[Paper Review] Parity Factors I: General Kotzig-Lovász Decomposition for Grafts

Nanao Kita|arXiv (Cornell University)|Dec 5, 2017

Advanced Graph Theory Research3 citations

TL;DR

This paper introduces a generalized Kotzig-Lovász decomposition for minimum $T$-joins in grafts, extending the classical decomposition from perfect matchings to parity-constrained joins. By defining an equivalence relation $\sim_{(G,T)}$ on vertices based on structural properties of minimum joins, the authors establish a canonical decomposition $\mathcal{P}(G,T)$ that refines the factor-components of a graft, generalizing the basilica decomposition theory to $T$-joins and providing a foundational tool for matching theory beyond perfect matchings.

ABSTRACT

This paper is the first from a series of papers that establish a generalization of the basilica decomposition for cardinality minimum joins in grafts. Joins in grafts are also known as $T$-joins in graphs, where $T$ is a given set of vertices, and minimum joins in grafts can be considered as a generalization of perfect matchings in graphs provided in terms of parity. The basilica decomposition is a canonical decomposition applicable to general graphs with perfect matchings, and the general Kotzig-Lovász decomposition is one of the three central concepts that compose this theory. The classical Kotzig-Lovász decomposition is a canonical decomposition for a special class of graphs known as {\em factor-connected graphs} and is famous for its contribution to the study of the matching polytope and lattice. The general Kotzig-Lovász decomposition is a nontrivial generalization of its classical counterpart and is applicable to general graphs with perfect matchings. As a component of the basilica decomposition theory, the general Kotzig-Lovász decomposition has contributed to the derivation of further results in matching theory, such as a characterization of barriers or an alternative proof of the tight cut lemma. In this paper, we present an analogue of the general Kotzig-Lovász decomposition for minimum joins in grafts.

Motivation & Objective

• To generalize the classical Kotzig-Lovász decomposition from factor-connected graphs to general grafts with $T$-joins.
• To establish a canonical decomposition for minimum $T$-joins using an equivalence relation $\sim_{(G,T)}$ on vertices.
• To embed this decomposition within the broader framework of the basilica decomposition theory for $T$-joins.
• To provide a refinement of the factor-component structure that captures deeper structural properties of $T$-joins beyond isolated components.

Proposed method

• Define an equivalence relation $\sim_{(G,T)}$ on $V(G)$ based on the existence of paths with non-positive weight relative to a minimum $T$-join $F$, using a distance decomposition from a root vertex.
• Construct a path $\hat{P}$ in $G$ from a path $P$ in a comb-bipartite graft $Q_r'$, ensuring $w_F(\hat{P}) \leq -1$ to verify transitivity of $\sim_{(G,T)}$.
• Prove that $\sim_{(G,T)}$ is an equivalence relation by verifying reflexivity, symmetry, and transitivity using path constructions and weight constraints.
• Define the general Kotzig-Lovász decomposition $\mathcal{P}(G,T)$ as the family of equivalence classes under $\sim_{(G,T)}$, which partitions the vertex set of the graft.
• Show that $\mathcal{P}(H;G,T)$, the restriction of $\mathcal{P}(G,T)$ to a factor-component $H$, is a refinement of $\mathcal{P}(H, T \cap V(H))$, indicating a non-trivial global structure.
• Use Sebö’s distance decomposition and path lifting techniques to relate local properties in subgraphs to global equivalence classes.

Experimental results

Research questions

• RQ1 How can the classical Kotzig-Lovász decomposition for perfect matchings be generalized to $T$-joins in grafts?
• RQ2 What canonical decomposition structure exists for minimum $T$-joins that captures the global structure of parity-constrained edge sets?
• RQ3 Is there a well-defined equivalence relation on vertices of a graft that characterizes the structural components of minimum $T$-joins?
• RQ4 How does the decomposition of a graft’s $T$-joins relate to the decompositions of its individual factor-components?
• RQ5 Can the basilica decomposition framework be extended from perfect matchings to $T$-joins via a generalized Kotzig-Lovász decomposition?

Key findings

• The binary relation $\sim_{(G,T)}$ is proven to be an equivalence relation on $V(G)$, establishing a canonical decomposition of the vertex set into equivalence classes.
• The family $\mathcal{P}(G,T)$ of equivalence classes forms the general Kotzig-Lovász decomposition of the graft $(G,T)$, which is a refinement of the decomposition of individual factor-components.
• For each factor-component $H$, the restriction $\mathcal{P}(H;G,T)$ is a proper refinement of $\mathcal{P}(H, T \cap V(H))$, indicating a more detailed structural hierarchy than component-wise decomposition.
• The decomposition generalizes Sebö’s announced result on the $T$-join analogue of the classical Kotzig-Lovász decomposition, confirming its validity and extending its scope.
• The construction of a path $\hat{P}$ with $w_F(\hat{P}) \leq -1$ from a path in a comb-bipartite graft provides a key technical tool for proving transitivity and structural consistency.
• The decomposition is not a disjoint union of component-wise decompositions but exhibits a global, inter-component structure, highlighting its non-trivial refinement over classical analogues.