Pith. sign in

REVIEW 1 major objections 6 minor 52 references

Remote Channel Synthesis

T0 review · 1 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This paper proves that the closure of the achievable rate region for remote channel synthesis is exactly a single-letter region, and that directly synthesizing the observed channel is strictly suboptimal when common randomness is scarce.

desk verdict The single-letter characterization in Theorem 3.1 is new and sound; the abstract's vector-scheme suboptimality claim rests on an unproven sketch that needs fixing or qualification. read the letter →

arxiv 2507.15757 v1 pith:7TMNC4BL submitted 2025-07-21 cs.IT math.IT

Yassine Hamdi , Deniz Gündüz This is my paper · ORCID
classification cs.ITmath.IT MSC 94A1594A1794A24
keywords remotechannelsynthesisstrongcoordinationcommonrandomnesssingle-lettercharacterizationlikelihoodencoderdirectpropervectorschemeWynerinformation
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

Remote channel synthesis asks how many bits per symbol an encoder must send, and how much common randomness it needs, for a decoder to produce an output sequence $Y^n$ whose joint law with an unobserved source $X^n$ is nearly a prescribed i.i.d. law $q_{X,Y}^{\otimes n}$, when the encoder only sees a noisy version $Z^n$ of $X^n$. The paper's central claim is a complete single-letter answer: a rate pair $(R,R_c)$ is optimal exactly when there is an auxiliary random variable $W$ with $X-Z-W-Y$, alphabet size at most $|Y||Z|+1$, satisfying $R \geq I(Z;W)$ and $R+R_c \geq I(X,Y;W)$. The proof runs through the likelihood-encoder achievability scheme and a converse that reduces any block code to a single-letter distribution. The second main claim is that the obvious scheme, synthesizing a compatible channel from $Z$ to $Y$ and stopping, is strictly suboptimal for small common randomness in most nontrivial cases, so optimal schemes must use vector coding across the block.

What carries the argument

The load-bearing object is the auxiliary variable $W$ in the Markov chain $X-Z-W-Y$, which stands for the codebook index (message plus common randomness plus time-sharing) that mediates between the encoder's observation and the decoder's output. The region is defined by two mutual information constraints, $R \geq I_p(Z;W)$ and $R+R_c \geq I_p(X,Y;W)$, together with the fixed marginals $p_{X,Z}=q_{X,Z}$ and $p_{X,Y}=q_{X,Y}$; the auxiliary alphabet bound $|W| \leq |Y||Z|+1$ comes from a Carath\'eodory theorem for connected sets applied to the set of achievable marginal-entropy pairs. Achievability is carried by a random codebook drawn from $p_W$, the likelihood encoder, and soft covering, while the converse uses $W=(M,J,T)$ extracted from an arbitrary code and the same entropy-continuity argument used for the standard channel synthesis problem.

What would settle it

Take binary $Z$ and $Y$ with fixed feasible marginals $q_{X,Z}, q_{X,Y}$, form the set $\tilde{E}$ of marginal-entropy points defined in Claim B.5, and check whether every point produced by a valid block code lies in the convex hull of $\tilde{E}$ with at most $|Y||Z|+1$ atoms. A single feasible instance whose Carath\'eodory number exceeds $|Y||Z|+1$ would falsify the alphabet bound and hence the converse of Theorem 3.1.

