Pith. sign in

REVIEW 4 major objections 3 minor 37 references

Reliability-Dependent Scaling Laws of Deterministic Identification over Binary Symmetric Channels

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

Pith's one-line read For deterministic identification over a binary symmetric channel, the optimal rate approaches capacity 1 whenever the error probabilities vanish subexponentially; only exponentially fast vanishing leaves a permanent rate penalty.

desk verdict The main theorem is not proven: the converse contains a non sequitur and the constants are not δ-only, though the Hamming-shell decomposition is a nice idea. read the letter →

arxiv 2608.03282 v1 pith:JILG2IPE submitted 2026-08-04 cs.IT math.ITmath.PR

classification cs.ITmath.ITmath.PR MSC 94A1594B65
keywords deterministicidentificationbinarysymmetricchannelsecond-orderasymptoticsrate–reliabilitytradeoffminimumdistanceboundHammingshellslargedeviationstotalvariation
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

Deterministic identification asks whether a receiver can decide, not decode, whether a particular message was sent; over a binary symmetric channel this paper shows the maximal number of identifiable messages obeys a sharp rate law set by how fast the smaller of the two error probabilities vanishes. The central claim, Theorem 1, is that when that error decays like $e^{-n^\alpha}$ with $0\le\alpha\le 1$, the optimal rate lies between $1-h(C_1 n^{-(1-\alpha)/2})$ and $1-h(C_2 n^{\alpha-1})$, with constants depending only on the crossover probability. So for any $\alpha<1$ the rate climbs to the identification capacity 1, with a backoff controlled by $\alpha$, while exponential decay ($\alpha=1$) leaves a fixed entropy penalty. The proof derives the lower bound from Hamming-shell concentration of channel noise and the Gilbert–Varshamov packing bound, and the upper bound from a total-variation separation argument that forces large minimum Hamming distance, converted to a rate loss by the Hamming bound.

What carries the argument

The engine is the minimum error parameter $\eta_n = \min\{\lambda_{1,n},\lambda_{2,n}\}$, which compresses the two error constraints into one reliability exponent. On the achievability side, the paper uses Hamming shells — decoding regions $D_u(r)$ collecting outputs whose noise weight is within $r$ of $n\delta$ — and chooses $r$ and the code's minimum distance $d_m$ on the same scale $n^{(1+\alpha)/2}$ in the moderate-deviation regime, so that both type-I and type-II errors fall below $\eta_n$ by concentration estimates (large-deviation, moderate-deviation, or CLT depending on $\alpha$); the Gilbert–Varshamov bound then converts the normalized distance into the entropy rate $1-h(d_m/n)$. On

What would settle it

Take $\lambda_1=1/2$ (constant) and $\lambda_2=e^{-n}$. The total-variation separation bound of Proposition 2 forces only constant Hamming distance $t \ge \ln 4 / c_0$, whereas Proposition 3 would claim $t \ge n/c_0$. Performing this calculation on a BSC($\delta$) shows the converse scaling $d_C \ge -\ln\eta_n/c_0$ does not follow from the stated lemmas when the two error probabilities decay at very different rates.

Watch

Extended reading notes

Core claim

Theorem 1 states that for $\mathrm{BSC}(\delta)$ with $0<\delta<1/2$ and reliability parameter $\eta_n = \min\{\lambda_{1,n},\lambda_{2,n}\}$ satisfying $-\ln\eta_n \asymp n^\alpha$, $\alpha\in[0,1]$, the optimal DID rate $R_m(n)$ satisfies $1-h(C_1 n^{-(1-\alpha)/2}) \le R_m(n) \le 1-h(C_2 n^{\alpha-1})$. The upper and lower bounds are both entropy backoffs from rate 1; for $\alpha<1$ they vanish, establishing that reliability below exponential decay does not change the first-order capacity, only the second-order convergence speed. At $\alpha=1$ the two bounds do not meet, but both show a nonvanishing penalty of the form $h(\text{constant})$, so identifier message length still grows linearl

Load-bearing premise

The converse only forces a large minimum distance if the type-I and type-II error probabilities decay at the same rate, so their minimum $\eta_n$ truly controls the sum that appears in the total-variation bound; without that comparability, the claimed minimum-distance constraint can fail.

