CylinderPlane Geometry & Applications

• CylinderPlane is a geometric construct where cylindrical structures interact with planes to yield unique sections, boundaries, and coordinate representations.
• It applies to analytic models in hyperplane sections, vortex flow, capillary and mean-curvature problems, and cutting-edge 3D-aware neural feature mapping.
• Research in CylinderPlane spans Euclidean geometry, combinatorial duals, arithmetic decompositions, and algebraic interpretations, highlighting its broad impact.

CylinderPlane denotes a family of constructions in which cylindrical geometry is constrained by, intersected with, or parameterized against a plane. In the literature represented here, the theme ranges from central hyperplane sections of generalized cylinders and oblique cylinder–plane cuts to vortex interaction with plane walls, capillary and mean-curvature problems in slabs and right cylinders, and, in a recent machine-learning usage, a cylindrical-coordinate implicit representation for 3D-aware image generation (Dirksen, 2016, Moura et al., 2012, Jia et al., 21 Jul 2025). The common structure is that a cylinder supplies axial or azimuthal organization, while a plane supplies a section, boundary, asymptotic model, or coordinate chart.

1. Euclidean section geometry and developed cuts

A standard higher-dimensional model is the generalized cylinder

$$Z=\left[-\frac12,\frac12\right]^n\times rB_2^m\subset\mathbb R^{n+m},$$

where $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$. For $n=1,m=2$, this is the usual $3$-dimensional circular cylinder $[-\tfrac12,\tfrac12]\times rB_2^2$. Central sections are taken by hyperplanes

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$

and the exact section-volume formula factorizes into sinc terms from the cube factor and a normalized Bessel term from the ball factor. In particular,

$j_{1/2}(s)=\frac{\sin s}{s},$

so the one-dimensional ball factor recovers the classical cube-slicing term (Dirksen, 2016).

The ordinary $3$-dimensional cylinder already exhibits nontrivial cylinder–plane behavior. For

$Z=\left[-\frac12,\frac12\right]\times rB_2^2$

and a normal

$$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$

the section area $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$0 has three regimes: an ellipse / disk-type section, a truncated ellipse, and a rectangle. The maximal section is a rectangle if

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$1

and a truncated ellipse if

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$2

A common misconception is therefore that an oblique plane cut of a cylinder is always an ellipse; in this model, truncated ellipses and rectangles occur as genuine extremal sections (Dirksen, 2016).

For an oblique circular cylinder in $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$3, Finch defines the solid as the convex hull of the two unit disks

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$4

with $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$5, $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$6. The defining planes are $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$7 and $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$8, and the ruling direction is $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$9. The volume is

$n=1,m=2$0

The total area is

$n=1,m=2$1

and the mean width is

$n=1,m=2$2

The lateral area is interpreted by a plane normal to $n=1,m=2$3: the perpendicular section is an ellipse with semi-major axis $n=1,m=2$4 and semi-minor axis

$n=1,m=2$5

This is a direct cylinder–plane reduction of a surface-area computation (Finch, 2012).

A fabrication-oriented variant studies two intersecting cylinders of equal radius whose surface patch is bounded on all sides by planes orthogonal to the primary cylinder axis. The developed interior cut and the developed exterior plane–cylinder cuts are given by closed-form sinusoidal formulas in the unwrapped variable $n=1,m=2$6, so the flattened boundaries are generally not arcs or straight lines (May, 2013). A different geometric interpretation appears in the Kepler problem: a noncircular negative-energy orbit lies in its orbital plane and also on exactly two cylinders belonging to the ellipse, with axes passing through the ellipse center and directed by

$n=1,m=2$7

In that setting, the orbital ellipse is realized directly as a plane–cylinder intersection rather than only as a plane–cone intersection (McConnell, 2012).

