Pith. sign in

REVIEW 2 major objections 5 minor 32 references

Tighter Asymptotic Key Rates for Intensity-Correlated Decoy-State QKD via Nonlinear Programming

T0 review · 2 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Solving the full nonlinear decoy-state parameter-estimation problems before linearising yields tighter key-rate bounds than canonical reference points, and coincidence of the two stages certifies optimality.

desk verdict A solid, reproducible improvement for correlated-source decoy-state QKD; the relaxation worry is answerable, and the main fix is presentational. read the letter →

arxiv 2602.04966 v2 pith:3GC2CWY7 submitted 2026-02-04 quant-ph

Matej Pivoluska , Mateus Araújo This is my paper
classification quant-ph PACS 03.67.Dd
keywords decoy-stateQKDintensitycorrelationsCauchy-Schwarzconstraintsnonlinearprogrammingouterlinearisationasymptotickeyrateinterior-pointsolverphase-randomizedweakcoherentpulses
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 attacks a realistic weakness in decoy-state QKD: real sources drift in intensity from round to round, and the correlation can leak intensity information, so the standard decoy-state analysis—which assumes intensity-independent yields—is no longer valid. Earlier work added Cauchy-Schwarz constraints that tie n-photon yields across intensities, but these are nonlinear and were linearised around reference points chosen from a channel model. The paper's proposal is to first solve the fully nonlinear constrained parameter-estimation problems with a general-purpose interior-point solver, then use that candidate solution as the linearisation point for the outer linear program that certifies a valid asymptotic key-rate lower bound. In numerical simulations for both coarse-grained model-independent and fine-grained truncated-Gaussian correlations, this two-stage workflow consistently beats the canonical reference points, and when the two stages return the same value it certifies that the bound is optimal.

What carries the argument

The load-bearing object is the two-stage optimisation workflow. Stage one solves the full Cauchy-Schwarz-constrained nonlinear programs (P1)-(P3) with a smooth interior-point solver; stage two replaces each CS inequality by its first-order Taylor expansion around that solution and solves the resulting linear program. The CS constraints themselves are the mechanism that couple n-photon yields across different intensity settings: for each pair of intensities a,b and photon number n they bound Y_{n,b} (and H_{n,b}) in terms of Y_{n,a} through square-root functions G±(y,τ), where τ encodes the maximum intensity-information leakage over the correlation range. The work that these constraints do is

What would settle it

Find a tuple of yields that satisfies the nonlinear Cauchy-Schwarz inequalities and the linear observation constraints but violates one of the tangent-line inequalities at the reference point produced by the nonlinear solve; that would show the outer LP is cutting into the feasible set and the claimed lower bound is not guaranteed. A cleaner version: solve the same instances with a certified exact method (e.g., a mixed-integer second-order cone formulation) and check whether its optimum matches the two-stage value; a mismatch would mean the certificate is not reliable.

Watch

Extended reading notes

Core claim

The central claim is that the quality of an outer-linearisation bound depends on where you linearise, and the best linearisation point is the solution of the actual nonlinear problem. The paper formulates the three decoy-state parameter-estimation tasks—lower bounds on the Z- and X-basis single-photon yields and an upper bound on the single-photon error rate—as nonlinear programs whose only nonlinearities are the Cauchy-Schwarz square-root constraints. Solving these with an interior-point nonlinear solver and then linearising the CS inequalities around that solution gives valid asymptotic key-rate lower bounds. The simulations show these bounds dominate the canonical channel-model-based refe

Load-bearing premise

The outer linearisation must be a valid outer relaxation—the tangent lines must never cut into the set of yield values allowed by the Cauchy-Schwarz constraints—but the paper never proves this and is inconsistent about whether those constraints are convex, so the final lower bound would not be guaranteed if they are not.

Editorial extensions

If this is right

  • The two-stage workflow can be dropped into existing CS-based decoy-state security analyses as a replacement for canonical linearisation points, improving key rates without changing the security proof.
  • When the nonlinear value and the outer-LP value coincide, the paper's construction certifies that the computed key-rate lower bound is optimal for that parameter-estimation problem.
  • For fine-grained truncated-Gaussian correlation models inspired by real hardware, the gains over canonical points are larger exactly when measured statistics deviate from the simple channel model.
  • The approach removes the need for iterative re-linearisation heuristics, making the computation faster and reproducible.
  • The improvement holds for both the coarse-grained model-independent formulation and the fine-grained model, across the tested distances and correlation ranges.