Editorial extensions

If this is right

  • For any $\alpha<1$, the optimal DID rate over the BSC converges to 1, so deterministic identification with vanishing errors still identifies exponentially many messages ($\log N \sim n$) at first-order capacity.
  • The convergence is slower for stronger reliability: the achievability backoff scales as $h(C n^{-(1-\alpha)/2})$, interpolating between $n^{-1/2}$ at bounded errors ($\alpha=0$) and a constant at exponential errors ($\alpha=1$).
  • At $\alpha=1$, the rate is bounded away from capacity by a fixed entropy term, but message length remains linear in $n$ — reliability changes the rate value, not the linear scaling, in contrast to continuous-alphabet channels.
  • In the bounded-error case $\alpha=0$, the second-order gap is of order $1/\sqrt{n}$, matching the usual central-limit scaling.
  • The constants $C_1$ and $C_2$ depend only on the channel crossover probability $\delta$, not on the particular error sequences, so the scaling law is universal across error profiles with the same exponent $\alpha$.

Reading between the lines

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

  • As an extension the authors do not state: a sharper converse in the mismatched-decay case would need to replace $\eta_n$ by the sum $\lambda_{1,n}+\lambda_{2,n}$ in the distance constraint, which would weaken the bound exactly when the two errors are not comparable.
  • The same Hamming-shell construction should generalize to any symmetric discrete memoryless channel with a concentration property, replacing the binary entropy $h$ with the channel's output entropy; this is a testable conjecture rather than a claim of the paper.
  • The gap between $C_1$ and $C_2$ in Theorem 1 suggests the exact second-order constant may be determined by matching the moderate-deviation exponent of Bernoulli tails; a numerical computation of the true $R_m(n)$ for moderate $n$ could indicate whether either constant is loose.
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

4 major / 3 minor

Summary. The paper studies rate–reliability tradeoffs for deterministic identification (DID) over binary symmetric channels (BSCs) when the type-I and type-II error probabilities decay with blocklength. It introduces the parameter η_n = min{λ1,n, λ2,n} and claims that if -ln η_n ≍ n^α, α∈[0,1], then the optimal DID rate satisfies 1-h(C1 n^{-(1-α)/2}) ≤ R_m(n) ≤ 1-h(C2 n^{α-1}) with constants C1,C2 depending only on the channel crossover δ. The achievability construction uses Hamming-shell concentration and large/moderate-deviation estimates; the converse uses a total-variation distinguishability bound and a minimum-distance argument combined with the Hamming bound.

Significance. If the claimed theorem were correct, it would give a clean BSC specialization of the general rate–reliability framework for DID, showing that the DID rate approaches capacity for α<1 and suffers a nonvanishing penalty for α=1. The paper usefully identifies Hamming-shell concentration as the relevant geometric mechanism, and the overall approach is natural. However, the main theorem is not established: the converse contains a logical gap in the passage from a sum constraint to a minimum-distance constraint, and the proof repeatedly defines constants that depend on the error-decay prefactor, contradicting the theorem statement. These are load-bearing issues, not presentation problems.

