Cylinder Conjecture in Geometry and Combinatorics

• Cylinder Conjecture is a multifaceted topic in geometry, combinatorics, and topology that studies the maximal arrangements of mutually tangent infinite cylinders in ℝ³.
• The approach uses algebraic and topological methods, including chirality and Ring matrices, to establish bounds with explicit constructions for N=7 and limits at N=18.
• It extends to finite field analogues, packing densities, spectral theory, and moduli spaces, offering insights into discrete configurations and isoperimetric minimization.

The Cylinder Conjecture encompasses a family of problems in geometry, combinatorics, topology, algebra, and spectral theory. Across these domains, "cylinder conjecture" typically refers to extremal, packing, or structural questions about cylinders in Euclidean space or abstract analogues. The foundational version, attributed to J.E. Littlewood, asks for the maximal number of infinite congruent cylinders in $\mathbb{R}^3$ all pairwise tangent—a question that, despite decades of study, remains only partially resolved. The terminology is also used for distinct but related conjectures including discrete analogues on finite fields, isoperimetric minimization, stable configurations in moduli spaces, and topological characterizations of configurations or morphisms. This article presents a comprehensive overview of the main forms of the Cylinder Conjecture and their contemporary status.

1. Congruent Infinite Cylinders in $\mathbb{R}^3$: The Geometry and Extremal Problem

The archetype Cylinder Conjecture asks: What is the maximum number $N$ of congruent infinite cylinders in $\mathbb{R}^3$ so that each touches all the others? More precisely, each cylinder $$C(\ell,r) = \{x \in \mathbb{R}^3 : \operatorname{dist}(x,\ell) = r\}$$ (radius $r>0$) is determined by its axis $\ell$, an affine line. Two congruent cylinders are said to touch (be tangent) if their boundaries meet but their interiors do not overlap. The touching condition reduces to $\operatorname{dist}(\ell_i, \ell_j) = 2r$ between axes.

Littlewood's heuristic predicted $N=7$ on degrees-of-freedom arguments. Classical explicit arrangements with $N=6$ exist; Bozóki, Lee, and Rónyai, via a computer-assisted algebraic solution of a 20-variable polynomial system, constructed and certified two distinct topological arrangements of $N=7$ congruent unit-radius cylinders, showing $N\geq7$ is possible (Bozóki et al., 2013). Extensions to $N=8$ have been explored, but Ambrus and Bezdek demonstrated such candidate configurations fail the tangency constraint.

Prior upper bounds included Bezdek's $N\leq24$ via geometric and angular packing arguments. The current best is $N\leq 18$, established by Koizumi (Koizumi, 24 Jun 2025) using a geometric-chirality-graph approach: the axes $\ell_i$ are described by $(w_i,v_i)$ pairs (base point, direction), and the pairwise tangency translates to a system $d(\ell_i,\ell_j)=1$. Assigning signed chiralities produces a signed complete graph, and Ramsey-theoretic arguments (no monochromatic $K_5$ in the chirality graph, Ramsey number $R(4,4)=18$) force $n\leq18$. Thus, the extremal value is bracketed by $7 \leq N \leq 18$ (Koizumi, 24 Jun 2025), with two explicit configurations for $N=7$ and no known realization for $N=8$.

2. Topological, Algebraic, and Graph-Theoretic Characterizations

Pikhitsa and Pikhitsa (Pikhitsa et al., 2017) developed a topological-combinatorial framework involving the chirality matrix (Seidel matrix) and the Ring matrix to classify line configurations associated with the axes of cylinders. The chirality matrix $P$ encodes the orientation relations between pairs of axes, and the Ring matrix $R$ records entanglement via encaging cycles. Forbidden submatrices (e.g., $K_5$, $P_{250}$) in $P$ preclude the existence of a mutually touching configuration. Their combinatorial analysis shows:

• Only two (mirror-image) topological configurations of $N=7$ congruent round cylinders exist.
• Allowing arbitrary radii, but fixed circular cross-sections, admits $N=9$, with explicit 9-knot configurations constructed numerically.
• If arbitrary smooth convex cross-sections are allowed (e.g., elliptic cylinders), the upper bound becomes $N=10$; explicit 10-knot elliptic configurations are found, and no consistent solution for $N>10$ exists in this category (Pikhitsa et al., 2017).