Reading between the lines

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

  • The certificate logic only works when the nonlinear and linearised optima coincide; when they differ, the nonlinear value is merely a candidate and the outer bound, while valid, may be loose, so a user must check for agreement rather than assume it.
  • The same two-stage idea—solve the true nonconvex problem, then linearise around its solution—should transfer to other QKD parameter-estimation settings where the feasible set admits convex outer approximations, such as source flaws beyond intensity drift or measurement-device imperfections.
  • A natural testable extension is to apply the workflow to measured correlation data from real hardware rather than simulated data, to see whether the tighter bounds survive in practice; the paper only simulates the Gaussian model based on one hardware system.
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.

Referee Report

2 major / 5 minor

Summary. The paper addresses decoy-state BB84 with phase-randomized weak coherent pulses whose intensities are correlated across rounds. It reviews the Cauchy--Schwarz (CS) constraints of Zapatero et al. [10] that couple n-photon yields across intensity settings, and proposes a two-stage numerical workflow: (i) solve the full nonlinear CS-constrained estimation problems P1--P3 with the interior-point solver IPOPT, and (ii) linearize the CS constraints around the IPOPT solution and solve the resulting LP to obtain a certified lower bound on the asymptotic key rate. The paper reports simulations for a coarse-grained model-independent formulation and for a fine-grained truncated-Gaussian model, showing tighter rates than the canonical channel-model-based linearization points, and argues that when the two optimisation stages coincide the rate is certified optimal. A public code repository is provided for reproducibility.

Significance. The proposed two-stage workflow is a simple, practical improvement over canonical linearization: it preserves the security guarantee of an outer LP relaxation while potentially yielding tighter bounds. Public code availability is a genuine strength, and the optimality-certificate idea is logically sound provided the outer linearization is a true relaxation and the IPOPT point is feasible. The paper's central claim is defensible, but the manuscript leaves the relaxation property unproved and contains an inconsistent and incorrect statement that the CS constraints are non-convex; the relevant feasible set is in fact convex. With fixes to these points, this is a useful, though methodologically incremental, contribution to decoy-state QKD with correlated sources.

major comments (2)
  1. [Sec. II.I, Eq. (27)] The assertion that the outer linearisation 'always leads to valid key rates' is load-bearing but never proved. Validity requires that tangent lines to G- underestimate G- and tangent lines to G+ overestimate G+, i.e. that G- is convex and G+ concave. This is true (for fixed tau, g-'' = +sqrt(tau(1-tau))/(2[y(1-y)]^{3/2}) > 0 and g+'' = -... < 0, and the branch definitions in Eq. (21) glue smoothly), but the paper should state and prove this explicitly. The current text is inconsistent: the Introduction calls the CS constraints 'convex functions', while Sec. III.A calls them 'non-convex square-root nonlinearities'. Without a correct and explicit proof of the relaxation property, the 'certified lower bound' claim is unsupported.
  2. [Sec. III, optimality certificate] The claim that coincidence of the two optimisation stages certifies optimality depends on the IPOPT solution being feasible for the original nonlinear problem. If the IPOPT point is only approximately feasible, its objective value is not a valid upper bound, and the sandwich argument (relaxed LP value <= true optimum <= IPOPT value) fails. The paper does not report IPOPT feasibility tolerances or maximum constraint violations. Please report the IPOPT tolerances used and, for the displayed curves, verify and state the maximum constraint violation of the reference points, e.g. max violation < 1e-8.
minor comments (5)
  1. [Figs. 1 and 2] The captions and text do not make clear whether the curves labeled 'IPOPT' are the raw IPOPT objective values or the final certified outer-LP values after linearizing at the IPOPT point. Since the optimality certificate rests on comparing these two values, the figures should show both (or state that they are numerically indistinguishable and provide the max difference).
  2. [Sec. II.H title] 'Cauchy-Schwartz' should be 'Cauchy-Schwarz'.
  3. [Eq. (29)] The derivative g'+- has a singularity at y=0 and y=1. The text should state that reference points are taken in (0,1) or describe how boundary cases are handled, especially because IPOPT could return values at the bounds.
  4. [Eq. (44)] The notation 'min ... (= Z_{1,mu})' inside the optimization is confusing. It should be written as an objective equal to Z_{1,mu} with the equality displayed after the minimization, not inside the min.
  5. [References] Reference [3] is given as 'AIAA Journal 61, 1895 (2023)'; this journal title appears incorrect for a quantum-key-distribution paper. Please verify the source.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the IPOPT-assisted outer-linearization workflow is a genuine relaxation-and-certificate scheme, not a fit or self-citation chain.

