A spectral correlation inequality for increasing Boolean functions
Pith reviewed 2026-06-27 16:41 UTC · model grok-4.3
The pith
Increasing Boolean functions satisfy Cov(f,g) ≥ 2 ∑_{S≠∅} |S| ˆf(S)^2 ˆg(S)^2
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
For all increasing Boolean functions f and g from the hypercube to {0,1}, Cov(f,g) ≥ 2 ∑_{S≠∅} |S| ˆf(S)^2 ˆg(S)^2. The proof proceeds by applying the reverse Bonami-Beckner inequality to the pair of increasing functions and then invoking Young's inequality. The paper also determines the exact optimal constant c_ρ,n in the pointwise inequality ⟨f, T_ρ g⟩ ≥ c_ρ,n ‖f * g‖_2^2, which equals 1 for ρ ≤ 1/2, (2(1-ρ))^n for 1/2 < ρ < 1, and 0 at ρ=1; integrating this yields the improved factor 4(n+1)/(2n).
What carries the argument
Reverse Bonami-Beckner inequality applied to increasing functions, combined with Young's convolution inequality to produce the spectral lower bound on covariance.
Load-bearing premise
The reverse Bonami-Beckner inequality applies directly to pairs of increasing functions in the form required for the spectral bound.
What would settle it
An explicit pair of increasing functions f and g on {0,1}^n for which the computed covariance falls below twice the indicated spectral sum would disprove the inequality.
read the original abstract
Talagrand's correlation inequality provides a quantitative strengthening of the Harris--Kleitman inequality for increasing Boolean functions. Motivated by a Fourier-analytic conjecture of Friedgut, Kahn, Kalai, and Keller, we prove that $$ \mathrm{Cov}(f,g)\ge 2\sum_{S\neq\emptyset}|S|\hat f(S)^2\hat g(S)^2 $$ holds for all increasing Boolean functions $f,g:\{0,1\}^n\to\{0,1\}$. The proof combines the reverse Bonami--Beckner inequality with Young's convolution inequality. We also establish a sharp pointwise inequality: for every $n\ge1$, every $0\le\rho\le1$, and every $f,g:\{0,1\}^n\to[0,1]$, the optimal constant $c_{\rho,n}$ for which $$ \left\langle f,T_\rho g \right\rangle\ge c_{\rho,n}\|f*g\|_2^2 $$ holds for all such $f,g$ is $1$ for $0\le\rho\le1/2$, $(2(1-\rho))^n$ for $1/2<\rho<1$, and $0$ for $\rho=1$. Integrating this pointwise inequality yields, for $n\ge1$, the slightly improved bound $$ \mathrm{Cov}(f,g)\ge 4\cdot\frac{n+1}{2n}\sum_{S\neq\emptyset}|S|\hat f(S)^2\hat g(S)^2. $$
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves that Cov(f,g) ≥ 2 ∑_{S≠∅} |S| ˆf(S)^2 ˆg(S)^2 for all increasing Boolean functions f,g : {0,1}^n → {0,1}. The proof combines the reverse Bonami-Beckner inequality with Young's convolution inequality. It also establishes the sharp pointwise inequality ⟨f, T_ρ g⟩ ≥ c_{ρ,n} ||f*g||_2^2 for f,g : {0,1}^n → [0,1], where c_{ρ,n} equals 1 for 0≤ρ≤1/2, (2(1-ρ))^n for 1/2<ρ<1, and 0 for ρ=1; integrating this yields the improved bound Cov(f,g) ≥ 4·(n+1)/(2n) ∑_{S≠∅} |S| ˆf(S)^2 ˆg(S)^2.
Significance. If the central inequality holds, the result supplies a Fourier-analytic strengthening of Talagrand's correlation inequality and resolves a conjecture of Friedgut-Kahn-Kalai-Keller in the monotone setting. The sharpness of the pointwise inequality (optimal in each ρ regime) is a clear strength, as is the explicit integration to a slightly better constant. The reliance on standard external tools (reverse Bonami-Beckner and Young) rather than ad-hoc constructions is also a positive feature.
minor comments (1)
- The abstract states the main inequality for range {0,1} but the pointwise inequality is stated for [0,1]-valued functions; a brief remark clarifying whether the main proof extends verbatim to the [0,1] case would aid readability.
Simulated Author's Rebuttal
We thank the referee for their positive assessment of the manuscript, including the recognition of the sharpness of the pointwise inequality and the explicit integration yielding the improved constant, and for recommending acceptance.