2. Vortex flow and rotating-cylinder interaction with plane walls

In ideal fluid mechanics, the configuration of a circular cylinder above a plane wall is a doubly connected domain. The physical plane is the upper half-plane above $n=1,m=2$8, the cylinder has center

$n=1,m=2$9

and boundary

$3$0

and the imposed flow consists of a uniform stream of speed $3$1 and a point vortex of circulation $3$2 at $3$3. The fluid is assumed inviscid, incompressible, and irrotational everywhere except at the point vortex, so it admits a complex potential

$3$4

Because the wall and cylinder generate an infinite image hierarchy, the exact solution is constructed by conformal mapping to an annulus and the Schottky–Klein prime function. The resulting total potential is

$3$5

The sample streamline plot for

$3$6

shows a closed recirculation bubble in front of the cylinder, streamlines deflected upward around the cylinder, and recovery to uniform flow far away. This upstream recirculation contrasts with the classical Föppl picture for a cylinder in an unbounded domain, where stationary vortices occur as a symmetric pair behind the cylinder (Moura et al., 2012).

A three-dimensional viscous analogue replaces the singular line vortex by a vertical rotating cylinder of radius $3$7 meeting the horizontal plane $3$8 at right angle. In cylindrical coordinates $3$9, the stationary axisymmetric Navier–Stokes problem is posed in

$[-\tfrac12,\tfrac12]\times rB_2^2$0

with boundary conditions

$[-\tfrac12,\tfrac12]\times rB_2^2$1

and

$[-\tfrac12,\tfrac12]\times rB_2^2$2

The main analytical difficulty is the corner $[-\tfrac12,\tfrac12]\times rB_2^2$3, where no-slip on the plane and nonzero prescribed swirl on the cylinder are incompatible. Weak existence is proved first by a Hopf extension localized near the cylinder and then by a contradiction argument of Leray–Schauder type for the original boundary data (Han et al., 2016).

The numerical solutions exhibit the wall-induced secondary circulation known as boundary layer pumping. Near the plane, viscosity suppresses swirl, centrifugal support weakens, radial inflow develops along the wall, and incompressibility converts that inflow into an updraft near the cylinder. With uniform cylinder rotation the flow is single-celled. With height-dependent rotation on the cylinder, the model produces two-celled vortex structures. Simulations at

$[-\tfrac12,\tfrac12]\times rB_2^2$4

with

$[-\tfrac12,\tfrac12]\times rB_2^2$5

approach the line-vortex/plane behavior as $[-\tfrac12,\tfrac12]\times rB_2^2$6 (Han et al., 2016).

3. Mean-curvature flow, capillarity, and cylindrical rigidity between planes

For volume-preserving mean curvature flow in a slab,

$[-\tfrac12,\tfrac12]\times rB_2^2$7

the stationary reference hypersurface is the cylinder

$[-\tfrac12,\tfrac12]\times rB_2^2$8

Writing a nearby hypersurface as a normal graph $[-\tfrac12,\tfrac12]\times rB_2^2$9, the evolution becomes

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$0

If

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$1

then the linearization has an $$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$2-dimensional center space corresponding exactly to nearby cylinders, and every sufficiently small

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$3

perturbation exists for all time and converges exponentially in $$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$4 to a nearby cylinder. The perturbation need not be axially symmetric (Hartley, 2012).

For capillary surfaces in a right solid cylinder $$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$5 with boundary cylinder $$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$6 and base plane

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$7

the contact-angle condition is

$$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$8

along $$H_a^t=\{x\in\mathbb R^{n+m}:\langle a,x\rangle=t\},$$9. Alexandrov reflection with horizontal planes $j_{1/2}(s)=\frac{\sin s}{s},$0 and, in the circular case, vertical planes through the axis yields strong symmetry and classification statements. If a compact embedded capillary surface inside $j_{1/2}(s)=\frac{\sin s}{s},$1 is minimal and its boundary is a graph on $j_{1/2}(s)=\frac{\sin s}{s},$2, then the surface is a horizontal planar domain. For complete embedded minimal capillary surfaces outside the cylinder, a planar end parallel to the cylinder forces the whole surface to be part of a plane parallel to $j_{1/2}(s)=\frac{\sin s}{s},$3, while, in the circular case, a catenoidal end forces the surface to be part of a catenoid with axis $j_{1/2}(s)=\frac{\sin s}{s},$4 (López et al., 2014).

