cap matrix idea may give nontrivial parallel-repetition bounds for other monogamy-of-entanglement games, wherever the overlap of acceptance projectors depends on the orientation of changed coordinates.
Inference: the exact two-copy value is pinned between two algebraic numbers of the same quadratic-field form; if the true value is one of the endpoints, it would suggest a general frozen-eigenvalue mechanism for two-copy repetition values.
Inference: because challenge-independent strategies are exactly multiplicative, the super-multiplicative gain isolates a coordination resource: the players must know not just which positions are tested but in which direction, which is testable in smaller instances.
Inference: an immediate extension is to compute the three-copy cap matrix $N_3$ and compare $\|N_3\|/8$ against $\cos^6(\pi/8)$ to see how quickly the directional improvement grows with $n$.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this paper proves strong parallel repetition fails for the two-copy unrestricted quantum cloning game, and it gives a strictly improved upper bound for every n. The results are genuinely new relative to Colisson, Palais, Escolà-Farràs, and Speelman, who left both endpoints open. The explicit lower-bound strategy (5+sqrt17)/16 is a real construction, and the cap-matrix technique—retaining directional t_A/t_B information instead of collapsing to Hamming distance—is a clean idea that provably beats the previous bound.

What's good: the operator-norm formulation in Section 2 is standard but clean; the directional overlap Lemma 3.1 is a nice refinement of the prior projector argument; the cap matrix N2 has spectrum that I independently checked (trace 4, Frobenius squared 53/8, eigenvalues as stated); the Halmos argument for Proposition 4.5 is elegant; and the Perron–Frobenius step in 5.2 is rigorous. The paper is honest about scope: the counterexample is for the unrestricted game and does not touch the routing QPV protocol. For a subfield result, the significance is real: it resolves an explicit open tightness question and shows strong parallel repetition fails in a natural monogamy game.

Soft spots: the lower bound in Theorem 4.1 depends entirely on two finite computations: the characteristic polynomial of the 64x64 integer matrix K, and the assertion that the written-out vector w satisfies Kw = (5+sqrt17)w 'by direct computation.' Neither is shown, and the cited verify_counterexample.py is not embedded in the text. This is not a hidden black box—the vector w is fully specified, so a referee can multiply K times w and check the eigenvalue equation by hand or with any CAS. But the paper should, for acceptance, provide the script or include the resulting product as an appendix; as written, a reader has to trust the assertion or redo the work. That's a presentation or verifiability issue, not a correctness flaw unless the computation is actually wrong. I did not find an error.

The other assumptions—the Stinespring reduction to local unitaries, the norm-matrix bound, the challenge-independent exact value—are all standard and appear sound. No circularity.

This paper is for quantum information readers working on monogamy-of-entanglement games, parallel repetition, and QPV. It deserves a serious referee; the claims are checkable and the main result is an open-question resolution. My recommendation: send it to peer review, and in the review ask the authors to make the computer verification openly available or to include the explicit eigenvector equation in full.

Referee Report

2 major / 4 minor

Summary. The paper studies parallel repetition of the two-player quantum cloning game QCG_2. Using a block Gram matrix formulation, the value is expressed as 1/2^n times the operator norm of a matrix built from local response unitaries. The authors prove directional caps on the off-diagonal blocks (Lemma 3.1), assemble them into a cap matrix, and for n=2 obtain the upper bound (11+√65)/32 < cos^4(π/8). They then construct an explicit challenge-dependent strategy involving a 64×64 integer matrix K; using stated computer-assisted computations (a characteristic polynomial and an eigenvector w) they claim λmax(K)=5+√17 and hence value (5+√17)/16 > 9/16, disproving strong parallel repetition. They also prove that challenge-independent strategies have exact value (3/4)^n for every n, and that the cap-matrix bound improves the previous upper bound for every n.