This work identifies essential algebraic and topological obstructions via forbidden minors in the Seidel matrix, revealing the interplay between geometry and graph theory in the structure of feasible cylinder arrangements.

3. Packing Density of Non-Parallel Infinite Cylinders

A separate branch, originating from Kuperberg's conjecture, addresses the optimal packing density of infinite, congruent, non-parallel cylinders in $\mathbb{R}^3$. Bezdek and Kuperberg established that the supremal upper density (using upper density $\delta^+$—the limsup ratio of covered to total volume in balls of radius $r$ as $r \to \infty$) for parallel axis cylinder packings equals $\pi/\sqrt{12}$, mirroring the optimal planar hexagonal disk packing (Eliyahu, 2023).

Kuperberg conjectured that the same optimal density should be attainable in the non-parallel case. This was affirmed: for any $\varepsilon>0$, there exists a non-parallel cylinder packing with lower density $\delta^- \geq \pi/6 - \varepsilon$, and there exists a non-parallel packing achieving upper density $\delta^+ = \pi/\sqrt{12}$, matching the parallel benchmark (Eliyahu, 2023). These results are achieved via careful construction of axis directions and leveraging dual circle packing correspondences. The exact supremal value of the lower density for non-parallel packings remains unresolved, with $\pi/6 \leq \delta^-_{sup} \leq \pi/\sqrt{12}$.

4. Discrete and Finite Geometry Analogs: Cylinder Sets in Finite Affine Space

The Cylinder Conjecture has a powerful discrete analogue in finite affine and projective spaces, notably over $\mathbb{F}_p^3$. The Strong Cylinder Conjecture, posed by Ball, asserts that any $p$-divisible subset $S \subseteq \mathbb{F}_p^3$ (i.e., $|S \cap H| \equiv 0 \pmod{p}$ for all affine hyperplanes $H$), with $|S| = p^2$, is a cylinder—specifically, a union of $p$ parallel lines (Kiss et al., 14 Jan 2026).

The conjecture has been completely resolved: every $p$-divisible subset of size $p^2$ in $\mathbb{F}_p^3$ must indeed be a cylinder (Kiss et al., 14 Jan 2026). The proof strategy combines linear and integer decomposition of indicator functions in the space of cylinders, reduction to orthogonality in polynomial representations, and elimination of possible multisets that violate the cylinder structure.

Generalizations for arbitrary finite fields, higher dimensions, and divisible codes have been explored, leading to analogously formulated "generalized cylinder conjectures" for divisible linear codes. Positive results hold for $q=2,3,5,7$ and related projective geometry settings, with explicit counterexamples (via field-reduction) for non-prime fields (Kurz et al., 2020).

5. Cylinder Conjectures in Moduli and Dynamics: Masur–Veech Volumes

In Teichmüller dynamics and algebraic geometry, the Cylinder Conjecture posits specific asymptotic contributions of $k$-cylinder square-tiled surfaces to the Masur–Veech volumes of strata. For the stratum $\mathcal{H}(\mu)$, the conjecture is that the contribution from $k$-cylinder surfaces decays as $d^{-k}$, with $d = \dim_{\mathbb{C}} \mathcal{H}(\mu)$ (Delecroix et al., 2019).

This has been verified for $k=1$: the explicit formula for the one-cylinder contribution $c_1(\mu)$, combined with asymptotics for the total volume, shows that $P_1(\mu) = c_1(\mu)/\Vol(\mathcal{H}(\mu)) \sim 1/d$ as $d \to \infty$. This result forms the cornerstone for conjectural expansions involving higher $k$, with recursive and equidistribution techniques supporting the expected order $O(d^{-k})$ (Delecroix et al., 2019).

6. Other Domains: Spectral, Isoperimetric, and Topological Cylinder Conjectures

Spectral Geometry

Pólya's conjecture on the eigenvalues of the Laplacian extends to cylindrical domains and infinite cylinders, with results characterizing the range of geometric parameters (radius, height) for which the Pólya or Li–Yau eigenvalue inequalities hold (Freitas et al., 4 Jun 2025, Guo et al., 21 Nov 2025). On the infinite cylinder $\mathbb{S}^1 \times \mathbb{R}$, the conjecture holds for geodesic disks up to radius $R_1 \approx 3.76085$ and for strips $S_h$ in explicitly characterized intervals for $h$ (Freitas et al., 4 Jun 2025). These results refine the classical spectral bounds and clarify the role of isoperimetric domains on cylinders.

