On the Recognition of Fuzzy Circular Interval Graphs

Fuzzy circular interval graphs are a generalization of proper circular arc graphs and have been recently introduced by Chudnovsky and Seymour as a fundamental subclass of claw-free graphs. In this paper, we provide a polynomial-time algorithm for recognizing such graphs, and more importantly for building a suitable representation.

💡 Research Summary

**
The paper addresses the algorithmic problem of recognizing fuzzy circular interval graphs (FCIGs), a class that generalizes proper circular arc (or circular interval) graphs and forms one of the two fundamental building blocks of claw‑free graphs, as shown by Chudnovsky and Seymour. While FCIGs had been defined structurally and were known to be claw‑free, no polynomial‑time recognition algorithm that also constructs a concrete representation was previously available.

The authors first recall the definitions of circular interval graphs (CIGs) and FCIGs. A CIG is obtained by placing a set of vertices on a circle and selecting a family of non‑nested intervals; two vertices are adjacent precisely when they lie in a common interval. An FCIG augments this model by allowing “fuzzy” intervals whose endpoints correspond to two vertices; for such a pair the adjacency may be chosen arbitrarily, while all other adjacencies are still dictated by the interval family.

A central technical tool is the notion of proper and homogeneous pairs of cliques. A proper pair (K₁, K₂) consists of two disjoint cliques such that every vertex of K₁ (resp. K₂) is neither complete nor anti‑complete to K₂ (resp. K₁). A homogeneous pair requires that every vertex outside K₁∪K₂ is either complete or anti‑complete to each of the two cliques. The paper also introduces almost‑proper pairs (a relaxation of proper pairs) and the concept of a tight representation with respect to a pair of cliques, meaning that all vertices of one clique are mapped to a single point on the circle and the interval connecting the two points belongs to the interval family.

The algorithm proceeds in three conceptual phases:

1. Normalization of fuzzy pairs. Lemma 2.6 shows that any FCIG can be transformed in O(n²) time into an equivalent representation where every fuzzy pair of cliques is both proper and homogeneous. This is achieved by moving vertices along the circle while preserving adjacency constraints.

2. Reduction of proper‑homogeneous pairs. For a proper‑homogeneous pair (K₁, K₂) the authors define a reduction operation: collapse each clique into a single “super‑vertex” and replace the two cliques and the fuzzy interval between them by a single interval. Lemma 4.1 proves that this operation preserves the FCIG property. By repeatedly applying this reduction to all proper‑homogeneous pairs (there can be at most O(n) of them), the original graph G is transformed into a graph G′ that contains no proper‑homogeneous pair.

3. Final recognition and reconstruction. Lemma 2.7 guarantees that a graph without proper‑homogeneous pairs is exactly a circular interval graph. Consequently, the authors invoke any of the known linear‑time algorithms for CIG recognition (e.g., the O(n + m) algorithm of Deng, Hell, and Huang) to obtain a proper circular arc representation of G′. Because each reduction step records how the super‑vertices correspond to original cliques, the representation of G′ can be lifted back to a full (Φ, I) representation of the original FCIG, including the fuzzy adjacencies.

The complexity analysis shows that each reduction step costs O(n + m) time, and at most O(n) steps are needed, leading to an overall bound of O(n² m). In practice, for sparse graphs the algorithm behaves close to linear time. Memory consumption remains linear, O(n + m).

Key technical contributions include:

• The introduction of almost‑proper pairs and the proof (Theorem 3.14) that any such pair can be made tight in O(n²) time, which is essential for the reduction step.
• Lemma 3.7, which guarantees that any clique of size ≥2 is either covered by a single interval or forces the existence of three intervals covering the whole circle; this structural insight underpins the detection of fuzzy dominating pairs.
• A constructive proof that the reduction operation does not introduce new forbidden configurations, preserving the claw‑free nature and the FCIG definition.

The paper’s significance lies in providing the first constructive polynomial‑time algorithm for FCIG recognition that simultaneously yields a concrete geometric representation. This bridges a gap between structural graph theory (where FCIGs were known only via decomposition theorems) and algorithmic applications. Potential applications include scheduling problems on circular time slots, bio‑informatics where circular genomes induce circular interval structures, and network design where “fuzzy” adjacency captures optional links.

In conclusion, the authors deliver a complete algorithmic framework: normalize fuzzy pairs, iteratively reduce proper‑homogeneous pairs, recognize the resulting circular interval graph, and reconstruct the original fuzzy representation. The method is theoretically sound, practically implementable, and opens avenues for further algorithmic work on related graph classes such as fuzzy linear interval graphs or broader claw‑free subclasses.