Colorful Quadrangulations in Graph Theory
- The paper introduces a unifying framework where quadrilateral structures enforce global rigidity in graph colorings through precise parity, homological, and algorithmic techniques.
- Colorful quadrangulations are defined by the interplay of proper vertex colorings, cell labelings, and surface embeddings, with applications in both enumeration and computational graph theory.
- Key methods include bijective enumerations, combinatorial degree arguments, and constructive embeddings that yield efficient algorithms for 3-color precoloring extensions.
Colorful quadrangulations form a cluster of closely related notions linking graph coloring, surface embeddings, cubical topology, bijective enumeration, and algorithmic precoloring extension. In the literature, the phrase ranges from rainbow quadrilateral faces in proper $4$-colorings of embedded quadrangulations, through centrally labelled cells in Sperner-type quadrangulations, to integer-labeled quadrangulations whose faces have label pattern and hence induce a proper $3$-coloring modulo $3$ (Hoffmann-Ostenhof et al., 2016, Musin, 2014, Budd, 29 Sep 2025). Across these models, the common theme is that quadrilateral structure interacts unusually tightly with orientation, homology, cup products, and label dynamics.
1. Core notions and competing models
In the surface-theoretic setting, a quadrangulation is a graph embedded so that every face boundary has length $4$. One standard orientable version takes to be finite, loopless, $2$-edge-connected, and embedded on a closed orientable surface so that every face boundary is a cycle of length $4$ (Hoffmann-Ostenhof et al., 2016). A closely related topological-graph-theoretic version views a quadrangulation of the closed orientable surface as a $2$-cell embedding in which every region is bounded by a simple circuit of length 0 (Lawrencenko, 2012).
Several higher-dimensional analogues coexist. In Musin’s cubical framework, a quadrangulation of a PL 1-manifold is a cell decomposition by cubes, with piecewise multilinear maps replacing affine simplicial maps (Musin, 2014). In the “normal quadrangulation” terminology, a normal quadrangulation of a 2-manifold 3 is the 4-skeleton of a cubical complex whose underlying space is homeomorphic to 5 (Kaiser et al., 29 Mar 2025). By contrast, the Kaiser–Stehlík notion uses a generalized simplicial complex 6 and requires that every maximal simplex induce a nonempty complete bipartite graph on its vertices; this is the higher-dimensional projective-space model behind the bound 7 for non-bipartite quadrangulations of 8 (Kaiser et al., 2013).
“Colorful” also has more than one precise meaning. In orientable surface quadrangulations with a proper vertex-9-coloring $3$0, a rainbow face is a face using all four colors, and a $3$1-face records the clockwise cyclic order of those colors (Hoffmann-Ostenhof et al., 2016). In cubical Sperner theory, the key object is a centrally labelled cell, defined by the condition that the piecewise multilinear map $3$2 hits the cube center $3$3 in the interior of the cell; in dimension $3$4, these are exactly fully colored quadrangles together with opposite-labeled edges $3$5 and $3$6 (Musin, 2014). In the labeled-map model of rigid quadrangulations, a colorful quadrangulation is a $3$7-labeled quadrangulation in which labels differ by $3$8 across each edge and every face has labels $3$9, so that labels modulo $3$0 give a proper $3$1-coloring and every face uses all three colors (Budd, 29 Sep 2025).
A further abstraction comes from colorful polytopes: given a properly edge-colored $3$2-regular graph $3$3, the associated colorful polytope $3$4 has $3$5-faces equal to the connected components of $3$6-colored subgraphs, and those $3$7-faces always have even length. In rank $3$8, $3$9 is a quadrangulation exactly when every $4$0-colored connected component is a $4$1-cycle (Araujo-Pardo et al., 2012).
2. Local colorful configurations and exact parity laws
For proper vertex-$4$2-colorings of quadrangulations on orientable surfaces, the basic local object is the rainbow face. The central counting theorem states that for any permutation $4$3 of $4$4,
$4$5
where $4$6 counts rainbow faces whose clockwise boundary colors are $4$7 (Hoffmann-Ostenhof et al., 2016). An immediate corollary is that the total number of rainbow faces is even.
The proof is an exact local-to-global cancellation argument. Each edge is oriented from the endpoint of larger color to the endpoint of smaller color, and for each face $4$8 one defines
$4$9
where 0 and 1 are the boundary edges whose orientation agrees with, or is opposite to, the clockwise direction around 2. Because each edge contributes 3 to one incident face and 4 to the other, one has
5
All 6-color and 7-color faces satisfy 8, while 9-faces contribute $2$0 and $2$1-faces contribute $2$2. This forces equality of the two counts (Hoffmann-Ostenhof et al., 2016). Orientability is essential twice: it is needed for a globally coherent clockwise orientation and for the sign cancellation along shared edges.