Gaussian Isoperimetry

Barthe's Cylinder Conjecture in the context of minimal Gaussian perimeter asserts that the isoperimetric minimizers among symmetric sets are round cylinders of the form $r S^k \times \mathbb{R}^{n-k}$. This has been resolved up to a single exceptional family (mean-convex, negative–$\lambda$ solutions) (Heilman, 2022). The main classification theorem shows, under all but this exceptional sign condition, minimizers are convex or round cylinders, sharpening Gaussian isoperimetric theory.

Categorical and Model-Theoretic Cylinder Conjectures

In higher category theory, Joyal's Cylinder Conjecture addresses homotopical properties of the category $\mathrm{Cyl}(A,B)$ of cylinders between simplicial sets $A,B$. Campbell proved that a cylinder is fibrant if and only if the canonical map $X \to A\star B$ is an inner fibration, establishing a key equivalence between model structures and inner fibration conditions (Campbell, 2019). This has applications in covariant equivalences and advanced model category theory.

7. Counterexamples and New Classes: Non-Cyclic Cylinder Decompositions

A recent development is the construction of infinite families of counterexamples to the Apisa–Wright Cylinder Conjecture, which posited that all 1-cylinder pillowcase-tiled surfaces (branched covers of the pillowcase) are cyclic covers. Explicitly, for odd $n \geq 5$, one can construct 1-cylinder covers whose deck group is $A_n$, thus non-cyclic (Abdalla et al., 21 Oct 2025). This demonstrates the flexibility of cylinder decompositions in flat geometry and has implications for classification in Teichmüller dynamics.

Table 1: Summary of the Main Cylinder Conjecture Types

Domain	Conjecture Statement/Result	Status & Best Bounds	Reference
$\mathbb{R}^3$ cylinders	Maximal $N$ pairwise-touching congruent cylinders	$7 \leq N \leq 18$; $N=7$ explicitly	(Koizumi, 24 Jun 2025, Bozóki et al., 2013)
Cylinder packing density	Max non-parallel congruent packing, $\delta^+$	$\delta^+ = \pi/\sqrt{12}$	(Eliyahu, 2023)
Finite field geometry	$p$-divisible set of size $p^2$ is a cylinder	True for all $p$, complete proof	(Kiss et al., 14 Jan 2026)
Moduli spaces	$k$-cylinder surfaces contribute $O(d^{-k})$	$k=1$ proved, $k>1$ partial evidence	(Delecroix et al., 2019)
Pillowcase-tiled surfaces	1-cylinder $\Rightarrow$ cyclic cover	False: infinite non-cyclic examples	(Abdalla et al., 21 Oct 2025)
Gaussian isoperimetry	Symmetric minimizer is a round cylinder	True except for mean-convex/neg $\lambda$	(Heilman, 2022)
Spectral/isoperimetric	Pólya/Li–Yau on cylinder, disk, strip	Complete for all geometric parameter ranges	(Freitas et al., 4 Jun 2025, Guo et al., 21 Nov 2025)

8. Open Problems and Future Directions

Substantial gaps and challenges remain:

• Determining the exact maximal $N$ for mutually touching congruent infinite cylinders in $\mathbb{R}^3$; the interval $7 \leq N \leq 18$ persists (Koizumi, 24 Jun 2025).
• Classifying all topological types of feasible arrangements when radii or cross-sections are variable, and understanding limits of continuous deformation between types (Pikhitsa et al., 2017).
• Achieving explicit combinatorial formulas and asymptotics for $k$-cylinder contributions to Masur–Veech volumes for $k>1$ (Delecroix et al., 2019).
• Extending the arithmetic and combinatorial classification of divisible sets for non-prime fields, higher-dimensional affine/projective spaces, and more general code structures (Kurz et al., 2020).
• Completing the classification of Gaussian isoperimetric minimizers in the exceptional mean-convex, negative–$\lambda$ case (Heilman, 2022).

The Cylinder Conjecture thus interfaces with quantitative and structural questions in discrete geometry, algebra, topology, analysis, and dynamics. Advances in each direction often stimulate new cross-disciplinary results and conjectures.