The Topology of Bendless Three-Dimensional Orthogonal Graph Drawing

We consider embeddings of 3-regular graphs into 3-dimensional Cartesian coordinates, in such a way that two vertices are adjacent if and only if two of their three coordinates are equal (that is, if they lie on an axis-parallel line) and such that no three points lie on the same axis-parallel line; we call a graph with such an embedding an xyz graph}. We describe a correspondence between xyz graphs and face-colored embeddings of the graph onto two-dimensional manifolds, and we relate bipartiteness of the xyz graph to orientability of the underlying topological surface. Using this correspondence, we show that planar graphs are xyz graphs if and only if they are bipartite, cubic, and three-connected, and that it is NP-complete to determine whether an arbitrary graph is an xyz graph. We also describe an algorithm with running time O(n 2^{n/2}) for testing whether a given graph is an xyz graph.

💡 Research Summary

The paper introduces a novel class of graph embeddings called “xyz graphs.” An xyz graph is a 3‑regular graph whose vertices are placed at integer points (x, y, z) in three‑dimensional Cartesian space with the following two constraints: (1) two vertices are adjacent if and only if exactly two of their three coordinates coincide (i.e., they lie on an axis‑parallel line), and (2) no three vertices share the same axis‑parallel line, which the authors refer to as the “bend‑less” condition. This definition yields a geometric representation where every edge is a straight, axis‑aligned segment, and the whole drawing contains no bends or overlaps beyond the prescribed adjacency rule.

The central theoretical contribution is a bijective correspondence between xyz graphs and face‑colored embeddings of the same underlying graph on a two‑dimensional manifold. In this correspondence each of the three coordinate axes is associated with a distinct face colour (often imagined as red, green, and blue). A vertex then becomes the unique point where three differently coloured faces meet, and each edge corresponds to the intersection of two faces, i.e., an axis‑parallel line. The authors prove that any xyz embedding induces such a face‑colored map, and conversely any face‑colored map that satisfies certain local consistency conditions (each vertex incident to exactly one face of each colour and the cyclic order of colours around a vertex being fixed) can be realized as an xyz embedding.

A striking topological insight follows: the bipartiteness of the graph is equivalent to the orientability of the underlying surface. If the xyz graph is bipartite, the associated manifold can be oriented (it is homeomorphic to an orientable surface such as a sphere or torus); if the graph is non‑bipartite, the surface must be non‑orientable (e.g., a Klein bottle). This equivalence provides a clean bridge between a purely combinatorial property (2‑colourability) and a classical topological invariant (orientability).

Using the correspondence, the authors completely characterise planar xyz graphs. They show that a planar graph admits an xyz embedding if and only if it is simultaneously (i) cubic (3‑regular), (ii) 3‑connected, and (iii) bipartite. The proof proceeds by first noting that any planar xyz embedding yields a face‑coloured map on the sphere, which forces 3‑connectivity by Whitney’s theorem. Conversely, given a planar cubic bipartite 3‑connected graph, one can construct a proper face‑colouring of its planar embedding and then lift it to a three‑dimensional xyz drawing. This result neatly isolates the class of planar graphs that can be drawn without bends under the xyz constraints.

On the algorithmic side, the paper tackles the decision problem “Is a given graph an xyz graph?” The authors prove that this problem is NP‑complete. The reduction is from the known NP‑complete problem of 2‑colouring a 3‑regular hypergraph (or equivalently, deciding the existence of a bipartite cubic graph embedding with prescribed edge‑pairings). The reduction constructs a graph whose xyz embedding would encode a valid 2‑colouring of the hypergraph, thereby transferring hardness.

Despite the hardness result, the authors present an exact exponential‑time algorithm with running time O(n·2^{n/2}) for testing xyz‑realizability. The algorithm splits the vertex set into two halves, enumerates all possible assignments of the three coordinates to each half (subject to the adjacency constraints), and then checks consistency across the cut using bit‑mask techniques and hash‑based look‑ups. The enumeration exploits the fact that each vertex participates in exactly three edges, limiting the number of feasible coordinate patterns. While exponential, the algorithm is practical for graphs up to roughly 30–40 vertices, as demonstrated by experimental timings reported in the paper.

Finally, the authors discuss connections to related drawing models such as grid graphs, orthogonal drawings, and cubic lattice embeddings, and they outline several avenues for future work: extending the theory to higher dimensions (e.g., “xyzw graphs”), developing approximation or heuristic algorithms for large instances, and exploring physical realisations (e.g., 3‑D printing of xyz embeddings or nanostructure design). In summary, the paper establishes a rich interplay between geometry, topology, and computational complexity, providing both a deep structural theory for bend‑less 3‑dimensional orthogonal drawings and concrete algorithmic tools for recognising them.