major comments (4)
  1. [§III, Proof of converse, Proposition 3] The step 'λ1+λ2 ≥ 2e^{-c0t}. Since λ1+λ2 ≥ 2λ, it suffices to require λ ≥ e^{-c0t}' is a non sequitur. A lower bound on a sum gives no lower bound on its minimum; it gives a lower bound on the maximum. For example, λ1=e^{-2c0t}, λ2=1/2 satisfies the sum inequality but violates λ ≥ e^{-c0t}. Consequently Proposition 3, d_C ≥ -ln η_n / c0(δ), is unsupported for the class of error sequences allowed by Theorem 1, and the upper bound R_m(n) ≤ 1-h(C2 n^{α-1}) rests on this invalid step. An additional comparability assumption such as λ1,n ≍ λ2,n would be needed, but it is neither stated nor proved.
  2. [§III, Theorem 1 statement vs. proof] The theorem claims constants C1,C2 depend only on δ, but the proof introduces C1 depending on the prefactor of -ln η_n. In Case 1, C1 = sqrt((2c'_l c_l+1)L)/(c'_l(1-2δ)); in Case 2, C1 = (2+√(2m))√(δ(1-δ))/(1-2δ); in Case 3, C1 = (2Q^{-1}(l/2)+Q^{-1}(l/4))/(1-2δ). Moreover, the converse silently replaces -ln η_n/(2c0 n) by n^{α-1}/(2c0), thereby ignoring the multiplicative constants hidden in -ln η_n ≍ n^α. For -ln η_n = L n with L arbitrarily large, the required distance grows as Ln/c0, so no δ-only C2 can yield the stated upper bound. As stated, the theorem is false; at minimum the constants must depend on the prefactor or on the range of -ln η_n/n^α.
  3. [§III, Case 2 (moderate-deviation regime)] The proof chooses a=√(δ(1-δ)) and b_n=n^{(1+α)/2}, which gives c_M=1/2 and hence PI ≤ exp(-n^α/2). This is claimed to be ≤ η_n. When -ln η_n = m n^α with m>1/2, the inequality exp(-n^α/2) ≤ exp(-m n^α) fails for large n. To meet the target, a must be taken proportional to √(2m), which makes the resulting constant depend on m. The displayed C1 in this subsection indeed depends on m, confirming that the universal-constant claim cannot be derived from the given argument.
  4. [§III, Case 3 (bounded-error regime)] The proof assumes 'η_n = l for some constant l > 0' without loss of generality. But the theorem's assumption -ln η_n ≍ 1 allows η_n to oscillate between positive constants, not necessarily converge. The constant C1 = (2Q^{-1}(l/2)+Q^{-1}(l/4))/(1-2δ) is then not well-defined for the sequence, and no δ-only constant is justified. This is a further instance of the mismatch between the theorem statement and the proof.
minor comments (3)
  1. [§III, Lemma 5] The proof writes (t choose t/2) for an arbitrary Hamming weight t. If t is odd, t/2 is not an integer; the argument should use floor or ceiling. The exponential lower bound is still plausibly repairable, but the current notation is not correct for all t.
  2. [§III, Case 2] The condition 'if t ≤ 2α/(1+α)a' is unclear as typeset; the intended inequality is likely t ≤ 2α/((1+α)a). Please clarify the role of a and the derivation of the threshold.
  3. [§III, proof organization] The proof of achievability says 'We first consider the regime -ln η_n = Ln' and treats the three cases as if -ln η_n is exactly a monomial. Under the theorem's assumption -ln η_n ≍ n^α, the quantity may oscillate within a constant factor; the proof should explicitly state whether it handles all such sequences or only those with a limit.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular reasoning detected; the mathematical gap in the converse is a correctness flaw, not a self-referential derivation.

full rationale

The paper's derivation chain is self-contained against external results. The achievability proofs in Section III (Cases 1–3) choose explicit Hamming-shell radii r and minimum distances d_m from η_n via the large-deviation bound (Lemma 2, Durrett), the moderate-deviation bound (Lemma 3, Eichelsbacher–Löwe), and the Berry–Esseen-type CLT bound (Lemma 4, Chen–Goldstein–Shao); these are standard external concentration results, not fitted to the target rate. The rate lower bound then follows from the GV lower bound in Lemma 1. The converse uses Lemma 5 (a direct binomial/Stirling estimate) and Proposition 2, cited as [22, Theorem III.2] by Colomer, Deppe, Boche, and Winter, with no overlap with the present authors, so the load-bearing TV separation result is genuinely external. The Hamming upper bound in Lemma 1 completes the converse. No parameter is fitted and then renamed a prediction, and no conclusion is imported through a self-citation. The manuscript does contain a serious non-circular flaw: after deriving λ1+λ2 ≥ 2e^{-c0t}, the proof says 'Since λ:=min{λ1,λ2} satisfies λ1+λ2 ≥ 2λ, it suffices to require λ≥e^{-c0t},' which is a non sequitur; Proposition 3's d_C ≥ -ln η_n/c0(δ) is therefore not established for arbitrary λ1,n,λ2,n. This is a correctness risk, not circularity, and does not raise the circularity score.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The core ingredients are standard concentration and coding bounds, but the paper adds two hidden assumptions: comparability of the two error decays, and universality of the moderate-deviation constant. The unspecified constants c_l, c'_l, c0 are the real slack in the bounds.