Watch

Extended reading notes

Core claim

Theorem 3.1 states that the closure of the achievable rate region for remote channel synthesis equals the set $S^{\mathrm{(r.c.s.)}}$ of all $(R,R_c)$ for which there exists a distribution $p_{X,Z,W,Y}$ with $p_{X,Z}=q_{X,Z}$, $p_{X,Y}=q_{X,Y}$, the Markov chain $X-Z-W-Y$, $|W| \leq |Y||Z|+1$, and the inequalities $R \geq I_p(Z;W)$ and $R+R_c \geq I_p(X,Y;W)$. The region is nonempty exactly when $q_{X,Y}$ factors through $Z$ as $q_{X,Y}(x,y)=\sum_z q_Z(z)q_{X|Z}(x|z)q_{Y|Z}(y|z)$ for some conditional distribution $q_{Y|Z}$. Achievability uses a random codebook drawn from $p_W$ with the likelihood encoder and soft covering; the converse extracts $W=(M,J,T)$ from an arbitrary code and uses a Carath\'eodory-type argument to cap the auxiliary alphabet. The paper further proves that for symmetric binary marginals with $0<q_{X,Z}(X\neq Z)<q_{X,Y}(X\neq Y)<1/2$, when $R_c$ is small the minimum compression rate over direct channel synthesis is strictly larger than the true optimum, so any optimal scheme must be a proper vector scheme whose $Z^n,Y^n$ joint law does not approach a product distribution.

Load-bearing premise

The converse relies on a convex-geometry fact: the set of single-letter distributions with the required marginals and entropy values is connected and compact, so every point in its convex hull can be represented using at most $|Y||Z|+1$ auxiliary symbols; if that bound ever failed, the stated region would be too small.

Editorial extensions

If this is right

  • When $R_c=0$, the region reduces to the condition $R \geq \max(I_p(Z;W), I_p(X,Y;W))$, so the zero-common-randomness optimum is a finite-dimensional minimization over $W$, computable in principle.
  • Direct channel synthesis, which synthesizes a compatible scalar channel $q_{Y|Z}$, always lies inside the remote region, but under the binary symmetric conditions of Proposition 3.3 it costs strictly more compression rate when $R_c$ is small.
  • Any scheme achieving the optimal compression rate in the low-common-randomness regime must be a proper vector scheme: the joint law of $Z^n$ and $Y^n$ cannot approach a product distribution, so the decoder's outputs must remain correlated across time.
  • The region is nonempty exactly when $q_{X,Y}$ factors through $Z$ as $q_{X,Y}(x,y)=\sum_z q_Z(z)q_{X|Z}(x|z)q_{Y|Z}(y|z)$, which gives a direct feasibility test for remote coordination at any finite rate.

Reading between the lines

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

  • One implication the paper leaves implicit is that the auxiliary alphabet bound makes the entire rate region computable by numerical optimization over a finite-dimensional distribution polytope, so the theorem is not only an existence result but an algorithm-ready characterization.
  • The strict suboptimality of scalar direct synthesis at low common randomness is a design warning for learned compression and federated-learning schemes that treat the encoder's noisy observation as a clean source: block-level or vector codes are needed to reach the true rate when shared randomness is scarce.
  • A natural testable extension, not pursued in the paper, is to quantify the gap in Figure 4 for non-binary or asymmetric sources; the same entropy-slope argument given in Appendix D should go through whenever the relevant entropy arguments stay in the increasing concave region of the binary entropy function.
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

1 major / 6 minor

Summary. The paper studies remote channel synthesis: an encoder observes a noisy version Z^n of a remote source X^n and sends a message over a noiseless link to a decoder that outputs Y^n, with the goal of making (X^n,Y^n) close in total variation to an i.i.d. target q_{X,Y}^{⊗n}; the encoder and decoder may share common randomness at rate R_c. The main result, Theorem 3.1, states that the closure of the achievable rate region A^(r.c.s.) equals the single-letter region S^(r.c.s.) defined by the existence of p_{X,Z,W,Y} with marginals q_{X,Z}, q_{X,Y}, Markov chain X–Z–W–Y, |W|≤|Y||Z|+1, and inequalities R≥I(Z;W), R+R_c≥I(X,Y;W). The proof is given in Appendices B and C, closely following Cuff's converse and achievability arguments with the likelihood encoder and soft-covering lemma. The paper also studies direct channel synthesis (d.c.s.), proves in Proposition 3.3 that for a specific binary class direct synthesis is strictly sub-optimal at small common-randomness rates, and argues in Section IV that optimal schemes must be "proper vector schemes."

