Sharp Isoperimetric Inequalities on Hamming Cube

Updated 2 March 2026

The paper establishes sharp isoperimetric inequalities on the Hamming cube by precisely characterizing minimal boundary sizes for subcubes under edge, vertex, and Lβ functionals.
It employs combinatorial, analytic, and probabilistic techniques, including Bellman function methods and computer-assisted proofs, to derive tight constants and stability results.
The results directly lead to optimal Poincaré-type bounds and have broad implications in extremal combinatorics, Boolean analysis, and information theory.

Sharp isoperimetric inequalities on the Hamming cube characterize the minimal possible boundary size of subsets of $\{0,1\}^n$ under various notions (edge, vertex, and interpolating $L^\beta$ functionals), with sharp constants and precise stability analyses. This area integrates combinatorial, analytic, and probabilistic methods, connects with Poincaré and functional inequalities, and has deep implications in extremal combinatorics, information theory, and Boolean analysis.

1. Foundational Results and Definitions

The $n$ -dimensional Hamming cube $Q_n$ is the graph with vertex set $\{0,1\}^n$ ; two vertices are adjacent if their Hamming distance is one. For a subset $A \subset \{0,1\}^n$ , the edge-boundary $\partial_E A$ is the set of edges with one endpoint in $A$ and one in $A^c$ . The size $|\partial_E A|$ measures the "surface area" of $L^\beta$ 0 in the cube.

Classical Edge-Isoperimetric Inequality

The foundational result (Harper–Lindsey–Bernstein–Hart) states:

$L^\beta$ 1

with equality if and only if $L^\beta$ 2 is a coordinate subcube, i.e., the set of all vectors with certain coordinates fixed. The average out-degree $L^\beta$ 3 is minimized for subcubes, exhibiting a sharp combinatorial profile (Ellis, 2013, Rashtchian et al., 2019).

Generalizations

Vertex-boundary: The set of vertices outside $L^\beta$ 4 adjacent to $L^\beta$ 5; associated inequalities are sharp for Hamming balls and generalized Hamming balls (Keevash et al., 2018).
$L^\beta$ 6 Interpolation: For $L^\beta$ 7, the sequence of functionals

$L^\beta$ 8

interpolates between vertex ( $L^\beta$ 9) and edge ( $n$ 0) boundary, with critical behavior as $n$ 1 approaches $n$ 2 (Durcik et al., 2024, Durcik et al., 24 Feb 2026, Kahn et al., 2019).

2. Sharp Stability, Robustness, and Characterization

Beyond minimality, sharp isoperimetric inequalities now quantify stability: how sets with near-minimal boundary must closely resemble extremal configurations.

Stability at the edge-boundary: If $n$ 3 satisfies $n$ 4 for small $n$ 5, then $n$ 6 can be made into a subcube by at most $n$ 7 additions and deletions, and no improvement is possible in general (Ellis, 2013, Ellis et al., 2017).
Distance penalty trade-off: For $n$ 8 and $n$ 9 (distance from any subcube), the boundary is forced up by

$Q_n$ 0

sharply matching explicit constructions at dyadic $Q_n$ 1 (Ellis, 2013).

Spectral and probabilistic structure: Samorodnitsky's functional extension shows that near-isoperimetric sets are nearly log-regular in degree and spectrally close to subcubes, controlling eigenvalue concentration and mean exit times for random walks (Samorodnitsky, 2012).
Vertex isoperimetry and balls: For the vertex-boundary, near-minimality forces closeness (in symmetric-difference) to generalized Hamming balls, with sharp explicit dependence on the deviation parameter (Keevash et al., 2018).

3. Mixed $Q_n$ 2 Inequalities and the Critical Exponent

Recent work has advanced mixed boundary inequalities with exponents $Q_n$ 3, interpolating between edge and vertex forms and seeking dimension-free sharp constants.

Kahn–Park Inequality ( $Q_n$ 4):

$Q_n$ 5

with equality for codimension-1 and codimension-2 subcubes; the exponent is best possible (Kahn et al., 2019).

Critical exponent breakthrough: For $Q_n$ 6, the sharp inequality

$Q_n$ 7

is attained precisely by subcubes, established via a Bellman-type two-point induction, convex analysis, and certified by rigorous computer-assisted interval arithmetic (Durcik et al., 2024, Durcik et al., 24 Feb 2026). At $Q_n$ 8 (the "critical exponent"), further intricate analysis yields:

$Q_n$ 9

and at small $\{0,1\}^n$ 0,

$\{0,1\}^n$ 1

(Durcik et al., 2024, Durcik et al., 24 Feb 2026).

Envelope construction: The Bellman envelope approach constructs candidate extremal functions via piecewise combinations of $\{0,1\}^n$ 2, cubic interpolants, and a rescaled Gaussian isoperimetric profile. Rigorous numerical methods confirm sharp constant regions and show that this envelope is optimal up to the threshold $\{0,1\}^n$ 3 (Durcik et al., 2024).