free parameters (4)
  • c_l = not quantified
    Shell-radius proportionality constant in Case 1; stated to depend only on δ but no value is given, and C1 depends on it.
  • c'_l = not quantified
    Lower-bound slope for c'_L(x); C1 depends on it, value never specified.
  • c0(δ) = not quantified
    TV decay constant in Lemma 5 and the converse; proof's supporting equality is false and no construction is supplied.
  • L, m, l (error-exponent preconstants) = unspecified in theorem
    The assumption -ln η_n ≍ n^α carries arbitrary preconstants; the proof's C1/C2 depend on them although Theorem 1 promises δ-only constants.
assumptions (7)
  • standard math Large deviation bound for Bernoulli sums (Lemma 2, Durrett)
    Used for the α=1 achievability; standard.
  • standard math Moderate deviation bound for i.i.d. sums (Lemma 3, Eichelsbacher-Löwe)
    Used for α∈(0,1); however the application fixes c_M=1/2 independent of the error exponent m.
  • standard math Stein/CLT concentration bound (Lemma 4, Chen-Goldstein-Shao)
    Used for α=0.
  • standard math GV lower bound and Hamming upper bound (Lemma 1)
    Converts minimum-distance constraints into rate bounds.
  • domain assumption Total-variation distinguishability converse of [22, Thm III.2] (Proposition 2)
    The cited theorem from the authors of that paper is used as an input; it is a known result but not proved here.
  • ad hoc to paper λ1,n and λ2,n decay at comparable rates so that η_n=min orders λ1,n+λ2,n
    Not stated as an assumption anywhere, but required for the converse's inference about the minimum distance; without it the counterexample λ1=e^{-Ln}, λ2=0.5 breaks the conclusion.
  • ad hoc to paper The moderate-deviation concentration constant can be fixed to 1/2 regardless of the prefactor m in -ln η_n = m n^α
    This is effectively an extra assumption, false for m>1/2, that the proof relies on in Case 2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Reliability-Dependent Scaling Laws of Deterministic Identification over Binary Symmetric Channels." pith.science (2026). https://pith.science/paper/JILG2IPE

@misc{pith2026260803282,
  author       = {Pith},
  title        = {Pith review of: Reliability-Dependent Scaling Laws of Deterministic Identification over Binary Symmetric Channels},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JILG2IPE}},
  note         = {Machine review of arXiv:2608.03282}
}
read the original abstract

In this paper, we study the asymptotic behavior of deterministic identification (DID) over binary symmetric channels (BSCs) under vanishing error constraints. By introducing a minimum error parameter, we characterize how different error-decay regimes affect the achievable DID rate. General achievability and converse bounds are derived, with explicit asymptotic characterizations in the large-deviation, moderate-deviation, and central-limit regimes. The achievability analysis combines coding-theoretic constructions with probabilistic concentration techniques, while the converse links statistical distinguishability to the minimum-distance structure of DID codes via total variation and Hamming-type bounds. Our results show that the asymptotic behavior of DID over BSCs is governed by a Hamming-shell concentration geometry of channel outputs, offering insights into the finite-blocklength behavior of deterministic identification over discrete-output channels.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

37 extracted references · 34 canonical work pages

  1. [29]

    Rate-reliability tradeoff for deterministic identification,

    P. Colomer, C. Deppe, H. Boche, and A. Winter, “Rate-reliability tradeoff for deterministic identification,”IEEE Trans. Commun., vol. 73, no. 12, pp. 14107–14123, Dec. 2025

  2. [1]

    Integration of energy, computation and communication in 6G cellular internet of things,

    Q. Qi, X. Chen, C. Zhong, and Z. Zhang, “Integration of energy, computation and communication in 6G cellular internet of things,”IEEE Commun. Lett., vol. 24, no. 6, pp. 1333-1337, Jun. 2020

  3. [2]

    A comprehensive survey on Internet of Things (IoT) toward 5G wireless systems,

    L. Chettri and R. Bera, “A comprehensive survey on Internet of Things (IoT) toward 5G wireless systems,”IEEE Internet Things J., vol. 7, no. 1, pp. 16–32, Jan. 2020

  4. [3]

    Secure identification for wiretap channels; robustness, super-additivity and continuity,

    H. Boche and C. Deppe, “Secure identification for wiretap channels; robustness, super-additivity and continuity,”IEEE Trans. Inf. Forensics Security, vol. 13, no. 7, pp. 1641–1655, Jul. 2018

  5. [4]

    On the influence of scattering from traffic signs in vehicle- to-x communications,