Significance. If the main theorem is correct, it provides a complete single-letter characterization of the compression and common-randomness rate region for remote strong coordination, extending Cuff's channel synthesis to the case where the encoder observes only a noisy version of the source. The proof of Theorem 3.1 is detailed and largely sound: the Carathéodory cardinality argument in Claim B.5 is valid because the relevant set is a continuous image of a connected compact set in an affine space of dimension |Y||Z|+1, and the achievability direction correctly uses the soft-covering lemma and the likelihood encoder. The paper also gives a concrete algebraic proof of strict sub-optimality of direct channel synthesis in a nontrivial binary class (Proposition 3.3), with a checkable entropy inequality in Appendix D. However, the secondary claim about the necessity of "proper vector schemes" (Proposition 4.3) is not rigorously established, so the advertised vector-scheme interpretation currently rests on an unproven claim.

major comments (1)
  1. [Section IV (proof of Proposition 4.3, Claim 4.5)] The proof of Claim 4.5 is only a sketch, and its key step is false as stated: the sentence "there exist only a finite number of different conditional distributions ρ_{Y|Z}" is not true for general finite alphabets with |Y|,|Z|≥2, since the set of stochastic matrices is a continuum. The asserted simultaneous existence of t_k and n_k satisfying (20) and (21) is not demonstrated, and even if such a sequence existed, it would not select a single t satisfying (19) without an additional compactness or continuity argument. In the binary setting of Proposition 3.3 the compatible channel is actually unique, so the claim may be repairable by proving that uniqueness and connecting it to (19), but the manuscript does not do so. As written, Proposition 4.3, which underlies the vector-scheme interpretation in the abstract, is not established.
minor comments (6)
  1. [Appendix C and Proposition 3.2] The text states S^(r.c.s.) ⊆ A^(r.c.s.) and A^(r.c.s.) = S^(r.c.s.), but Theorem 3.1 only proves that the closure of A^(r.c.s.) equals S^(r.c.s.). The achievability proof shows that for every ε>0 the pair (R+ε,R_c) is achievable, which gives S^(r.c.s.) ⊆ closure(A^(r.c.s.)), not S^(r.c.s.) ⊆ A^(r.c.s.) without an additional closedness argument. This should be corrected to "closure of A" throughout.
  2. [Appendix C after Eq. (48)] The sentence "From (2.1), we have p_Z ≡ q_Z" refers to a nonexistent equation; it should refer to the defining property of D^(r.c.s.) in (7), where p_{X,Z} ≡ q_{X,Z}.
  3. [Section IV (proof of Proposition 4.3)] The sentence beginning "Since there are only a finite number of possible single-letter conditional distributions, then (e.g., from [3, Lemma V.1]) P_{Z^n,Y^n} is nearly i.i.d. by part" is garbled and incomplete; it should be rewritten, and it should not rely on the false finiteness assertion.
  4. [Claim 4.5] There is a typo: "There exits t ∈ N" should read "There exists t ∈ N".
  5. [Figure 4] The plot is referenced as a lower bound on the gap, but the caption only defines θ and τ; it would be clearer to state the explicit expression being plotted and the range of parameters used.
  6. [Abstract and Section I] The phrase "in most cases" in the abstract is not formalized; Proposition 3.3 is proved for a specific binary class satisfying (13). The wording should be aligned with the actual theorem, e.g., by saying "for a class of binary sources".

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the single-letter characterization is derived from external soft-covering and single-letterization lemmas, with no fit or self-citation chain reducing the result to its inputs.