Significance. If the main claims hold, the paper settles a natural open question for this monogamy game: strong parallel repetition fails already at two copies, and the previously known interval is strictly tightened. The conceptual contribution is the directional refinement of the overlap bound into a cap matrix, and the construction of an explicit challenge-dependent lower-bound strategy. The upper-bound and challenge-independent proofs are clean, parameter-free, and non-circular. The principal weakness is that the central counterexample rests on a nontrivial finite computation that is only asserted in Remark 4.2; the referenced ancillary script is not included. Once a machine-checkable certificate is supplied, the central result would be independently verifiable and the paper would be suitable for publication.

major comments (2)

[Theorem 4.1 / Remark 4.2] The proof of the lower bound relies on two computational assertions: det(λI−K) = λ^48(λ−2)^2(λ−3)^3(λ−4)^2(λ−5)^3(λ^2−7λ+4)(λ^2−10λ+8)(λ^2−11λ+22), and Kw = (5+√17)w 'by direct computation'. Neither computation is displayed, and the ancillary script mentioned in Remark 4.2 is not present in the submitted text. The eigenvector assertion is the load-bearing certificate for the claim (5+√17)/16 > 9/16; the characteristic polynomial is used to identify the largest eigenvalue and hence the exact strategy value. A typo in either assertion would invalidate the exact-value claim, and an error in the vector check would break the counterexample. Please provide the code or an exact-arithmetic transcript, or an independent analytic verification (e.g., modular factorization for the characteristic polynomial and a short script verifying Kw).
[Section 2, Eq. (4)] The reduction from arbitrary local response channels to challenge-dependent local unitaries is stated very tersely. The Stinespring argument is standard, but as written it does not explicitly justify why retaining the environment in the private registers preserves the success probability, since a general channel would require a partial trace over the environment. Please spell out the argument: for any POVM E on the output registers, Tr[(I_env ⊗ E) U(ρ⊗|0><0|)U†] = Tr[E Φ(ρ)]. This is not a correctness concern, but it supports the central operator-norm formula and should be made precise.

minor comments (4)

[Eq. (1)] The expression '(1/2 + 1/2√2)^n' is ambiguous. It should read $\frac12 + \frac{1}{2\sqrt2}$, not $\frac12 + \frac{\sqrt2}{2}$.
[Remark 4.3] The claim that replacing −iY by any anti-diagonal unitary W_{α,β} leaves the strategy value unchanged is not proved. Since this is a remark rather than a load-bearing step, it would help to add one sentence explaining the diagonal phase conjugation.
[Theorem 4.1 proof] The statement about the largest root of the characteristic polynomial could be made explicit: the roots of λ^2−10λ+8 are 5±√17, and all other roots are smaller. This is immediate but would help the reader.
[Remark 4.2] The claimed value ∥w∥² = 1972−476√17 is not derived. It is not needed for the main argument, but if kept it should be verified or removed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: explicit constructions and independent norm estimates carry the claims; the only gap is an unshown finite computation, which is a verification issue, not circularity.

full rationale

The paper's derivation chain is self-contained and not circular. The lower bound (Theorem 4.1) is an explicit strategy: the gate G, the matrix K = 4H, the stated characteristic polynomial, and the explicit vector w with Kw = (5+√17)w. The claimed value (5+√17)/16 is obtained by computing λmax(K)/16 from these finite, explicit objects; it is not assumed or fitted. The upper bound (Corollary 5.1) follows from the norm-matrix bound (Lemma 2.1), the directional cap bound (Lemma 3.1), and the spectrum of N2; no prior bound is used as input. Proposition 4.5 is proved directly via Halmos's two-subspaces theorem, and Proposition 5.2 uses Perron–Frobenius and character diagonalization to compare caps with a diagonalizable matrix. Citations are external and standard: [1] is a different research group and is used only as a benchmark or starting point; [5] and [6] are textbook and classical results. The only notable weakness is that the characteristic polynomial and the eigenvector relation are delegated to an ancillary script (Remark 4.2), meaning the proof relies on an unshown finite computation; this is a reproducibility/correctness concern, not a circular one, because those assertions are independent of the theorems they support. No step reduces by construction to its own inputs.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The derivations rest on standard quantum-information facts (Stinespring dilation, Halmos two-subspaces, Perron-Frobenius, tensor-product norms) plus the explicit definition of the unrestricted game model. The one paper-specific input is the characteristic polynomial factorization of K, a computer-assisted assertion; it is certified by the explicit eigenvector w (which alone implies the lower bound beats (3/4)²). No free parameters are fitted; no new physical entities are postulated.