Musin’s Sperner-type framework enlarges the catalogue of local colorful events. For a quadrangulation $2$3 of a polygon, a labeling $2$4 with no boundary edge labeled $2$5 or $2$6 determines a combinatorial degree
$2$7
and Theorem C states that $2$8 contains at least $2$9 centrally labelled cells (Musin, 2014). In dimension $4$0, centrally labelled cells are either quadrangles containing all four labels or edges with opposite labels $4$1 or $4$2. For a Sperner-labeled pile $4$3, this yields the direct quadrangulated analogue of Sperner’s lemma: there exists either a quadrangle whose vertices have the four different colors, or an edge whose endpoints are colored $4$4 or $4$5 (Musin, 2014).
A third local pattern appears in the 2025 Youngs-generalization framework. If $4$6 has quadrangulated $4$7-skeleton and there exists an odd closed walk $4$8 whose integral homology class is torsion, then any $4$9-coloring 0 must contain a rainbow square; in particular, 1 is not 2-chromatic (Enami et al., 29 May 2025). Here the colorful cell is again a 3-cycle, but now forced by a homological obstruction rather than by counting or boundary degree.
3. Topological and homological coloring obstructions
Youngs’ theorem is the historical starting point for the projective case: any quadrangulation of 4 is not 5-chromatic (Enami et al., 29 May 2025). In the higher-dimensional Kaiser–Stehlík model, this becomes a dimension-sensitive lower bound: if 6 is a non-bipartite quadrangulation of 7, then
8
(Kaiser et al., 2013). The proof lifts the quadrangulation from 9 to an antipodally symmetric triangulation of $2$0, extracts an antisymmetric $2$1-coloring upstairs, and thereby constructs a $2$2-equivariant map $2$3 into the box complex. Lovász’s inequality $2$4 then yields the bound (Kaiser et al., 2013). The bound is sharp, since the same paper shows that complete graphs $2$5 and Mycielski graphs $2$6 embed as non-bipartite quadrangulations in $2$7 (Kaiser et al., 2013).
The 2025 torsion-homology generalization subsumes several earlier non-$2$8-colorability theorems. If $2$9 is a CW complex with even 00-skeleton, each 01-cell attached along some 02 with 03, and 04 contains an odd closed walk whose integral homology class is torsion, then
05
For quadrangulated 06-skeleta, 07, so 08, hence 09 (Enami et al., 29 May 2025). This recovers Youngs’ theorem, the Archdeacon–Hutchinson–Nakamoto–Negami–Ota and Mohar–Seymour non-orientable surface criterion, and the Kaiser–Stehlík projective-space theorem within one statement. The paper’s structural explanation is that “odd closed walk + torsion in 10” is the common obstruction (Enami et al., 29 May 2025).
Normal quadrangulations behave differently. For the cubical notion of Hachimori–Nakamoto–Ozeki, the 2025 projective-space paper proves that no non-bipartite normal quadrangulation of 11 is 12-colorable for any 13 (Kaiser et al., 29 Mar 2025). The proof constructs codimension-one submanifolds 14 from a hypothetical 15-coloring and derives a contradiction from nontrivial cup products in
16
However, the same paper proves that for every 17 there exists a normal quadrangulation of 18 whose 19-keleton contains 20, so chromatic number is arbitrarily large in dimension 21 (Kaiser et al., 29 Mar 2025). Thus the statement “non-bipartite implies 22-chromatic” survives in dimension 23, survives only as “not 24-colorable” for normal cubical quadrangulations in higher dimensions, and fails completely as a bounded-chromatic phenomenon already for 25.
4. Constructive frameworks, embeddings, and minimality
One major constructive approach is spinal quadrangulation. For any non-trivial connected graph 26, White and Craft’s theorem gives a quadrangulation
27
where 28 is the 29-fold interlacement and 30 is the first Betti number (Lawrencenko, 2012). This construction preserves chromatic number,
31
and doubles order,
32
As a consequence, for any integers 33 and 34 with
35
there exists a spinal quadrangulation of 36 with chromatic number 37 (Lawrencenko, 2012). Edge-deleted complete graphs 38 tune the genus precisely, and the resulting quadrangulations are minimal whenever
39
A second extremal construction concerns complete graphs themselves. The minimum orientable genus for an embedding of 40 with all face degrees at least 41 is
42
while the minimum nonorientable genus is
43
for all 44, with one small exceptional distinction between 45 and the even-faced orientable genus 46, and between 47 and 48 (Liu et al., 2016). In the congruence classes forced by Euler counting, these minimum-genus embeddings are genuine quadrangulations: orientably iff 49 or 50, and nonorientably iff 51 or 52 (Liu et al., 2016). These embeddings provide sharpness examples for the Even Map Color Theorem.
Planar simple quadrangulations admit another canonical structure. A face-rooted quadrangulation has girth 53 if and only if it admits the canonical minimal ordinary orientation with indegree 54 at each inner vertex, obtained in the general 55-angulation theory as the 56 specialization of 57-orientations (Bernardi et al., 2010). Under the master bijection of Bernardi and Fusy, simple quadrangulations correspond to mobiles with black vertices of degree 58 and white vertices of degree 59, recovering Schaeffer’s bijection and Brown’s counting formulas (Bernardi et al., 2010).