full rationale

The paper derives Theorem 3.1 by giving a standard single-letter region S^(r.c.s.) in terms of an auxiliary random variable W, then proving achievability via Cuff's soft-covering lemma ([3, Corollary IV.1]) and the converse via the usual W=(M,J,T) construction, entropy continuity ([3, Lemma VI.3]), and a Caratheodory-for-connected-sets cardinality bound ([3, Lemma VI.1]). These are external, machine-checkable-style results from the published literature; they do not incorporate the target rate region as an input. The auxiliary variable W is optimized subject to the given marginals and Markov chain; it is not fitted to the achievable region, and the region is not defined in terms of A^(r.c.s.). The paper's use of its own prior work ([48]) is confined to notation in Appendix C and is not load-bearing. The only notable weakness is Proposition 4.3 / Claim 4.5, where the proof sketch asserts that 'there exist only a finite number of different conditional distributions rho_{Y|Z}' — false for general finite alphabets since the stochastic simplex is a continuum — and the simultaneous existence of t_k, n_k satisfying (20)-(21) is not demonstrated. This is a correctness gap in the advertised vector-scheme sub-optimality claim, not a circularity: the claim does not reduce to its inputs by definition, and Theorem 3.1 is independent of it. There is no fitted parameter renamed as a prediction, no self-citation invoked to forbid alternatives, and no equation equal to its own input by construction. Accordingly the circularity score is 0.

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

No free parameters are fitted; the rates are characterized by mutual information expressions over an auxiliary variable W, which is an optimization variable, not a model parameter. The paper inherits standard lemmas from Cuff's channel synthesis and Wyner's common information. No new physical or mathematical entities are postulated; the 'vector scheme' is a definition, not an entity.

assumptions (7)
  • domain assumption The source (X_i,Z_i) are i.i.d. with known joint distribution q_{X,Z}.
    Standard i.i.d. model used throughout; the code is designed for a known distribution.
  • domain assumption The target pair (q_{X,Z}, q_{X,Y}) is feasible, i.e., there exists a channel q_{Y|Z} such that q_{X,Y}(x,y)=sum_z q_Z(z) q_{X|Z}(x|z) q_{Y|Z}(y|z).
    Necessary for the problem to be solvable; stated in Section III after Theorem 3.1. If not feasible, the region is empty.
  • standard math The soft-covering lemma (Cuff [3, Cor. IV.1]) is valid and applicable to the likelihood encoder with random codebooks.
    Used in Appendix C to prove achievability (Claim C.1). It is an external theorem from the channel synthesis literature.
  • standard math The entropy continuity bound [3, Lemma VI.3] connects total-variation closeness of the n-letter distribution to a O(epsilon log(1/epsilon)) bound on per-letter dependence.
    Used in the converse (Claim B.4, equations (30)-(32)) to bound conditional mutual information terms.
  • standard math The Caratheodory theorem for connected sets (Fenchel-Bunt) allows representing points in the convex hull of the connected set E-tilde with |Y||Z|+1 atoms.
    Used in Claim B.5 to bound the alphabet size of the auxiliary variable W and thus obtain a single-letter characterization.
  • standard math The Wyner common information of DSBS(theta) is 1-h(theta), achieved by a doubly symmetric binary source construction with crossover theta-tilde = 1/2 - 1/2 sqrt(1-2theta).
    Used in Appendix D, Case 5 of Proposition 3.3's proof, following [7, Section 3].
  • domain assumption The source alphabets X, Z, Y are finite.
    Stated in Section II.B; the entropy calculations and Caratheodory argument require finiteness.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Remote Channel Synthesis." pith.science (2026). https://pith.science/paper/7TMNC4BL

@misc{pith2026250715757,
  author       = {Pith},
  title        = {Pith review of: Remote Channel Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7TMNC4BL}},
  note         = {Machine review of arXiv:2507.15757}
}
abstract

We consider the problem of synthesizing a memoryless channel between an unobserved source and a remote terminal. An encoder has access to a partial or noisy version $Z^n = (Z_1, \ldots, Z_n)$ of a remote source sequence $X^n = (X_1, \ldots, X_n),$ with $(X_i,Z_i)$ independent and identically distributed with joint distribution $q_{X,Z}.$ The encoder communicates through a noiseless link to a decoder which aims to produce an output $Y^n$ coordinated with the remote source; that is, the total variation distance between the joint distribution of $X^n$ and $Y^n$ and some i.i.d. target distribution $q_{X,Y}^{\otimes n}$ is required to vanish as $n$ goes to infinity. The two terminals may have access to a source of rate-limited common randomness. We present a single-letter characterization of the optimal compression and common randomness rates. We also show that when the common randomness rate is small, then in most cases, coordinating $Z^n$ and $Y^n$ using a standard channel synthesis scheme is strictly sub-optimal. In other words, schemes for which the joint distribution of $Z^n$ and $Y^n$ approaches a product distribution asymptotically are strictly sub-optimal.