assumptions (6)

standard math Arbitrary local response channels can be reduced to challenge-dependent local unitaries acting on fixed output registers (Stinespring dilation; environment absorbed into private registers).
Section 2, between Eqs. (3) and (4); cited to Watrous [5]. This underpins the operator-norm formula (4).
standard math Halmos's two-subspaces theorem: for projections P,Q, the sum P+Q has eigenvalues 1±s on two-dimensional blocks where s is a singular value of PQ.
Proposition 4.5 proof; cited [6].
standard math Perron-Frobenius theorem for entrywise positive symmetric matrices gives a positive Perron vector and strict norm inequalities.
Proposition 5.2 proof, used to get the strict inequality ||N_n|| < ||N̂_n||.
standard math Single-coordinate Bell calculus: J_A^†J_B = (1/2)Λ and isometric/coisometric norm preservation give ||Γ_AΓ_B|| = 1/2.
Section 3, Eq. (9); derived in-text.
domain assumption Unrestricted game model: the initial state ρ ranges over all density operators on R⊗A⊗B⊗E_AE_B, with no constraint on the referee's marginal.
Section 2 ('No constraint is placed on the shared initial state or on the referee's marginal'). The central claims are about this model; the paper notes the routing model of [1, Section 5] is different.
ad hoc to paper The characteristic polynomial factorization of the 64×64 integer matrix K is correct as stated.
Theorem 4.1 proof; a computer-assisted assertion. Certified by the claimed ancillary script and partly by the explicit eigenvector w, which alone suffices for the lower bound > 9/16.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Parallel Repetition in the Two-Player Quantum Cloning Game." pith.science (2026). https://pith.science/paper/MKS3YAVB

@misc{pith2026260800930,
  author       = {Pith},
  title        = {Pith review of: Parallel Repetition in the Two-Player Quantum Cloning Game},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MKS3YAVB}},
  note         = {Machine review of arXiv:2608.00930}
}

abstract

We study parallel repetition in the two-player quantum cloning game, a monogamy-of-entanglement game motivated by quantum position verification. Colisson Palais, Escol\`a-Farr\`as, and Speelman bounded the value of $n$ copies between $(3/4)^n$ and $\cos^{2n}(\pi/8)$. For two copies, we prove that neither bound is tight. An explicit challenge-dependent strategy achieves value $(5+\sqrt{17})/16>9/16$, so strong parallel repetition fails for the unrestricted game. A block Gram matrix argument gives the upper bound $(11+\sqrt{65})/32<\cos^4(\pi/8)$ and strictly improves the previous parallel-repetition upper bound for every $n$. For every $n$, challenge-independent strategies have optimal value $(3/4)^n$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

8 extracted references · 1 canonical work pages

[1]

Colisson Palais, L

L. Colisson Palais, L. Escolà-Farràs, and F. Speelman,A quantum cloning game with applications to quantum position verification, inTQC 2025, LIPIcs 350, 2:1–2:17 (2025), doi:10.4230/LIPIcs.TQC.2025.2

work page doi:10.4230/lipics.tqc.2025.2 2025
[2]

Chandran, V

N. Chandran, V. Goyal, R. Moriarty, and R. Ostrovsky,Position based cryptography, inCRYPTO 2009, LNCS 5677, 391–407 (2009), doi:10.1007/978-3-642-03356-8_23

doi:10.1007/978-3-642-03356-8_23 2009
[3]

Buhrman, N

H. Buhrman, N. Chandran, S. Fehr, R. Gelles, V. Goyal, R. Ostrovsky, and C. Schaffner,Position- based quantum cryptography: impossibility and constructions, SIAM J. Comput. 43(1), 150–178 (2014), doi:10.1137/130913687

doi:10.1137/130913687 2014
[4]

