How to draw combinatorial map? From graphs and edges to corner rotations and permutations

In this article we consider combinatorial maps approach to graphs on surfaces, and how between them can be establish terminological uniformity in favor of combinatorial maps in way rotations are set as base structural elements and all other notions are derived from them. We set this approach as rotational prevalence principle. We consider simplest way how to draw combinatorial map, and ask how this approach in form of rotational prevalence could be used in graphs drawing practice and wider in algorithms. We try to show in this paper that the use of corners in the place of halfedges is much more natural than that of halfedges. Formally there is no difference between both choices, but corner approach is much more clear and concise, thus we advocate for that.

💡 Research Summary

The paper revisits combinatorial maps as a unified framework for representing graphs embedded on surfaces, advocating a “rotational prevalence” principle in which rotations (permutations) are the sole primitive structural elements. Traditional half‑edge or dart models describe the incidence relations among vertices, edges, and faces using two permutations: the vertex rotation σ and the face rotation α. The authors propose to replace the half‑edge with a more natural primitive called a corner. A corner is a triple (v, e, f) that simultaneously records the vertex, the incident edge, and the incident face. Under this definition σ cycles all corners incident to the same vertex, while α cycles all corners belonging to the same face. Their composition τ = σ·α pairs each corner with its opposite corner on the same edge, reproducing the usual half‑edge pairing but without the redundancy inherent in the half‑edge representation.

The “rotational prevalence” principle states that every graph operation—vertex/face/edge insertion or deletion, embedding modification, isomorphism testing—can be expressed solely through permutation manipulations. Because permutations can be accessed in constant time, adjacency queries become O(1) per step, which dramatically improves scalability for large graphs, multigraphs, and graphs with loops or non‑simple faces. The paper demonstrates that the corner‑based model handles these pathological cases uniformly, preserving the algebraic structure of the map.

A central contribution is a minimalistic drawing algorithm for combinatorial maps. The algorithm places each corner as a point in the plane, then draws edges by following σ and α: following σ around a vertex draws the cyclic order of incident edges, while following α around a face draws the boundary of that face. No separate topological validation is required; the permutation structure guarantees a consistent embedding. This visualisation technique directly reflects the underlying algebraic data, making it easier for developers to debug and for educators to illustrate map concepts.

The authors explore how this rotation‑corner framework can be integrated into existing graph drawing pipelines. In force‑directed or layered layout algorithms, adjacency information is needed repeatedly. By querying σ for all corners of a vertex, the algorithm instantly obtains the set of incident edges, and by querying α it obtains the set of neighboring vertices around a face. This reduces the overhead of constructing adjacency lists and enables dynamic updates (e.g., edge insertions) without recomputing the entire layout. Moreover, the corner representation lends itself to compact storage: each corner stores three integer identifiers, and the two permutations can be kept as arrays of indices, yielding a memory‑efficient structure suitable for massive graphs.

From a database perspective, the paper argues that indexing corners rather than half‑edges simplifies graph query processing. A query such as “find all edges incident to vertex v” translates to traversing σ starting at any corner with vertex identifier v, which is a simple linear walk of length equal to the vertex degree. Similarly, “list all vertices on face f” is a walk of α. These walks can be executed directly in the storage engine, allowing graph‑oriented SQL extensions or NoSQL graph stores to exploit the permutation structure for fast joins and pattern matching.

Finally, the authors emphasize that while the corner model is formally equivalent to the half‑edge model, it offers clearer conceptual semantics. Corners correspond to the intuitive notion of a “corner of a face” that a human can visualize, whereas half‑edges are more abstract. This conceptual clarity, combined with the algorithmic benefits of rotation‑centric design, makes a strong case for adopting corner‑based combinatorial maps in both theoretical research and practical software systems. The paper concludes by calling for broader adoption, further tool development, and exploration of rotational prevalence in other combinatorial structures.