Circularity Check
No circularity; derivation relies on external inequalities without self-reference or reduction to inputs
full rationale
The central claim is obtained by combining the reverse Bonami-Beckner inequality with Young's convolution inequality, both standard external tools. The increasing assumption ensures non-negativity of f and g but does not create a self-definitional loop or fitted prediction. The pointwise inequality is separately established and integrated to yield a slight improvement, but neither step reduces by construction to the target spectral form. No self-citations appear in the load-bearing steps, and the derivation remains self-contained against external benchmarks.
Axiom & Free-Parameter Ledger
axioms (2)
- standard math Fourier analysis decomposes Boolean functions into Walsh-Fourier basis with the usual inner product and Parseval identity.
- standard math The reverse Bonami-Beckner inequality and Young's convolution inequality hold in the stated forms for functions on the hypercube.
Forward citations
Cited by 1 Pith paper
-
The sharp diagonal spectral correlation inequality on the discrete cube
Proves Cov(f,g) ≥ 4 ∑_{∅≠S} |S| ˆf(S)^2 ˆg(S)^2 for increasing Boolean f,g on {0,1}^n, with the factor 4 sharp and all equality cases determined.
Reference graph
Works this paper leans on
-
[1]
Inequalities in Fourier analysis.Ann
William Beckner. Inequalities in Fourier analysis.Ann. of Math. (2), 102(1):159–182, 1975
1975
-
[2]
David Beltran, Paata Ivanisvili, Jos´ e Madrid, and Lekha Patil. Optimal Young’s convolu- tions inequality and its reverse form on the hypercube.arXiv preprint arXiv:2507.06115, 2025
arXiv 2025
-
[3]
´Etude des coefficients de Fourier des fonctions deL p(G).Ann
Aline Bonami. ´Etude des coefficients de Fourier des fonctions deL p(G).Ann. Inst. Fourier (Grenoble), 20(fasc. 2):335–402 (1971), 1970
1971
-
[4]
Fan Chang and Yu Chen. Talagrand-type correlation inequalities for submodular and supermodular functions on the hypercube.arXiv preprint arXiv:2510.22307, 2025
Pith/arXiv arXiv 2025
-
[5]
Chv´ atal
V. Chv´ atal. Intersecting families of edges in hypergraphs having the hereditary property. InHypergraph Seminar (Proc. First Working Sem., Ohio State Univ., Columbus, Ohio, 1972; dedicated to Arnold Ross), Lecture Notes in Math., Vol. 411, pages 61–66. Springer, Berlin-New York, 1974
1972
-
[6]
Chv´ atal’s conjecture and corre- lation inequalities.J
Ehud Friedgut, Jeff Kahn, Gil Kalai, and Nathan Keller. Chv´ atal’s conjecture and corre- lation inequalities.J. Combin. Theory Ser. A, 156:22–43, 2018
2018
-
[7]
T. E. Harris. A lower bound for the critical probability in a certain percolation process. Proc. Cambridge Philos. Soc., 56:13–20, 1960
1960
-
[8]
On the correlation of increasing families
Gil Kalai, Nathan Keller, and Elchanan Mossel. On the correlation of increasing families. J. Combin. Theory Ser. A, 144:250–276, 2016
2016
-
[9]
Lower bound on the correlation between monotone families in the average case.Adv
Nathan Keller. Lower bound on the correlation between monotone families in the average case.Adv. in Appl. Math., 43(1):31–45, 2009
2009
-
[10]
Biased halfspaces, noise sensitivity, and local Chernoff inequalities.Discrete Anal., 2019
Nathan Keller and Ohad Klein. Biased halfspaces, noise sensitivity, and local Chernoff inequalities.Discrete Anal., 2019. Paper No. 13, 50 pp
2019
-
[11]
Geometric influences II: correlation inequalities and noise sensitivity.Ann
Nathan Keller, Elchanan Mossel, and Arnab Sen. Geometric influences II: correlation inequalities and noise sensitivity.Ann. Inst. Henri Poincar´ e Probab. Stat., 50(4):1121– 1139, 2014
2014
-
[12]
Kleitman
Daniel J. Kleitman. Families of non-disjoint subsets.J. Combinatorial Theory, 1:153–155, 1966
1966
-
[13]
Cambridge University Press, New York, 2014
Ryan O’Donnell.Analysis of Boolean functions. Cambridge University Press, New York, 2014
2014
-
[14]
How much are increasing sets positively correlated?Combinatorica, 16(2):243–258, 1996
Michel Talagrand. How much are increasing sets positively correlated?Combinatorica, 16(2):243–258, 1996. 13
1996
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.