Generalized Twisted Drawings

• Generalized twisted drawings are simple drawings of complete graphs characterized by a fixed center of radial monotonicity and a transversal ray that crosses every edge exactly once.
• They yield precise enumerative results, such as exactly 2n-4 empty triangles, and guarantee the existence of large plane substructures like Hamiltonian paths or cycles.
• Their structural characterizations via antipodal vi-cells, transversal curves, and rotation systems support efficient recognition algorithms and provide new extremal graph drawing bounds.

Searching arXiv for papers on generalized twisted drawings and related characterizations. Generalized twisted drawings are a class of simple drawings of complete graphs $K_n$ in which the edges are constrained by a radial monotonicity condition around a point $O$ and by the existence of a ray from $O$ that intersects every edge exactly once. They were introduced to obtain improved bounds on plane substructures in simple drawings of $K_n$, but subsequent work showed that the notion supports a substantial structural and algorithmic theory: it admits characterizations via antipodal cells, transversal curves, and rotation systems; it yields exact enumerative results for empty triangles; and it interacts naturally with broader parity-based approaches to topological graph drawing (Aichholzer et al., 2022, Aichholzer et al., 22 Aug 2025).

1. Geometric definition and scope

A simple drawing of a graph is a drawing in which vertices are distinct points, edges are Jordan arcs between their endpoints, no edge passes through another vertex, and any two edges meet at most once, either at a common endpoint or in a proper crossing. In the generalized twisted literature, the primary setting is simple drawings of complete graphs $K_n$.

A drawing is c-monotone if there exists a point $O$ such that every ray emanating from $O$ intersects every edge at most once. A simple drawing is generalized twisted if there is a point $O$ such that the drawing is c-monotone with respect to $O$ and there exists a ray $r$ emanating from $O$0 that crosses every edge exactly once. Later work formulates the same notion as a property up to strong isomorphism: a simple drawing $O$1 is generalized twisted if it is strongly isomorphic to a drawing with a point $O$2 and a ray $O$3 satisfying those two conditions (Aichholzer et al., 2022, Aichholzer et al., 22 Aug 2025).

The definition is deliberately geometric and drawing-specific. The point $O$4 acts as a center for a radial monotonicity constraint, while the special ray $O$5 functions as a global transversal. This combines a local restriction—no ray from $O$6 may meet an edge twice—with a global one—some ray must meet every edge. The resulting class is broader than the classical twisted drawings: every twisted drawing is weakly isomorphic to a generalized twisted drawing, but not every generalized twisted drawing is weakly isomorphic to a twisted drawing. The literature also notes that generalized twisted drawings exist for every $O$7 (Aichholzer et al., 2022).

2. Structural constraints and plane substructures

The class is rigid enough to force nontrivial plane substructures. A central local statement is that if $O$8 is a generalized twisted drawing of $O$9 with vertices labeled $O$0 counterclockwise around $O$1, then the edges $O$2 and $O$3 do not cross. This crossing prohibition is the basic combinatorial pattern from which larger plane structures are derived (Aichholzer et al., 2022).

From that lemma, it follows that every generalized twisted drawing of $O$4 contains a plane Hamiltonian path. The construction is explicit: when the vertices are ordered counterclockwise around $O$5, one can write down a Hamiltonian path whose nonadjacent edges correspond to alternating index patterns, and the $O$6 lemma rules out crossings among those nonadjacent edges. For odd $O$7, a similar argument gives a plane Hamiltonian cycle, and the literature records the conjecture that every generalized twisted drawing contains a plane Hamiltonian cycle (Aichholzer et al., 2022).

Generalized twisted drawings are also structurally dense in a crossing sense. The papers note that every generalized twisted drawing is crossing maximal: every $O$8-vertex induced subdrawing contains a crossing. This rules out the possibility that generalized twistedness is merely a weak perturbation of planar structure. The class is instead organized around a highly constrained crossing pattern that still coexists with large plane subdrawings.

A common misconception is that generalized twisted drawings are simply twisted drawings in a different geometric presentation. The available results show otherwise. The generalized class preserves the transversal-ray mechanism and many extremal consequences of twisted drawings, but allows more geometric flexibility and more weak isomorphism types than the classical twisted family (Aichholzer et al., 2022).