full rationale

The claimed derivation chain is: keep the nonlinear CS constraints (Eqs. 20-23) from [10]; solve the resulting NLPs (P1)-(P3) with IPOPT; use the returned point only as the reference point for the first-order outer linearization (Eq. 27); solve the resulting LP; report the LP value as a certified lower bound. The IPOPT point is not a parameter fitted to make the key-rate prediction come out; it is a candidate feasible point for the same estimation problem that defines the bounds feeding Eq. (6). The final LP bound is not forced to equal the IPOPT objective by construction - the LP can return a strictly smaller value, and optimality is claimed only when the two values coincide, which is the standard relaxation-gap certificate (provided Eq. (27) is a true outer approximation). The outer-relaxation property is not proven in the paper and Sec. III.A even calls the constraints "non-convex square-root nonlinearities", but this is a correctness/clarity risk, not a circular reduction: the required convexity/concavity of G-/G+ can be checked directly from Eq. (22), independent of the paper's conclusions. The only self-citation is [7] in a general background list; it is not load-bearing. No step defines a predicted quantity in terms of an input, no fitted parameter is renamed a prediction, and no uniqueness or ansatz is imported from the authors' own prior work.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

All premises are standard assumptions in decoy-state QKD security analysis or mathematical properties of the CS constraints taken from [10,16]. No free parameters are fitted to produce the claimed improvement; simulation parameters (µ,ν,ω,pµ,η,pd,δA,fEC,ξ,δmax,Ncut) are fixed inputs from the channel model. No new physical entities are introduced.

assumptions (7)
  • domain assumption Poisson photon-number statistics conditional on actual intensity (Eq. 1)
    Assumption 1 in Sec. II.B; required for all subsequent bounds.
  • domain assumption Bounded relative deviation |1 - α_k/a_k| ≤ δ_max (Eq. 2)
    Assumption 2; used to derive record-independent bounds on photon-number probabilities.
  • domain assumption Finite correlation range ξ (Assumption 3)
    Limits the memory of intensity correlations; needed for finite-dimensional formulations.
  • domain assumption Perfect phase randomization, no state-preparation flaws beyond intensity drift, no side-channels (Sec. II.A)
    Standard idealizations in this security analysis context.
  • domain assumption Basis-independent detection efficiency (M^{Z,f}=M^{X,f}, Sec. II.A)
    Required for the key-rate formula and decoy-state bounds to hold in the stated form.
  • standard math Outer linearization of CS constraints is a valid relaxation (Sec. II.I, Eq. 27): G- convex, G+ concave
    The tangent lines must lie outside the feasible set; the paper relies on this property for the validity of the final bound but does not prove it, citing [10].
  • domain assumption Key-rate formula Eq. (6) is valid
    Taken from [10]; the optimization problems supply bounds that plug into this formula.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tighter Asymptotic Key Rates for Intensity-Correlated Decoy-State QKD via Nonlinear Programming." pith.science (2026). https://pith.science/paper/3GC2CWY7

@misc{pith2026260204966,
  author       = {Pith},
  title        = {Pith review of: Tighter Asymptotic Key Rates for Intensity-Correlated Decoy-State QKD via Nonlinear Programming},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3GC2CWY7}},
  note         = {Machine review of arXiv:2602.04966}
}
abstract

Decoy-state QKD with phase-randomized weak coherent pulses is typically analyzed assuming independent, precisely prepared intensities. Real sources, however, can exhibit correlated intensity drift across rounds, potentially leaking intensity information and breaking the standard decoy-state reduction to linear programs. Cauchy--Schwarz (CS) constraints can restore security by coupling $n$-photon yields across intensities, but they introduce nonlinear square-root constraints that are commonly handled via outer linearisation around channel-model-based reference points. We propose a reproducible alternative: first solve the full CS-constrained parameter-estimation problems using the interior-point nonlinear solver IPOPT, then use the resulting candidate solution as the linearisation point for the outer optimisation that certifies a valid lower bound on the asymptotic key rate. Simulations for both coarse-grained model-independent correlations and fine-grained truncated-Gaussian models show consistently tighter key-rate bounds than canonical reference points, and in some cases allow certifying optimality when both optimisation stages coincide.

