Regular Labelings and Geometric Structures

Three types of geometric structure—grid triangulations, rectangular subdivisions, and orthogonal polyhedra—can each be described combinatorially by a regular labeling: an assignment of colors and orientations to the edges of an associated maximal or near-maximal planar graph. We briefly survey the connections and analogies between these three kinds of labelings, and their uses in designing efficient geometric algorithms.

💡 Research Summary

The paper surveys a unifying combinatorial concept—regular labelings—and shows how it underlies three seemingly disparate geometric structures: planar grid triangulations, rectangular partitions, and orthogonal polyhedra. A regular labeling is an assignment of colors and orientations to the edges of a maximal (or near‑maximal) planar graph that satisfies simple local constraints at each vertex. The authors review each of the three instances, describe the constraints, prove key structural properties (most notably the acyclicity of monochromatic or bichromatic subgraphs, which become st‑planar graphs), and explain how these properties enable efficient geometric constructions.

Schnyder labelings for grid triangulations.
For a maximal planar graph G with a fixed outer face, each interior edge receives one of three colors (red, blue, green) and an orientation such that every interior vertex has exactly one outgoing edge of each color in cyclic order, while the three outer vertices each have only incoming edges of a single color. This forces each monochromatic subgraph to be a tree, and any two-color subgraph (with one color reversed) to be an st‑planar digraph. Consequently the three outgoing monochromatic paths from any vertex partition the interior faces into three contiguous regions. Counting faces (or vertices) in these regions yields barycentric coordinates that embed G on an integer grid of size (n‑2)×(n‑2). The construction runs in linear time and attains the best known worst‑case grid size for planar straight‑line drawings. Moreover, local “twist” operations on a triangle (reversing its three edges and cyclically permuting colors) generate a distributive lattice on the set of all Schnyder labelings; this lattice coincides with the lattice of out‑degree‑three orientations of G.

Rectangular labelings for rectangular partitions.
A rectangular partition of a large rectangle induces an “extended dual” graph that is planar, maximal except for a quadrilateral outer face, and has no separating triangles. Edges are colored red (horizontal adjacency) or blue (vertical adjacency) and oriented from the higher to the lower (or left to right) feature. At each interior vertex the incident edges of the same color and orientation appear consecutively, in the cyclic order in‑blue, in‑red, out‑blue, out‑red. These constraints again guarantee that each monochromatic subgraph is a tree and each bichromatic subgraph (with one color reversed) is st‑planar. The lengths of the longest red or blue directed paths crossing a segment give the x‑ or y‑coordinates of the corresponding rectangle, producing a geometric realization of the partition. As with Schnyder labelings, local twist operations on a quadrilateral (flipping colors inside and reversing one set of directions) connect all rectangular labelings into a distributive lattice. By Birkhoff’s representation theorem the lattice can be compactly encoded as the family of lower sets of a partial order whose elements correspond to quadrilaterals and twist counts; this DAG has O(n²) size and can be built in quadratic time. The compact representation enables rapid enumeration of all labelings and underlies fixed‑parameter tractable algorithms for area‑universal partitions and polynomial‑time algorithms for partitions respecting prescribed edge orientations.

Polyhedral labelings for orthogonal polyhedra (corner polyhedra).
A corner polyhedron is an orthogonal polyhedron where three faces are “back” faces and all other faces are oriented positively. Its dual graph is maximal planar, Eulerian, and has a distinguished outer triangle. The regular labeling colors each dual edge red, blue, or green according to the axis parallel to the corresponding polyhedron edge, and orients the edge according to the coordinate order of the two incident faces. Each triangle contains all three colors, interior vertices have edges of exactly two colors, and the outer three vertices exhibit alternating orientations. Selecting any two colors and reversing one yields an st‑planar digraph; st‑numberings of these three bichromatic digraphs provide the three coordinate values for the families of parallel face planes, thereby reconstructing the corner polyhedron from its labeling. The labeling’s local constraints again guarantee a distributive lattice of all possible polyhedral labelings, linked by twist operations on quadrilaterals of the dual.

Common themes and implications.
All three regular labeling families share a common combinatorial skeleton: a (near‑)maximal planar graph with a distinguished outer face, a coloring/orientation scheme satisfying simple cyclic adjacency rules, and the resulting monochromatic/bichromatic subgraphs being acyclic (st‑planar). These properties give rise to distributive lattices of labelings, where elementary twist moves correspond to covering relations. The lattice viewpoint provides a powerful algebraic tool: it yields compact representations, enables enumeration, and supports algorithmic transformations between the three geometric domains. By exposing these deep analogies, the paper not only unifies existing algorithms for straight‑line planar drawings, rectangular cartograms, treemaps, VLSI floorplanning, and orthogonal polyhedron construction, but also suggests new avenues for cross‑application of techniques—such as using lattice‑based enumeration to explore design spaces in architecture or graphics, or transferring fixed‑parameter algorithms from rectangular partitions to polyhedral layout problems. Overall, the survey highlights regular labelings as a versatile bridge between combinatorial graph theory and practical geometric computation.