Figures

Figures reproduced from arXiv: 2507.15757 by the authors.

Figure 1
Figure 1. One shot channel synthesis with common randomness. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. Our remote channel synthesis setup. the remote sequence Xn, such that (Xn, Zn) follows an i.i.d. distribution q ⊗n X,Z. This problem bears a resemblance with the remote rate-distortion problem [30], [31], [32, Chap. 3, Sec. 5], [33], in which a constraint of the form E[d(Xn , Y n )] ≤ ∆ is imposed instead of (1), where d is a measure of distortion. We obtain a single-letter characterization of the optimal rate pairs… view at source ↗
Figure 4
Figure 4. Lower bound on the gap between the optimal rates in [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Graphical model for Q. See Equations (42) and (43). denote this random codebook distribution by QC(n) . Given a realization c (n) , the codeowrd of index (m, j) is denoted w n(c (n) , m, j). B. Distribution Q and soft-covering lemma For every positive integer n, we def…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 49 canonical work pages

  1. [44]

    Remote joint strong coordination and reliable communication,

    G. Cervia, T. J. Oechtering, and M. Skoglund, “Remote joint strong coordination and reliable communication,” in 2020 IEEE International Symposium on Information Theory (ISIT) , 2020, pp. 932–937

    work page 2020

  2. [1]

    Compressing Images by Encoding Their Latent Representations with Relative Entropy Coding,

    G. Flamich et al. , “Compressing Images by Encoding Their Latent Representations with Relative Entropy Coding,” in Annual Conference on Neural Information Processing Systems , 2020, Conference Paper

    work page 2020

  3. [2]

    Entanglement-assisted capacity of a quantum channel and the reverse Shannon theorem,

    C. H. Bennett et al. , “Entanglement-assisted capacity of a quantum channel and the reverse Shannon theorem,” IEEE Transactions on Information Theory, vol. 48, no. 10, pp. 2637–2655, 2002

    work page 2002

  4. [3]

    Distributed Channel Synthesis,

    P. Cuff, “Distributed Channel Synthesis,” IEEE Transactions on Infor- mation Theory, vol. 59, no. 11, 2013

    work page 2013

  5. [4]

    The Quantum Reverse Shannon Theorem and Re- source Tradeoffs for Simulating Quantum Channels,

    C. H. Bennett et al., “The Quantum Reverse Shannon Theorem and Re- source Tradeoffs for Simulating Quantum Channels,” IEEE Transactions on Information Theory , vol. 60, no. 5, pp. 2926–2959, 2014

    work page 2014

  6. [5]

    Exact Channel Synthesis,

    L. Yu and V . Y . F. Tan, “Exact Channel Synthesis,” IEEE Transactions on Information Theory , vol. 66, no. 5, pp. 2799–2818, 2020

    work page 2020

  7. [6]

    Optimal Redundancy in Exact Channel Synthesis,

    S. M. Sriramu and A. B. Wagner, “Optimal Redundancy in Exact Channel Synthesis,” in IEEE International Symposium on Information Theory (ISIT), 2024

    work page 2024

  8. [7]

    The common information of two dependent random vari- ables,

    A. Wyner, “The common information of two dependent random vari- ables,” IEEE Transactions on Information Theory , vol. 21, no. 2, pp. 163–179, 1975

    work page 1975

Show all 52 references
  1. [8]

    Compression of Sources of Probability Distributions and Density Operators,

    A. Winter, “Compression of Sources of Probability Distributions and Density Operators,” 2002, arXiv:quant-ph/0208131

    2002 arXiv

  2. [9]