The colorful-polytope viewpoint gives yet another constructive criterion. If 60 is a properly edge-colored cubic graph, then the rank-61 colorful polytope 62 is a quadrangulation exactly when every connected component of every 63-colored subgraph is a 64-cycle (Araujo-Pardo et al., 2012). The same underlying graph can produce either a spherical quadrangulation or a toroidal map with mixed squares and octagons, depending on the chosen edge-coloring, as illustrated by two distinct colorings of the cube graph (Araujo-Pardo et al., 2012).
5. Algorithmic colorability and precoloring extension
For fixed surfaces, colorful quadrangulations are also algorithmically tractable. The central theorem of Dvořák, Král’, and Thomas gives, for every fixed surface 65 and integer 66, a linear-time algorithm that takes a quadrangulation 67 of 68 with at most 69 boundary vertices and a coloring of those boundary vertices by 70, and decides whether the precoloring extends to a 71-coloring of 72; if it does, the algorithm outputs such a coloring (Dvorak et al., 2015).
The structural invariant behind this algorithm is winding number. For a coloring 73, one defines
74
If 75 is a quadrangulation of an orientable surface and 76 are boundary cycles oriented clockwise, then every 77-coloring satisfies
78
(Dvorak et al., 2015). On non-orientable surfaces, this is replaced by the parity-compliance condition
79
where 80 is defined from local face and cuff orientations (Dvorak et al., 2015). For 81-generic quadrangulations, these conditions are not merely necessary but sufficient.
The disk case is even more explicit. If 82 is a quadrangulation of a disk with boundary cycle 83, then a boundary 84-coloring 85 of winding number 86 extends if and only if
87
(Dvorak et al., 2015). In boundary-linked disks, 88 follows automatically from 89.
A related but more specialized algorithmic theory handles near-quadrangulations of the cylinder. Given a simple connected plane graph 90 with two distinguished facial cycles 91, the extension problem for a 92-coloring of 93 can be solved in time
94
where
95
96
(Dvořák et al., 2019). For genuine cylinder quadrangulations, 97, so the complexity is essentially linear. As a corollary, every triangle-free graph embedded in the torus with edge-width at least 98 is 99-colorable (Dvořák et al., 2019).
6. Enumeration, bijections, and discrete-flat models
The most explicit enumerative theory currently comes from the bijection between rigid quadrangulations and colorful integer-labeled quadrangulations. A rigid quadrangulation is a flat quadrangulation of the disk in which every inner vertex has degree $3$00, every concave boundary vertex has degree exactly $3$01, and every ray connects a concave corner to a straight boundary vertex (Budd, 29 Sep 2025). The corresponding colorful model consists of rooted $3$02-labeled quadrangulations in which labels differ by $3$03 across edges, the root edge points from label $3$04 to label $3$05, and every face has labels
$3$06
in cyclic order (Budd, 29 Sep 2025).
The main bijection
$3$07
identifies non-root convex corners of the rigid quadrangulation with vertices of the colorful quadrangulation, and the label of the corresponding vertex is exactly the turning number of the corner (Budd, 29 Sep 2025). The full dictionary is precise: non-root convex corners correspond to vertices, rows to even edges, columns to odd edges, and the root corner together with concave corners correspond to faces; right-tangential convex corners correspond to local minima, left-tangential convex corners to local maxima, and tangential sides correspond to level lines (Budd, 29 Sep 2025).
This bijection transfers the exact colorful generating functions of Bousquet-Mélou and Elvey Price to rigid quadrangulations. If $3$08 denotes rigid quadrangulations with $3$09 convex corners, then
$3$10
where $3$11 is the unique formal power series satisfying
$3$12
(Budd, 29 Sep 2025). The asymptotic growth is
$3$13
Boundary-refined colorful disk classes $3$14, B-patches, C-patches, and E-patches produce multivariate generating functions $3$15, $3$16, $3$17, and $3$18, and the colorful two-catalytic system becomes a rigid-combinatorial recursion under the bijection (Budd, 29 Sep 2025).
A second 2025 paper gives a direct tree bijection on the rigid side. Budd had shown that rigid quadrangulations are in bijection with colorful quadrangulations, and Zonneveld proves the corresponding generating series bijectively by encoding rigid quadrangulations with decorated binary trees (Zonneveld, 29 Sep 2025). If $3$19 denotes complete base-$3$20 rigid quadrangulations and
$3$21
then for $3$22,
$3$23
while for $3$24,
$3$25
(Zonneveld, 29 Sep 2025). The paper states that this direct bijective proof “opens the door to better understand the geometry of random rigid quadrangulations (and maybe even of random colorful quadrangulations), by studying the corresponding decorated trees” (Zonneveld, 29 Sep 2025).
Taken together, these strands show that colorful quadrangulations are not a single rigid object class but a research nexus. In some papers they are parity-constrained rainbow faces of $3$26-colored surface embeddings; in others they are centrally forced cells in cubical Sperner theory; in projective topology they are witnesses to homological and cup-product obstructions; and in bijective enumeration they are the solvable labeled-map avatars of rigid flat disks. The unifying feature is that quadrilateral structure makes color information globally rigid enough to be counted exactly, detected homologically, and manipulated algorithmically.