Figures

Figures reproduced from arXiv: 2602.04966 by the authors.

Figure 1
Figure 1. FIG. 1. Coarse grained key rate calculated using formulation described in Section [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Here we compare lower bounds on key rates for System B from [ [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

32 extracted references · 9 linked inside Pith

  1. [10]

    Zapatero, ´A

    V. Zapatero, ´A. Navarrete, K. Tamaki, and M. Curty, Security of quantum key distribution with intensity correlations, Quantum5, 602 (2021), arXiv:2105.11165 [quant-ph]

    arXiv 2021

  2. [28]

    Trefilov, X

    D. Trefilov, X. Sixto, V. Zapatero, A. Huang, M. Curty, and V. Makarov, Intensity correlations in decoy-state bb84 quantum key distribution systems, Optica Quantum3, 417 (2025)

    2025

  3. [1]

    G. S. Vernam, Cipher printing telegraph systems: For secret wire and radio telegraphic communications, Journal of the A.I.E.E.45, 109 (1926)

    1926

  4. [2]

    M. N. Wegman and J. L. Carter, New hash functions and their use in authentication and set equality, Journal of Computer and System Sciences22, 265 (1981)

    1981

  5. [3]

    Renner and R

    R. Renner and R. Wolf, Quantum advantage in cryptography, AIAA Journal61, 1895 (2023)

    2023

  6. [4]

    Wang, X.-L

    X.-B. Wang, X.-L. Hu, and Z.-W. Yu, Practical Long-Distance Side-Channel-Free Quantum Key Distribution, Physical Review Applied12, 054034 (2019)

    2019

  7. [5]

    Nahar, T

    S. Nahar, T. Upadhyaya, and N. L¨ utkenhaus, Imperfect phase randomization and generalized decoy-state quantum key distribution, Physical Review Applied20, 064031 (2023)

    2023

  8. [6]

    Curr´ as-Lorenzo, M

    G. Curr´ as-Lorenzo, M. Pereira, G. Kato, M. Curty, and K. Tamaki, A security framework for quantum key distribution implementations (2023), arXiv:2305.05930 [quant-ph]

    arXiv 2023

Show all 32 references
  1. [7]

    Pereira, G

    M. Pereira, G. Curr´ as-Lorenzo, and M. Ara´ ujo, Optimal key rates for quantum key distribution with partial source characterization (2025), arXiv:2510.13085 [quant-ph]

    2025

  2. [8]

    Nagamatsu, A

    Y. Nagamatsu, A. Mizutani, R. Ikuta, T. Yamamoto, N. Imoto, and K. Tamaki, Security of quantum key distribution with light sources that are not independently and identically distributed, Physical Review A93, 042325 (2016)

    2016

  3. [9]

    Pereira, G

    M. Pereira, G. Kato, A. Mizutani, M. Curty, and K. Tamaki, Quantum key distribution with correlated sources, Science Advances6, eaaz4487 (2020), arXiv:1908.08261 [quant-ph]

    2020 arXiv

  4. [11]

    Curr´ as-Lorenzo, S

    G. Curr´ as-Lorenzo, S. Nahar, N. L¨ utkenhaus, K. Tamaki, and M. Curty, Security of quantum key distribution with imperfect phase randomisation, Quantum Science and Technology9, 015025 (2023)

    2023

  5. [12]

    Pereira, G

    M. Pereira, G. Curr´ as-Lorenzo, A. Mizutani, D. Rusca, M. Curty, and K. Tamaki, Quantum key distribution with un- bounded pulse correlations, Quantum Science and Technology10, 015001 (2024)

    2024

  6. [13]

    Marwah and F

    A. Marwah and F. Dupuis, Proving security of BB84 under source correlations (2024), arXiv:2402.12346 [quant-ph]

    2024 arXiv

  7. [14]

    Li, Y.-G

    J.-X. Li, Y.-G. Shan, R. Wang, F.-Y. Lu, Z.-Q. Yin, S. Wang, W. Chen, D.-Y. He, G.-C. Guo, and Z.-F. Han, Secure quantum key distribution against correlated leakage source (2025), arXiv:2507.11251 [quant-ph]

    2025 arXiv

  8. [15]