A translator version of cylinder–plane rigidity is given by the characterization of the grim reaper cylinder. If a connected properly embedded translating soliton in $j_{1/2}(s)=\frac{\sin s}{s},$5, with uniformly bounded genus on compact sets, is $j_{1/2}(s)=\frac{\sin s}{s},$6-asymptotic outside a cylinder to two half-planes, then exactly one of the following holds: both half-planes lie in the same vertical plane and the surface is that plane, or the half-planes lie in distinct parallel planes and the surface is the grim reaper cylinder. In the nonplanar case, the slab width is forced to be $j_{1/2}(s)=\frac{\sin s}{s},$7, the surface is symmetric about

$j_{1/2}(s)=\frac{\sin s}{s},$8

and the final classification is obtained from the vanishing of the quotient $j_{1/2}(s)=\frac{\sin s}{s},$9 (Martin et al., 2015).

At a more global level, connected, nonsingular, analytic, closed, globally subanalytic CMC surfaces in $3$0 are only a plane, a round sphere, or a right circular cylinder. The proof eliminates unduloid asymptotics by combining exponential convergence of ends with the fact that globally subanalytic functions cannot track genuine exponential decay without eventually vanishing, and it eliminates catenoid-type minimal ends by a polynomial-growth obstruction to logarithmic inversion (Barbosa et al., 2014).

4. Classification by sections, slabs, and end behavior

Plane sections can themselves force cylindricality. If

$3$1

is a complete, connected, $3$2 immersion with the central cross-cut property and some plane in general position cuts the surface along a clean figure-8, then $3$3 is a central cylinder. The planar mechanism is sharp: a clean central loop with even rotation number $3$4 can never be central unless it passes through its center exactly twice and $3$5. In the figure-8 case, the center is pinned to the unique simple double-point, and the resulting central curve is forced to be a line segment, which yields a local and then global cylinder (Solomon, 2015).

In convex cylinders

$3$6

where $3$7 is a convex body in a hyperplane, the isoperimetric geometry is controlled by the base $3$8. For the right-cylinder case $3$9, there exists $Z=\left[-\frac12,\frac12\right]\times rB_2^2$0 such that every isoperimetric region of volume $Z=\left[-\frac12,\frac12\right]\times rB_2^2$1 is a slab

$Z=\left[-\frac12,\frac12\right]\times rB_2^2$2

and

$Z=\left[-\frac12,\frac12\right]\times rB_2^2$3

For the half-cylinder $Z=\left[-\frac12,\frac12\right]\times rB_2^2$4, large-volume minimizers are half-slabs $Z=\left[-\frac12,\frac12\right]\times rB_2^2$5 and

$Z=\left[-\frac12,\frac12\right]\times rB_2^2$6

for $Z=\left[-\frac12,\frac12\right]\times rB_2^2$7. In cylindrically bounded convex bodies asymptotic to $Z=\left[-\frac12,\frac12\right]\times rB_2^2$8, translated large-volume minimizers converge to half-slabs in the asymptotic half-cylinder and their free boundaries converge to flat sets

$Z=\left[-\frac12,\frac12\right]\times rB_2^2$9

A plausible implication is that, in this regime, the full unbounded geometry is effectively reduced to planar caps over the base (Ritoré et al., 2014).

For complete Riemannian surfaces without conjugate points, optimal growth conditions separate plane and cylinder behavior. On $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$0,

$$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$1

and equality holds if and only if the metric is flat. On $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$2, if both ends have subquadratic area growth, equivalently both ends open less than linearly in terms of the shortest noncontractible loop length, then the metric is flat. The proofs adapt Hopf’s method to metric balls in the plane and horocyclic exhaustions in the cylinder (Bangert et al., 2012).