3. Characterizations by cells, curves, and extensions

One of the main advances in the theory is that generalized twistedness can be recognized without directly referring to the original point-and-ray construction. The key topological object is a vi-cell, meaning a cell of the drawing whose boundary contains at least one vertex. Two cells are antipodal if for every triangle in the drawing, the two cells lie on opposite sides of that triangle.

For a simple drawing $O$9 of $K_n$0, the following conditions are equivalent:

1. $K_n$1 is weakly isomorphic to a generalized twisted drawing.
2. $K_n$2 contains two antipodal vi-cells.
3. $K_n$3 can be extended by a simple curve $K_n$4 such that $K_n$5 crosses every edge of $K_n$6 exactly once.
4. $K_n$7 can be extended by two vertices $K_n$8 and $K_n$9, and edges incident to them, such that the resulting drawing is a simple drawing of $K_n$0, the edge $K_n$1 crosses every edge of $K_n$2, and no edge incident to $K_n$3 crosses any edge incident to $K_n$4 (Aichholzer et al., 2022).

These equivalences replace the original geometric definition by several drawing-independent or nearly drawing-independent formulations. In particular, antipodal vi-cells encode generalized twistedness at the level of the cell decomposition, while the transversal-curve condition isolates the essential global feature: the existence of a simple curve that crosses every edge exactly once.

The extension by $K_n$5 and $K_n$6 is especially useful conceptually. In generalized twisted drawings, the cell containing the special point $K_n$7 and the unbounded cell form a pair of antipodal vi-cells. Conversely, if a pair of antipodal vi-cells exists, then one can construct a curve crossing every edge exactly once; this in turn can be promoted to an extension by vertices $K_n$8 and $K_n$9. This shows that generalized twistedness is not merely an artifact of a chosen geometric model but a topological property of the drawing class (Aichholzer et al., 2022).

The same viewpoint reappears in later rotation-system work. There, a previously known theorem is used heavily: a rotation system of a simple drawing $O$0 of $O$1 is generalized twisted if and only if $O$2 contains a pair of antipodal vi-cells. This supplies a direct bridge between cell structure and combinatorial encodings (Aichholzer et al., 22 Aug 2025).

4. Rotation systems and recognition of generalized twistedness

For a drawing, the rotation at a vertex is the cyclic order of the incident edges around that vertex, and the collection of all vertex rotations is the drawing’s rotation system. For complete graphs, rotation systems are particularly informative: two simple drawings of $O$3 have the same crossing-edge pairs if and only if they have the same or inverse rotation system. This motivates the distinction among abstract rotation systems, realizable rotation systems, and generalized twisted rotation systems (Aichholzer et al., 22 Aug 2025).

An abstract rotation system assigns a cyclic order to the edges at each vertex without presupposing any drawing. It is realizable if some simple drawing induces it. It is generalized twisted if it can be realized by a generalized twisted drawing. Not every abstract rotation system is realizable, and not every realizable rotation system is generalized twisted.

The first major characterization is local-to-global. For any $O$4, an abstract rotation system of $O$5 is generalized twisted if and only if all subrotation systems induced by five vertices are generalized twisted. This gives a concise combinatorial description of generalized twistedness and shows that, for sufficiently large complete graphs, the property is determined by $O$6-vertex restrictions. The theorem does not extend to $O$7: there is a $O$8 rotation system whose every $O$9-vertex subrotation system is generalized twisted, but the whole system is not. The same work records that $O$0 has exactly one generalized twisted rotation system, called $O$1, and that $O$2 has three generalized twisted rotation systems (Aichholzer et al., 22 Aug 2025).

A second characterization applies to realizable rotation systems. If $O$3 is the rotation system of a simple drawing $O$4 of $O$5, then $O$6 is generalized twisted if and only if there exist vertices $O$7 and a bipartition $O$8 of $O$9 such that:

1. for every $O$0, the edge $O$1 crosses $O$2;
2. for every $O$3, the edge $O$4 crosses $O$5;
3. for every $O$6 and $O$7, the edge $O$8 does not cross $O$9;
4. in the rotation of $O$0, starting at $O$1, all vertices of $O$2 appear before all vertices of $O$3;
5. in the rotation of $O$4, starting at $O$5, all vertices of $O$6 appear before all vertices of $O$7 (Aichholzer et al., 22 Aug 2025).

This bipartition theorem gives a global structural certificate for realizable inputs. It also underlies the current recognition algorithms.