    The Communication Complexity of Correlation,

    P. Harsha et al. , “The Communication Complexity of Correlation,” in IEEE Conference on Computational Complexity , 2007, Conference Proceedings

    2007

  3. [10]

    Strong Functional Representation Lemma and Applications to Coding Theorems,

    C. T. Li and A. E. Gamal, “Strong Functional Representation Lemma and Applications to Coding Theorems,” IEEE Transactions on Information Theory, vol. 64, no. 11, 2018

    2018

  4. [11]

    The Quantum Reverse Shannon Theorem and Re- source Tradeoffs for Simulating Quantum Channels,

    C. H. Bennett et al., “The Quantum Reverse Shannon Theorem and Re- source Tradeoffs for Simulating Quantum Channels,” IEEE Transactions on Information Theory , vol. 60, no. 5, 2014

    2014

  5. [12]

    Coordination capacity,

    P. W. Cuff et al. , “Coordination capacity,” IEEE Transactions on Information Theory, vol. 56, no. 9, pp. 4181–4206, 2010

    2010

  6. [13]

    Fast Relative Entropy Coding with A* Coding,

    G. Flamich, S. Markou, and J. M. Hernandez-Lobato, “Fast Relative Entropy Coding with A* Coding,” in International Conference on Machine Learning, 2022, Conference Paper

    2022

  7. [14]

    Greedy Poisson Rejection Sampling,

    G. Flamich, “Greedy Poisson Rejection Sampling,” in Annual Con- ference on Neural Information Processing Systems , 2023, Conference Paper

    2023

  8. [15]

    Faster Relative Entropy Coding with Greedy Rejection Coding,

    G. Flamich, S. Markou, and J. M. Hernández-Lobato, “Faster Relative Entropy Coding with Greedy Rejection Coding,” in Annual Conference on Neural Information Processing Systems , 2023, Conference Paper

    2023

  9. [16]

    Adaptive Greedy Rejection Sampling,

    G. Flamich and L. Theis, “Adaptive Greedy Rejection Sampling,” in IEEE International Symposium on Information Theory , 2023, Confer- ence Paper

    2023

  10. [17]

    Accelerating Relative Entropy Coding with Space Partitioning,

    H. Jiajun, F. Gergely, and H.-L. Jos Miguel, “Accelerating Relative Entropy Coding with Space Partitioning,” 2024, Conference Paper

    2024

  11. [18]

    Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters,

    H. Marton et al., “Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters,” in International Conference on Learning Representations, 2019

    2019

  12. [19]

    Compressing Images by Encoding Their Latent Representations with Relative Entropy Coding,

    G. Flamich et al. , “Compressing Images by Encoding Their Latent Representations with Relative Entropy Coding,” in Annual Conference on Neural Information Processing Systems , 2020

    2020

  13. [20]

    Universally Quantized Neural Compres- sion,

    E. Agustsson and L. Theis, “Universally Quantized Neural Compres- sion,” in Annual Conference on Neural Information Processing Systems , 2020

    2020

  14. [21]

    Lossy Compression with Gaussian Diffusion,

    L. Theis et al. , “Lossy Compression with Gaussian Diffusion,” 2022, arXiv:2206.08889

    2022 arXiv

  15. [22]

    Sparse Random Networks for Communication-Efficient Federated Learning,

    B. Isik et al. , “Sparse Random Networks for Communication-Efficient Federated Learning,” in International Conference on Learning Repre- sentations, 2023

    2023

  16. [23]

    Adaptive Compression in Federated Learning via Side Infor- mation,

    ——, “Adaptive Compression in Federated Learning via Side Infor- mation,” in International Conference on Artificial Intelligence and Statistics, 2024, Conference Paper

    2024

  17. [24]

    Dp-rec: Private & communication-efficient federated learning,

    A. Triastcyn, M. Reisser, and C. Louizos, “Dp-rec: Private & communication-efficient federated learning,” 2021, arXiv:2111.05454

