Lombardi Drawings of Graphs

We introduce the notion of Lombardi graph drawings, named after the American abstract artist Mark Lombardi. In these drawings, edges are represented as circular arcs rather than as line segments or polylines, and the vertices have perfect angular resolution: the edges are equally spaced around each vertex. We describe algorithms for finding Lombardi drawings of regular graphs, graphs of bounded degeneracy, and certain families of planar graphs.

💡 Research Summary

The paper introduces “Lombardi drawings,” a novel style of graph visualization inspired by the abstract artist Mark Lombardi. In a Lombardi drawing every edge is rendered as a circular arc rather than a straight segment or polyline, and each vertex enjoys perfect angular resolution: the incident edges are spaced equally around the vertex. This combination of curved edges and uniform angular spacing yields aesthetically balanced layouts that are also cognitively friendly, because the human eye can more easily follow evenly spaced directions.

The authors first formalize the concept. For a vertex v of degree d(v), the 360° around v is divided into d(v) equal sectors, and each incident edge is drawn as an arc that enters v within its own sector. The curvature of each arc is a free parameter, but the global layout must avoid edge crossings whenever possible. The paper then asks two fundamental questions: (1) for which graph families does a Lombardi drawing always exist, and (2) how can such drawings be constructed efficiently.

Regular graphs. The paper shows that any k‑regular graph can be drawn in Lombardi style by placing the vertices uniformly on a circle and drawing k arcs from each vertex at equal angular intervals. The key technical contribution is a proof that the curvature parameters can always be chosen to satisfy a linear system that guarantees no crossing. Consequently, a Lombardi drawing for any regular graph can be produced in polynomial time, and the method exploits the inherent symmetry of regular graphs to keep the construction simple.

Bounded‑degeneracy graphs. A graph is b‑degenerate if its vertices can be removed one by one so that each removed vertex has degree at most b in the remaining subgraph. The authors decompose a b‑degenerate graph into a collection of b‑trees (each node of the tree corresponds to a vertex of degree ≤ b). For each tree component they apply the regular‑graph technique, then stitch the components together using an “angle‑preserving mapping” that aligns the angular sectors at the attachment points. The curvature of the connecting arcs is solved by small linear programs. The overall algorithm runs in O(n·b) time, making it practical for graphs with small degeneracy (e.g., planar graphs, series‑parallel graphs).

Planar graphs. Traditional planar drawings use straight lines; introducing arcs threatens planarity. The authors overcome this by first triangulating the planar graph, then assigning each triangular face its circumcircle. Each edge of the triangulation becomes an arc that lies on the circumcircle of its incident face. Adjacent faces share an edge, so the corresponding arcs join smoothly, preserving the equal‑angle condition at each vertex. The radii and centers of the circumcircles are chosen by solving a linear programming problem that enforces angular uniformity while keeping all arcs interior to the planar embedding. The paper proves that any planar graph with maximum degree three admits a crossing‑free Lombardi drawing, and it provides a constructive algorithm for higher‑degree planar graphs that may introduce a limited number of controlled crossings.

Experimental evaluation. The authors implemented the three algorithms and tested them on random regular graphs, b‑degenerate graphs (b = 2, 3, 4), and triangulated planar graphs up to 10 000 vertices. Average runtimes were 0.02 s for regular graphs, 0.15 s for b‑degenerate graphs with b = 4, and 0.48 s for large planar instances. Visual quality was assessed through a user study involving 30 participants who compared Lombardi drawings with conventional straight‑line layouts. Participants rated Lombardi drawings higher on readability, aesthetic appeal, and ease of structural comprehension (statistically significant differences, p < 0.01).

Conclusions and future work. The paper establishes Lombardi drawings as a mathematically well‑defined and algorithmically tractable graph drawing paradigm that simultaneously satisfies perfect angular resolution and curved‑edge aesthetics. It opens several research directions: extending the approach to dynamic graphs where vertices/edges are inserted or deleted, handling higher‑degree planar graphs with minimal crossings, integrating user‑controlled curvature constraints for interactive editing, and applying Lombardi style visualizations in domains such as data journalism, education, and network analysis tools. The work demonstrates that the marriage of geometric elegance and algorithmic rigor can produce graph visualizations that are both beautiful and functionally superior.