    K. Guan, B. Ai, M. Liso Nicol´as, R. Geise, A. M¨oller, Z. Zhong, and T. K¨orner, “On the influence of scattering from traffic signs in vehicle- to-x communications,”IEEE Trans. Veh. Technol., vol. 65, no. 8, pp. 5835–5849, Aug. 2016

  6. [5]

    Millimeter-wave vehicular communication to support massive automotive sensing,

    J. Choi, V . Va, N. Gonzalez-Prelcic, R. Daniels, C. R. Bhat, and R. W. Heath, “Millimeter-wave vehicular communication to support massive automotive sensing,”IEEE Commun. Mag., vol. 54, no. 12, pp. 160–167, Dec. 2016

  7. [6]

    The tactile Internet: Applications and challenges,

    G. P. Fettweis, “The tactile Internet: Applications and challenges,”IEEE Veh. Technol. Mag., vol. 9, no. 1, pp. 64–70, Mar. 2014

  8. [7]

    Robust WHT-GFDM for the next generation of wireless networks,

    N. Michailow, L. Mendes, M. Matth ´e, I. Gaspar, A. Festag, and G. Fettweis, “Robust WHT-GFDM for the next generation of wireless networks,”IEEE Commun. Lett., vol. 19, no. 1, pp. 106–109, Jan. 2015

Show all 37 references
  1. [8]

    Identification via channels,

    R. Ahlswede and G. Dueck, “Identification via channels,”IEEE Trans. Inf. Theory, vol. 35, no. 1, pp. 15-29, Jan. 1989

  2. [9]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,”Bell Syst. Tech. J., vol. 27, no. 3, pp. 379-423, Jul. 1948

  3. [10]

    Approximation theory of output statistics,

    T. S. Han and S. Verd ´u, “Approximation theory of output statistics,” IEEE Trans. Inf. Theory, vol. 39, no. 3, pp. 752-772, May 1993

  4. [11]

    New converses in the theory of identification via chan- nels

    Y . Steinberg, “New converses in the theory of identification via chan- nels”,IEEE Trans. Inf. Theory, vol. 44, no. 3, pp. 984-998, May 1998

  5. [12]

    On identification capacity of infinite alphabets or continuous-time channels,

    M. V . Burnashev, “On identification capacity of infinite alphabets or continuous-time channels,”IEEE Trans. Inf. Theory, vol. 46, no. 7, pp. 2407–2414, Nov. 2000

  6. [13]

    T. S. Han,Information-Spectrum Methods in Information Theory, Springer Berlin Heidelberg, 2003

  7. [14]

    General nonasymptotic and asymptotic formulas in chan- nel resolvability and identification capacity and their application to the wiretap channel,

    M. Hayashi, “General nonasymptotic and asymptotic formulas in chan- nel resolvability and identification capacity and their application to the wiretap channel,”IEEE Trans. Inf. Theory, vol. 52, no. 4, pp. 1562-1575, Apr. 2006

  8. [15]

    Minimax converse for identification via channels,

    S. Watanabe, “Minimax converse for identification via channels,”IEEE Trans. Inf. Theory, vol. 68, no. 1, pp. 25-34, Jan. 2022

  9. [16]

    Z. Liu, Y . Li, H. Zhang, J. Wang, G. Yan and Z. Ma, ”Second-Order Identification Capacity of AWGN Channels,” inProc. IEEE Int. Symp. Inf. Theory, July 2024, pp. 309-314

  10. [17]

    Explicit construction of optimal constant- weight codes for identification via channels,

    S. Verdu and V . K. Wei, “Explicit construction of optimal constant- weight codes for identification via channels,”IEEE Trans. Inf. Theory, vol. 39, no. 1, pp. 30–36, Jan. 1993

  11. [18]

    Identification is easier than decoding,

    J. J´ aJ´ a, “Identification is easier than decoding,”in Ann. Symp. Found. Comp. Scien. (SFCS), 1985, pp. 43–50

  12. [19]

    Identification without randomization,

    R. Ahlswede and Ning Cai, “Identification without randomization,” IEEE Trans. Inf. Theory, vol. 45, no. 7, pp. 2636–2642, 1999

  13. [20]

    Secure identification for Gaussian channels,

    W. Labidi, C. Deppe, and H. Boche, “Secure identification for Gaussian channels,”in Proc. IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP), May 2020, pp. 2872–2876