Input	Characterization used	Time
Abstract rotation system of $O$8	Check all induced $O$9-vertex subrotation systems	$r$0
Realizable rotation system of $r$1	Search for $r$2 and a bipartition satisfying the five conditions	$r$3

The quadratic-time algorithm for realizable rotation systems proceeds by computing the empty star triangles at a chosen vertex, identifying a small constant set of candidate compatible vertex pairs, and testing the bipartition conditions. The literature emphasizes that this $r$4 method does not directly apply to arbitrary abstract rotation systems; the general abstract case currently uses the $r$5 local-checking algorithm (Aichholzer et al., 22 Aug 2025).

5. Extremal consequences: disjoint edges, long paths, and empty triangles

Generalized twisted drawings were introduced as a tool for proving new lower bounds in arbitrary simple drawings of complete graphs. One theorem shows that every simple drawing of $r$6 contains at least

$r$7

pairwise disjoint edges. Another shows that every simple drawing of $r$8 contains a plane path of length

$r$9

In both arguments, generalized twisted subdrawings serve as the mechanism that converts a structured local configuration into a guaranteed plane Hamiltonian path or another large plane substructure (Aichholzer et al., 2022).

Later work places these results in a broader extremal context. Generalized twisted drawings have been central in deriving the best known lower bounds on the size of a largest plane matching, the length of a longest plane cycle, and the length of a longest plane path in any simple drawing of $O$00. They are also the largest known class of drawings for which every drawing of $O$01 has exactly

$O$02

empty triangles, a quantity conjectured to be the minimum for all simple drawings (Aichholzer et al., 22 Aug 2025).

An empty triangle is a triangle with an empty side, where “empty” means that the side contains no vertices of the drawing. A star triangle $O$03 at $O$04 is one for which the edge $O$05 is not crossed by any edge incident to $O$06. In generalized twisted drawings of $O$07, every vertex is incident to exactly two empty star triangles, one having $O$08 on its empty side and the other having $O$09 on its empty side, and the two empty sides are disjoint. The counting argument then shows that every generalized twisted drawing of $O$10 contains exactly $O$11 empty triangles (García et al., 2022).

The exact count is significant because it generalizes the extremal behavior of the classical twisted drawings. Harborth had constructed simple drawings with exactly $O$12 empty triangles, and those extremal examples were classical twisted drawings. The generalized-twisted theorem shows that the same extremal value is attained by a much broader family, not by an isolated configuration (García et al., 2022).

6. Parity, rotations, and the Hanani–Tutte perspective

Although generalized twisted drawings are usually studied through geometric and combinatorial characterizations of simple drawings of $O$13, they also fit naturally into a parity-based view of graph drawings. The most relevant background result is the Unified Hanani–Tutte theorem: if a graph $O$14 has a drawing $O$15 in the plane where every pair of edges that are independent or have a common endpoint in a prescribed set $O$16 crosses an even number of times, then $O$17 has a planar embedding in which the cyclic orders of edges at vertices from $O$18 are the same as in $O$19 (Fulek et al., 2016).

The theorem interpolates between the classical strong and weak Hanani–Tutte theorems. If $O$20, the condition reduces to even crossings for independent edges, yielding the strong theorem. If $O$21, every pair of edges crosses evenly, and the resulting embedding preserves the rotation at every vertex, yielding the weak theorem. The relevant notions are standard: an edge is even if it crosses every other edge an even number of times, a vertex is even if all pairs of incident edges cross each other evenly, a drawing is independently even if every pair of independent edges crosses evenly, and the rotation of a vertex is the clockwise cyclic order of its incident edges (Fulek et al., 2016).

This theorem is not a definition of generalized twisted drawings, and the paper does not use “generalized twisted drawings” as its central formal object. Nevertheless, it provides the parity statement that most closely parallels the generalized-twisted viewpoint. The theorem shows that parity constraints on crossings can force planarity while preserving selected local rotation data. In the language used to motivate generalized twisted drawings, a drawing may look twisted, but if the parity structure is correct, it can be untwisted into a genuine planar embedding while retaining the intended cyclic orders at controlled vertices (Fulek et al., 2016).

A plausible implication is that generalized twisted drawings sit within a broader program in which topological graph classes are understood through combinatorial invariants rather than through fixed geometric representatives. In that broader picture, generalized twistedness is one concrete instance where a geometric definition was later converted into a rotation-system theory with local forbidden patterns, recognition algorithms, and topological certificates.