    2021 arXiv

  18. [25]

    Communication efficient private fed- erated learning using dithering,

    B. Hasırcıo ˘glu and D. Gündüz, “Communication efficient private fed- erated learning using dithering,” in IEEE International Conference on Acoustics, Speech and Signal Processing , 2024

    2024

  19. [26]

    Compression with Exact Error Distribution for Federated Learning,

    Hegazy et al., “Compression with Exact Error Distribution for Federated Learning,” in International Conference on Artificial Intelligence and Statistics, 2024, Conference Paper

    2024

  20. [27]

    Communication-Efficient Laplace Mechanism for Differential Privacy via Random Quantization,

    A. M. Shahmiri, C. W. Ling, and C. T. Li, “Communication-Efficient Laplace Mechanism for Differential Privacy via Random Quantization,” in IEEE International Conference on Acoustics, Speech and Signal Processing, 2024

    2024

  21. [28]

    Universal exact com- pression of differentially private mechanisms,

    Y . Liu, W.-N. Chen, A. Özgür, and C. T. Li, “Universal exact com- pression of differentially private mechanisms,” in Advances in Neural Information Processing Systems , vol. 37, 2024, pp. 91 492–91 531

    2024

  22. [29]

    Channel simulation: Theory and applications to lossy compression and differential privacy,

    C. T. Li, “Channel simulation: Theory and applications to lossy compression and differential privacy,” Foundations and Trends® in Communications and Information Theory , vol. 21, no. 6, pp. 847–1106,

  23. [30]

    Information Transmission With Addi- tional Noise,

    R. Dobrushin and B. Tsybakov, “Information Transmission With Addi- tional Noise,” IRE Transactions on Information Theory , vol. 8, no. 5, 1962

    1962

  24. [31]

    Transmission of Noisy Information to a Noisy Receiver With Minimum Distortion,

    J. Wolf and J. Ziv, “Transmission of Noisy Information to a Noisy Receiver With Minimum Distortion,” IEEE Transactions on Information Theory, vol. 16, no. 4, 1970

    1970

  25. [32]

    Berger, Rate Distortion Theory: A Mathematical Basis for Data Compression

    T. Berger, Rate Distortion Theory: A Mathematical Basis for Data Compression. NJ: Prentice Hall: Englewood Cliffs, 1971

    1971

  26. [33]

    Indirect Rate Distortion Problems,

    H. Witsenhausen, “Indirect Rate Distortion Problems,” IEEE Transac- tions on Information Theory , vol. 26, no. 5, 1980

    1980

  27. [34]

    Approximation theory of output statistics,

    T. Han and S. Verdu, “Approximation theory of output statistics,” IEEE Transactions on Information Theory , vol. 39, no. 3, pp. 752–772, 1993

    1993

  28. [35]

    Common information is far less than mutual information

    P. Gács and J. Korner, “Common information is far less than mutual information.” Problems of Control and Information Theory , vol. 2, pp. 149–162, 1973

    1973

  29. [36]

    Generating dependent random variables over networks,

    A. A. Gohari and V . Anantharam, “Generating dependent random variables over networks,” in 2011 IEEE Information Theory Workshop , Conference Proceedings, pp. 698–702

    2011

  30. [37]

    Channel Simulation via Interactive Communi- cations,

    M. H. Yassaee et al. , “Channel Simulation via Interactive Communi- cations,” IEEE Transactions on Information Theory , vol. 61, no. 6, pp. 2964–2982, 2015

    2015

  31. [38]

    Strong Coordination over Noisy Channels with Strictly Causal Encoding,

    G. Cervia et al., “Strong Coordination over Noisy Channels with Strictly Causal Encoding,” in Annual Allerton Conference on Communication, Control, and Computing (Allerton) , 2018

    2018

  32. [39]

    Communication for Generating Correlation: A Unify- ing Survey,

    M. Sudan et al., “Communication for Generating Correlation: A Unify- ing Survey,” IEEE Transactions on Information Theory , vol. 66, no. 1, pp. 5–37, 2020