4. Extensions, Poincaré Inequalities, and Applications

Functional and Poincaré inequalities: The sharp $\{0,1\}^n$ 4 isoperimetric inequalities directly yield optimal $\{0,1\}^n$ 5 Poincaré-type bounds for Boolean functions. For $\{0,1\}^n$ 6, one obtains

$\{0,1\}^n$ 7

and for $\{0,1\}^n$ 8 an improved constant:

$\{0,1\}^n$ 9

(Durcik et al., 2024, Durcik et al., 24 Feb 2026).

Cube partition inequalities: Progress and eventual resolution of the Kahn–Park cube-partition conjecture is built on the critical exponent inequality, with implications for partition problems and maximal independent set counting on $A \subset \{0,1\}^n$ 0 (Kahn et al., 2019, Durcik et al., 2024, Durcik et al., 24 Feb 2026).
Information theory link: The $A \subset \{0,1\}^n$ 1 isoperimetric result provides the sharp low-noise limit for Boolean channel Hellinger conjecture, linking boundary measures to expected sensitivity in the limit of vanishing noise (Durcik et al., 24 Feb 2026).
Edge isoperimetry for powers and related graphs: Sharp (up to constants) generalizations exist for the $A \subset \{0,1\}^n$ 2th power $A \subset \{0,1\}^n$ 3 and Kleitman-West graphs, with the classical $A \subset \{0,1\}^n$ 4 case being uniquely dimension-free and sharp in terms of initial segments and subcubes (Rashtchian et al., 2019).

5. Methodological Framework and Proof Techniques

Compression and shifting: Iterative "compression" or "shifting" sequences, which map arbitrary sets to down-sets and then left-compressed forms, reduce to structured extremals without increasing the boundary, a method tracing to Harper and later refinements (Rashtchian et al., 2019, Ellis, 2013).
Bellman function and two-point induction: Recent advances construct explicit piecewise Bellman functions, verifying via two-point inequalities on $A \subset \{0,1\}^n$ 5 that certain envelopes are sharp lower bounds for the isoperimetric profile (Durcik et al., 2024, Durcik et al., 24 Feb 2026).
Entropic and influence decompositions: Talagrand's and Samorodnitsky's arguments connect influences, entropy, and Fourier-analytic profiles, quantifying that almost-minimal boundary necessarily forces the measure to be near a subcube in various senses (Samorodnitsky, 2012, Eldan et al., 2022, Ellis, 2013).
Computer-assisted interval arithmetic: Positivity of two-point functions and explicit sharpness regions are certified using dyadic partitioning and rigorous interval methods (Arb/FLINT libraries), allowing extension to exponents close to the critical value (Durcik et al., 2024).

6. Extremal Configurations, Open Problems, and Further Directions

Extremals: Subcubes are the unique extremals for all sharp inequalities discussed for the edge boundary and $A \subset \{0,1\}^n$ 6 functionals with $A \subset \{0,1\}^n$ 7; for the vertex-boundary, Hamming balls and "generalized balls" appear (Keevash et al., 2018).
Sharpness at dyadic distances: Certain swap-constructions at dyadic distances from subcubes precisely realize the lower bounds with prescribed $A \subset \{0,1\}^n$ 8 penalty terms (Ellis, 2013).
Extensions to biased and product measures: Analogues of sharp stability and extremality exist for the $A \subset \{0,1\}^n$ 9-biased measure on $\partial_E A$ 0, with subcubes and lex-initial segments again providing equality, and stability quantified in terms of the bias (Ellis et al., 2017).
Open problems: Exact edge-isoperimetric profiles for higher-order Hamming cube powers ( $\partial_E A$ 1) remain unknown beyond constant factors. The extension of computer-assisted Bellman approaches to these and other graphs is a promising direction (Rashtchian et al., 2019, Durcik et al., 2024).
Broader impact: Sharp cube isoperimetry now underpins extremal combinatorics (stability in Erdős–Ko–Rado, matching and intersection theorems), random walks, threshold phenomena, Boolean function analysis, and information theory, providing both structure and quantitative robustness in high-dimensional discrete settings (Ellis et al., 2017, Eldan et al., 2022).

Table: Exponents and Extremals for Main Sharp Isoperimetric Inequalities

Exponent ( $\partial_E A$ 2)	Inequality Form	Extremal Sets
$\partial_E A$ 3	$\partial_E A$ 4	Subcubes
$\partial_E A$ 5	$\partial_E A$ 6	Codim-1, codim-2 subcubes
$\partial_E A$ 7	$\partial_E A$ 8	All subcubes
$\partial_E A$ 9 (critical)	$A$ 0	Subcubes (sharp), $A$ 1 const.

These advances yield a comprehensive structural and quantitative understanding of boundary-minimization phenomena on the discrete cube, with the Bellman-envelope program providing the central analytic machinery, and combinatorial stability theory ensuring that near-extremality robustly confines the structure to explicit families. The sharp tradeoffs and stability bounds remain a core engine for further developments in discrete analysis, spectral graph theory, and high-dimensional probability (Ellis, 2013, Kahn et al., 2019, Durcik et al., 2024, Durcik et al., 24 Feb 2026).