Tomamichel, S

M. Tomamichel, S. Fehr, J. Kaniewski, and S. Wehner,A monogamy-of-entanglement game with applica- tions to device-independent quantum cryptography, New J. Phys. 15, 103002 (2013), doi:10.1088/1367- 2630/15/10/103002

doi:10.1088/1367- 2013
[5]

Watrous,The Theory of Quantum Information, Cambridge University Press (2018), doi:10.1017/9781316848142

J. Watrous,The Theory of Quantum Information, Cambridge University Press (2018), doi:10.1017/9781316848142

doi:10.1017/9781316848142 2018
[6]

P. R. Halmos,Two subspaces, Trans. Amer. Math. Soc. 144, 381–389 (1969), doi:10.1090/S0002-9947- 1969-0251519-5

doi:10.1090/s0002-9947- 1969
[7]

Beigi and R

S. Beigi and R. König,Simplified instantaneous non-local quantum computation with applications to position-based cryptography, New J. Phys. 13, 093036 (2011), doi:10.1088/1367-2630/13/9/093036

doi:10.1088/1367-2630/13/9/093036 2011
[8]

Escolà-Farràs and F

L. Escolà-Farràs and F. Speelman,Quantum position verification in one shot: parallel repetition of the f-BB84 andf-routing protocols, arXiv:2503.09544 (2025), doi:10.48550/arXiv.2503.09544. 11

arXiv doi:10.48550/arxiv.2503.09544 2025

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Colisson Palais, L

L. Colisson Palais, L. Escolà-Farràs, and F. Speelman,A quantum cloning game with applications to quantum position verification, inTQC 2025, LIPIcs 350, 2:1–2:17 (2025), doi:10.4230/LIPIcs.TQC.2025.2

work page doi:10.4230/lipics.tqc.2025.2 2025

[2] [2]

Chandran, V

N. Chandran, V. Goyal, R. Moriarty, and R. Ostrovsky,Position based cryptography, inCRYPTO 2009, LNCS 5677, 391–407 (2009), doi:10.1007/978-3-642-03356-8_23

doi:10.1007/978-3-642-03356-8_23 2009

[3] [3]

Buhrman, N

H. Buhrman, N. Chandran, S. Fehr, R. Gelles, V. Goyal, R. Ostrovsky, and C. Schaffner,Position- based quantum cryptography: impossibility and constructions, SIAM J. Comput. 43(1), 150–178 (2014), doi:10.1137/130913687

doi:10.1137/130913687 2014

[4] [4]

Tomamichel, S

M. Tomamichel, S. Fehr, J. Kaniewski, and S. Wehner,A monogamy-of-entanglement game with applica- tions to device-independent quantum cryptography, New J. Phys. 15, 103002 (2013), doi:10.1088/1367- 2630/15/10/103002

doi:10.1088/1367- 2013

[5] [5]

Watrous,The Theory of Quantum Information, Cambridge University Press (2018), doi:10.1017/9781316848142

J. Watrous,The Theory of Quantum Information, Cambridge University Press (2018), doi:10.1017/9781316848142

doi:10.1017/9781316848142 2018

[6] [6]

P. R. Halmos,Two subspaces, Trans. Amer. Math. Soc. 144, 381–389 (1969), doi:10.1090/S0002-9947- 1969-0251519-5

doi:10.1090/s0002-9947- 1969

[7] [7]

Beigi and R

S. Beigi and R. König,Simplified instantaneous non-local quantum computation with applications to position-based cryptography, New J. Phys. 13, 093036 (2011), doi:10.1088/1367-2630/13/9/093036

doi:10.1088/1367-2630/13/9/093036 2011

[8] [8]

Escolà-Farràs and F

L. Escolà-Farràs and F. Speelman,Quantum position verification in one shot: parallel repetition of the f-BB84 andf-routing protocols, arXiv:2503.09544 (2025), doi:10.48550/arXiv.2503.09544. 11

arXiv doi:10.48550/arxiv.2503.09544 2025