    2020

  33. [40]

    Source coding for synthesizing correlated randomness,

    T. A. Atif et al., “Source coding for synthesizing correlated randomness,” IEEE Transactions on Information Theory , vol. 69, no. 1, pp. 626–649, 2023

    2023

  34. [41]

    One-Shot Coding over General Noisy Networks,

    Y . Liu and C. T. Li, “One-Shot Coding over General Noisy Networks,” in IEEE International Symposium on Information Theory (ISIT) , 2024

    2024

  35. [42]

    Broadcast Channel Synthesis from Shared Randomness,

    M. A. Managoli and V . M. Prabhakaran, “Broadcast Channel Synthesis from Shared Randomness,” in 2024 IEEE International Symposium on Information Theory (ISIT) , Conference Proceedings, pp. 1919–1924

    2024

  36. [43]

    Channel Simulation: Finite Blocklengths and Broad- cast Channels,

    M. X. Cao et al., “Channel Simulation: Finite Blocklengths and Broad- cast Channels,” IEEE Transactions on Information Theory , vol. 70, no. 10, pp. 6780–6808, 2024

    2024

  37. [45]

    El Gamal and Y .-H

    A. El Gamal and Y .-H. Kim, Network Information Theory. Cambridge (England): Cambridge University Press, 2011

    2011

  38. [46]

    On Extracting Common Random Bits From Corre- lated Sources on Large Alphabets,

    S. O. Chan et al., “On Extracting Common Random Bits From Corre- lated Sources on Large Alphabets,” IEEE Transactions on Information Theory, vol. 60, no. 3, pp. 1630–1637, 2014

    2014

  39. [47]

    The likelihood encoder for lossy compression,

    E. C. Song, P. Cuff, and H. V . Poor, “The likelihood encoder for lossy compression,” IEEE Transactions on Information Theory, vol. 62, no. 4, pp. 1836–1849, 2016

    2016

  40. [48]

    The Rate-Distortion-Perception Trade-off with Side Information,

    Y . Hamdi and D. Gündüz, “The Rate-Distortion-Perception Trade-off with Side Information,” in IEEE International Symposium on Informa- tion Theory, 2023. APPENDIX A USEFUL LEMMAS Lemma A.1: [3, Lemma V .1] Let Π and Γ be two distribu- tions on an alphabet U × L. Then ∥ΠU − ΓU ...

    2023

  41. [50]

    (45) The key difference with respect to the proof in [3] for standard channel synthesis is the following. We can introduce the remote source X via Lemma A.2 with L = X n, U= (J, Zn), and ΠL|U the memoryless channel Q qX|Z: from (42) and (45), we get EC(n) ∥QJ,Zn,X n|C(n) − pU ...

  42. [51]

    Let R be a non-negative real number and let (kn)n≥1 be a sequence of positive integers satisfying kn/2nR →n→∞1

    (46) Proof of Claim C.1: Lemma C.2: [3, Corollary IV .1] LetW be a finite alphabet and ρW a distribution on the latter. Let R be a non-negative real number and let (kn)n≥1 be a sequence of positive integers satisfying kn/2nR →n→∞1. For every positive integer n, let E (n) be a ...

  43. [52]

    Therefore, QZn|C(n),J ≡ QZn|C(n) J ,J

    (48) From (42), we have ∀j, QZn|C(n) j ,J=j = ψ(C(n) j ), where ψ : {wn(m)}m∈[kn] ∈ (W n)kn 7→ 1 kn knX m=1 nY t=1 pZ|W =wt(m) is a map which is defined independently of any index j. Therefore, QZn|C(n),J ≡ QZn|C(n) J ,J . Moreover, since from (42) we have QC(n),J ≡ QC(n) pU [...

  44. [2024]

    Available: http://dx.doi.org/10.1561/0100000141

    [Online]. Available: http://dx.doi.org/10.1561/0100000141

    doi

Pith tools

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