5. Combinatorial, arithmetic, and birational reinterpretations

In graph theory, the planar notion of a rectangular dual extends to the cylinder. A cylindrical rectangular dual is a contact representation on a flat cylinder by interior-disjoint rectangles such that no four rectangles share a point and the union of all rectangles forms a strip from the left to the right side of the fundamental polygon. A graph embedded on the cylinder admits such a dual if and only if it is a properly triangulated cylindrical (PTC) graph. This is the exact cylinder analogue of the properly triangulated planar characterization, but it allows non-contractible loops, non-contractible parallel edges, and non-contractible separating triangles, and it yields a linear-time test and construction (Biedl et al., 8 Jun 2025).

In $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$3, the term acquires an arithmetic meaning. A set is $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$4-divisible if every affine plane intersects it in $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$5 points. The paper proves that every $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$6-divisible multiset lies in the $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$7-linear span of cylinder type multisets and, equivalently, of differences of parallel lines

$$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$8

For total weight $$a=(\sqrt{1-\alpha^2},\alpha,0),\qquad \alpha\in[0,1],$$9, one has for any affine plane $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$00,

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$01

so $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$02 is a $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$03-linear combination of cylinders. Here a plane is itself a cylinder because it is the union of $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$04 pairwise disjoint parallel lines (Kiss et al., 14 Jan 2026).

In birational geometry, a cylinder on a smooth rational surface means a Zariski-open subset

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$05

for an affine curve $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$06. If $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$07 is an ample $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$08-divisor, an $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$09-polar cylinder is one with

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$10

For blow-ups of $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$11 at points in general position, if

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$12

then $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$13 does not contain $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$14-polar cylinders. The threshold is sharp in the sense that examples with

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$15

do contain $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$16-polar cylinders, while the general-position hypothesis cannot be dropped (Cheltsov, 2016). A common misconception is therefore that “cylinder” in algebraic geometry refers to an embedded Euclidean cylinder; here it refers instead to an open subset of product type $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$17.

6. CylinderPlane as a cylindrical implicit representation in 3D-aware generation

In a recent and title-level usage, CylinderPlane is an implicit $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$18-dimensional feature representation built on the cylindrical coordinate system

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$19

rather than the Cartesian one. The representation uses three feature planes: the cylindrical plane $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$20, the circular plane $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$21, and the rectangular plane $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$22. For a $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$23-dimensional point with cylindrical coordinates $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$24, the feature lookup is

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$25

The stated motivation is to separate features at different azimuth angles and thereby eliminate feature ambiguity in symmetric regions (Jia et al., 21 Jul 2025).

The method further introduces a nested cylinder representation

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$26

so that multiple cylinders at different scales can be composited. The paper argues that the combination of cylinders with different resolutions captures more critical locations and multi-scale features, facilitates fine detail learning, and improves robustness to different resolutions. Because cylindrical coordinates introduce a periodic seam at $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$27 and $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$28, the method adds CylinderPlane Boundary Regularization through feature matching across the seam and feature smoothing near the seam (Jia et al., 21 Jul 2025).

CylinderPlane is described as agnostic to implicit rendering methods and compatible with both volumetric neural rendering and DMTet-based differentiable mesh rasterization. On the reported Full-Head dataset with nearly $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$29K high-resolution full-head images, the quantitative comparison is: $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$30

$B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$31

A plausible implication is that, in this specific full-head setting, explicit azimuthal separation is better aligned with $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$32 synthesis than Cartesian Tri-plane feature sharing (Jia et al., 21 Jul 2025).

Across these literatures, CylinderPlane is less a single theorem than a recurring geometric template. Exact section formulas, wall-induced vortex topology, capillary and translator rigidity, isoperimetric slabs, combinatorial duals, $B_2^m=\{x\in\mathbb R^m:\lvert x\rvert_2\le 1\}$33-divisible decompositions, and cylindrical neural feature fields all rely on the same structural asymmetry: the cylinder preserves axial or angular organization, while the plane provides a section, boundary condition, asymptotic limit, or coordinate system. This suggests that the persistence of the theme is not accidental but reflects a durable reduction principle in analysis and geometry.