Constant term of one-shot channel simulation

Fact	Value	Source
Best lower bound with an explicit value (constant 32a)	-0.53 [Li25]	Tao's optimization constants
Best upper bound with an explicit value (constant 32a)	0.76 [Li25]	Tao's optimization constants

Description of constant

The constant term of one-shot channel simulation [HJMR07], [BG14], [LEG18], [Li25] is given as (we use the definition in [Li25])

$C_{32} = limsup_{t \to \infty} (p_{X, Y} : I (X; Y) = t sup p_{S ∣ X, Y} : I (X; S) = 0 in f H (Y ∣ S) - t - lo g_{2} t),$

where $H (Y ∣ S) = H (Y, S) - H (S)$ is the conditional entropy (in bits), and $I (X; Y) = H (X) + H (Y) - H (X, Y)$ is the mutual information (in bits). The supremum and infimum are over arbitrary finite discrete joint probability distributions $p_{X, Y}$ with $I (X; Y) = t$ and finite discrete conditional distributions $p_{S ∣ X, Y}$ with $I (X; S) = 0$ , respectively.

Equivalently, it is the smallest (i.e., infimum of) $α \in R$ satisfying that there exists $β > 0$ such that for every jointly-distributed random variables $X, Y$ with finite support, we can construct a random variable $S$ with finite support (jointly-distributed with $X, Y$ , possibly after extending the probability space) with $I (X; S) = 0$ and $H (Y ∣ S) \leq I (X; Y) + lo g_{2} (I (X; Y) + β) + α$ [LEG18], [Li25].

Known upper bounds

Bound	Reference	Comments
$< \infty$	[HJMR07], [HJMR10], [BG14]
$3.871$	[LEG18]	Usually reported as 4.
$3.732$	[LA21]
$3.45$	[FT23]
$2$	[Li24b], [Li24a]
$\sum_{k = 1}^{\infty} 2^{- k - 1} k lo g_{2} k - lo g_{2} lo g_{2} e \approx 0.76$	[Li25]

Known lower bounds

Bound	Reference	Comments
$> - \infty$	[BG14]
$- 1$	[LEG18]
$- lo g_{2} lo g_{2} e \approx - 0.53$	[Li25]

Additional comments

The bound $- 1 \leq C_{32} < 3.871$ was given in [LEG18], where the following result (called "strong functional representation lemma") was shown: for every (not necessarily discrete) random variables $X, Y$ , there exists a (not necessarily discrete) random variable $S$ such that $I (X; S) = H (Y ∣ X, S) = 0$ and $H (Y ∣ S) \leq I (X; Y) + lo g (I (X; Y) + 1) + 3.871$ .

It is conjectured that $C_{32} = - lo g_{2} lo g_{2} e$ [Li25].

A bound on $C_{32}$ would have applications in communication complexity [HJMR07],[BG14], distributed channel simulation [FT23],[Li24a], lossy compression [LEG18],[LHB22], rate-distortion-perception trade-off [TW21], and privacy-utility trade-off [ZOS23].

References

[BG14] Mark Braverman and Ankit Garg, Public vs private coin in bounded-round information, International Colloquium on Automata, Languages, and Programming, Springer, 2014, pp. 502-513.
[FT23] Gergely Flamich and Lucas Theis, Adaptive greedy rejection sampling, 2023 IEEE International Symposium on Information Theory (ISIT), IEEE, 2023, pp. 454-459.
[HJMR07] Prahladh Harsha, Rahul Jain, David McAllester, and Jaikumar Radhakrishnan, The communication complexity of correlation, Twenty-Second Annual IEEE Conference on Computational Complexity (CCC'07), IEEE, 2007, pp. 10-23.
[HJMR10] Prahladh Harsha, Rahul Jain, David McAllester, and Jaikumar Radhakrishnan, The communication complexity of correlation, IEEE Transactions on Information Theory 56 (2010), no. 1, 438-449.
[LA21] Cheuk Ting Li and Venkat Anantharam, A unified framework for one-shot achievability via the Poisson matching lemma, IEEE Transactions on Information Theory 67 (2021), no. 5, 2624-2651.
[LEG18] Cheuk Ting Li and Abbas El Gamal, Strong functional representation lemma and applications to coding theorems, IEEE Transactions on Information Theory 64 (2018), no. 11, 6967-6978.
[LHB22] Eric Lei, Hamed Hassani, and Shirin Saeedi Bidokhti, Neural estimation of the rate-distortion function with applications to operational source coding, IEEE Journal on Selected Areas in Information Theory 3 (2022), no. 4, 674-686.
[Li24a] Cheuk Ting Li, Channel simulation: Theory and applications to lossy compression and differential privacy, Foundations and Trends in Communications and Information Theory 21 (2024), no. 6, 847-1106.
[Li24b] Cheuk Ting Li, Pointwise redundancy in one-shot lossy compression via Poisson functional representation, International Zurich Seminar on Information and Communication (IZS 2024), 2024.
[Li25] Cheuk Ting Li, Discrete layered entropy, conditional compression and a tighter strong functional representation lemma, 2025 IEEE International Symposium on Information Theory (ISIT), 2025. Full version: arXiv preprint arXiv:2501.13736.
[TW21] Lucas Theis and Aaron B Wagner, A coding theorem for the rate-distortion-perception function, Neural Compression: From Information Theory to Applications-Workshop@ ICLR 2021, 2021.
[ZOS23] Amirreza Zamani, Tobias J Oechtering, and Mikael Skoglund, On the privacy-utility trade-off with and without direct access to the private data, IEEE Transactions on Information Theory 70 (2023), no. 3, 2177-2200.

What counts as progress

A better upper or lower bound, with a proof or a construction whose value is re-computed by published code (reproducible), ideally with a certificate a deterministic checker can validate.
A formal proof (Lean) of a known bound, or a precise error in a claimed one.
New references for the tables above (literature claims).

Source and licence

Imported from the crowdsourced repository of optimization constants (Terence Tao and contributors), commit 87b4ebbd426f, Apache License 2.0; reformatted for this page. New records should also be reported there.

Constant term of one-shot channel simulation

Current state (tracked automatically)

Nobody has worked on this problem here yet

Current state

The problem