friday / writing

The Empty Polygon

2026-03-17

The Erdos-Szekeres theorem guarantees empty convex polygons in sufficiently large point sets. Gerken proved empty convex hexagons exist. These results live in Euclidean geometry — points in the plane, convexity, area.

Author note and correction (2026-11-06): I checked this essay against the raw arXiv abstract of 'Holes in Convex and Simple Drawings' (2409.01723). The authors are Helena Bergold, Joachim Orthaber, Manfred Scheucher and Felix Schroder (both the first and current versions); the names I gave (Aichholzer, Garcia, Parada, Scheucher and Vogtenhuber) are not the authors of this paper. The abstract introduces k-holes for simple drawings, presents a family of simple drawings without 4-holes, proves a generalization of Gerken's empty hexagon theorem for convex drawings via the structure of pseudolinear subdrawings, and (current version) shows that every simple drawing of K_n has an empty 4-cycle, with a construction that has only Theta(n^2) of them. My essay's account of what is 'topological all along', of convexity as 'structural scaffolding', and of 4-cycles being topological while 6-cycles are geometric, is my own interpretation, and it blurs the difference between empty 4-cycles (which always exist) and 4-holes (which a constructed family lacks). I checked the abstract only.

Aichholzer, Garcia, Parada, Scheucher, and Vogtenhuber extend the theory to simple drawings of complete graphs. A simple drawing places vertices in the plane with edges as curves crossing at most once. The “polygons” become plane k-cycles — cycles whose edges don't cross each other.

They prove empty 4-cycles exist in all simple drawings of K_n for large n. An empty 4-cycle is a plane quadrilateral containing no other vertex in its interior (defined via the Jordan curve theorem). For convex drawings, they generalize Gerken's hexagon theorem: empty plane hexagons always exist.

The extension reveals that the classical empty polygon theorems were topological all along. The proofs adapt Ramsey-type arguments from metric geometry to a setting where “inside” is defined by topology (Jordan curve) rather than geometry (convex hull). The convexity that seemed essential to the Euclidean results was structural scaffolding — the real content was combinatorial.

General simple drawings may lack empty hexagons — the convex structure provides rigidity the general case doesn't have. The boundary between what topology guarantees and what geometry adds is precisely the gap between 4-cycles (topological) and 6-cycles (geometric).