    Mizutani, G

    A. Mizutani, G. Kato, K. Azuma, M. Curty, R. Ikuta, T. Yamamoto, N. Imoto, H.-K. Lo, and K. Tamaki, Quantum key distribution with setting-choice-independently correlated light sources, npj Quantum Information5, 1 (2019)

    2019

  9. [16]

    Sixto, V

    X. Sixto, V. Zapatero, and M. Curty, Security of Decoy-State Quantum Key Distribution with Correlated Intensity Fluc- tuations, Physical Review Applied18, 044069 (2022)

    2022

  10. [17]

    Li, F.-Y

    J.-X. Li, F.-Y. Lu, Z.-H. Wang, V. Zapatero, M. Curty, S. Wang, Z.-Q. Yin, W. Chen, D.-Y. He, G.-C. Guo, and Z.-F. Han, Quantum key distribution overcoming practical correlated intensity fluctuations (2025), arXiv:2501.13482 [quant-ph]

    2025 arXiv

  11. [18]

    Burenkov, B

    V. Burenkov, B. Qi, B. Fortescue, and H.-K. Lo, Security of high speed quantum key distribution with finite detector dead time (2010), arXiv:1005.0272 [quant-ph]

    2010 arXiv

  12. [19]

    Kamin and N

    L. Kamin and N. L¨ utkenhaus, Improved Decoy-state and Flag-state Squashing Methods (2024), arXiv:2405.05069 [quant- ph]

    2024 arXiv

  13. [20]

    Tupkary, S

    D. Tupkary, S. Nahar, P. Sinha, and N. L¨ utkenhaus, Phase error rate estimation with basis-efficiency mismatch for decoy- state BB84 (2024), arXiv:2408.17349 [quant-ph]

    2024

  14. [21]

    Marcomini, A

    A. Marcomini, A. Mizutani, F. Gr¨ unenfelder, M. Curty, and K. Tamaki, Loss-tolerant quantum key distribution with detection efficiency mismatch, Quantum Science and Technology10, 035002 (2025)

    2025

  15. [22]

    Grasselli, G

    F. Grasselli, G. Chesi, N. Walk, H. Kampermann, A. Widomski, M. Ogrodnik, M. Karpi´ nski, C. Macchiavello, D. Bruß, and N. Wyderka, Quantum key distribution with basis-dependent detection probability, Physical Review Applied23, 044011 (2025)

    2025

  16. [23]

    Nahar, D

    S. Nahar, D. Tupkary, and N. L¨ utkenhaus, Imperfect detectors for adversarial tasks with applications to quantum key distribution (2025), arXiv:2503.06328 [quant-ph]

    2025

  17. [24]

    Z. Wang, D. Tupkary, and S. Nahar, Phase error estimation for passive detection setups with imperfections and memory effects (2025), arXiv:2508.21486 [quant-ph]

    2025

  18. [25]

    H.-K. Lo, X. Ma, and K. Chen, Decoy State Quantum Key Distribution, Physical Review Letters94, 230504 (2005)

    2005

  19. [26]

    X. Ma, B. Qi, Y. Zhao, and H.-K. Lo, Practical decoy state for quantum key distribution, Phys. Rev. A72, 012326 (2005)

    2005

  20. [27]

    Rice and J

    P. Rice and J. Harrington, Numerical analysis of decoy state quantum key distribution protocols (2009), arXiv:0901.0013 15 [quant-ph]

    2009 arXiv

  21. [29]

    W¨ achter and L

    A. W¨ achter and L. T. Biegler, On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming, Mathematical Programming106, 25 (2006)

    2006

  22. [30]

    Yoshino, M

    K.-i. Yoshino, M. Fujiwara, K. Nakata, T. Sumiya, T. Sasaki, M. Takeoka, M. Sasaki, A. Tajima, M. Koashi, and A. Tomita, Quantum key distribution with an efficient countermeasure against correlated intensity fluctuations in optical pulses, npj Quantum Information4, 1 (2018)

    2018

  23. [31]

    Gurobi Optimization, LLC, Gurobi Optimizer Reference Manual (2026)

    2026

  24. [32]

    C. Coey, M. Lubin, and J. P. Vielma, Outer approximation with conic certificates for mixed-integer convex problems, Mathematical Programming Computation12, 249 (2020), arXiv:1808.05290 [math.OC]

    2020 arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.