  14. [21]

    Deterministic identification over channels with power constraints,

    M. J. Salariseddigh, U. Pereg, H. Boche, and C. Deppe, “Deterministic identification over channels with power constraints,”IEEE Trans. Inf. Theory, vol. 68, no. 1, pp. 1–24, Jan. 2022

  15. [22]

    Deterministic iden- tification over channels with finite output: A dimensional perspective on superlinear rates,

    P. Colomer, C. Deppe, H. Boche, and A. Winter, “Deterministic iden- tification over channels with finite output: A dimensional perspective on superlinear rates,”IEEE Trans. Inf. Theory, vol. 71, no. 5, pp. 3373–3396, May 2025

  16. [23]

    Optimal codes for deterministic identification over Gaussian channels: Closing the capacity gap,

    P. Colomer, C. Deppe, H. Boche, and A. Winter, “Optimal codes for deterministic identification over Gaussian channels: Closing the capacity gap,” arXiv preprint arXiv:2604.11782, 2026

  17. [24]

    Secure identification for Gaussian channels,

    W. Labidi, C. Deppe, and H. Boche, “Secure identification for Gaussian channels,” inProc. IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP), May 2020, pp. 2872–2876

  18. [25]

    Deterministic identification over fading channels,

    M. J. Salariseddigh, U. Pereg, H. Boche, and C. Deppe, “Deterministic identification over fading channels,” inProc. IEEE Inf. Theory Workshop (ITW), Apr. 2021, pp. 1–5

  19. [26]

    Deterministic identification over Poisson channels,

    M. J. Salariseddigh, U. Pereg, H. Boche, C. Deppe, and R. Schober, “Deterministic identification over Poisson channels,” inProc. IEEE Globecom Workshops (GC Wkshps), Madrid, Spain, 2021, pp. 1–6

  20. [27]

    6G and the Post-Shannon theory,

    J. Cabrera, H. Boche, C. Deppe, R. F. Schaefer, C. Scheunert, and F. H. Fitzek, “6G and the Post-Shannon theory,” inShaping Future 6G Networks: Needs, Impacts and Technologies. Hoboken, NJ, USA: Wiley, 2021, pp. 271–294

  21. [28]

    Deterministic iden- tification for Bernoulli channels and related channels with continuous input,

    P. Colomer, C. Deppe, H. Boche, and A. Winter, “Deterministic iden- tification for Bernoulli channels and related channels with continuous input,” arXiv preprint arXiv:2605.05168, 2026

  22. [30]

    Rate-reliability tradeoff for deterministic identification over Gaussian channels,

    P. Colomer, C. Deppe, H. Boche, and A. Winter, “Rate-reliability tradeoff for deterministic identification over Gaussian channels,” arXiv preprint arXiv:2602.12182, 2026

  23. [31]

    A comparison of signalling alphabets,

    E. N. Gilbert, “A comparison of signalling alphabets,”The Bell system technical journal, vol. 31, no. 3, pp. 504–522, 1952

  24. [32]

    Estimate of the number of signals in error correcting codes,

    R. R. Varshamov, “Estimate of the number of signals in error correcting codes,”Docklady Akad. Nauk, SSSR, vol. 117, pp. 739–741, 1957

  25. [33]

    I. B. Djordjevic,Quantum Information Processing, Quantum Computing, and Quantum Error Correction: An Engineering Approach, 2nd ed. Cambridge, MA, USA: Academic Press, 2021

  26. [34]

    Durrett,Probability: Theory and Examples, 5th ed.Cambridge: Cambridge University Press, 2019

    R. Durrett,Probability: Theory and Examples, 5th ed.Cambridge: Cambridge University Press, 2019

  27. [35]

    Moderate deviations for i.i.d. random variables,

    P. Eichelsbacher and M. L ¨owe, “Moderate deviations for i.i.d. random variables,”ESAIM Probab. Stat., vol. 7, pp. 209–218, 2003

  28. [36]

    L. H. Y . Chen, L. Goldstein, and Q.-M. Shao,Normal Approximation by Stein’s Method. New York: Springer, 2011

  29. [37]

    Wells,The Penguin Dictionary of Curious and Interesting Numbers

    D. Wells,The Penguin Dictionary of Curious and Interesting Numbers. Middlesex, UK: Penguin Books, 1986

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.