Table of Contents
Fetching ...

Packing $K_r$s in bounded degree graphs

Michael McKay, David Manlove

TL;DR

This work provides a complete complexity classification for fixed-$r$ packing problems in bounded-degree graphs, distinguishing vertex-disjoint and edge-disjoint variants. The authors introduce thresholds on the maximum degree $oldsymbol{ abla}$, showing linear-time solvability when $oldsymbol{ abla} < 3oldsymbol{r}/2 - 1$, polynomial-time solvability when $oldsymbol{ abla} < 5oldsymbol{r}/3 - 1$, and APX-hardness when $oldsymbol{ abla} \\ge \\lceil 5oldsymbol{r}/3 ceil - 1$, with the same regimes extending to EDKr for $oldsymbol{r} \\ge 6$ and adapting to $oldsymbol{r} \\le 5$ via an additional bound $oldsymbol{ abla} \\le 2oldsymbol{r}-2$. The core methodology leverages intersection graphs $oldsymbol{ his$K_r^G}$ and $oldsymbol{ his$K'_r^G}$ and, when claw-free, polynomial-time maximum independent-set algorithms to obtain packing solutions; APX-hardness is established via L-reductions from MIS-3-TF and Max 2-SAT$_{\le 3}$. These results yield a tight complexity landscape for fixed $r$ and advance the understanding of clique packing in bounded-degree graphs, with potential extensions to weighted variants and parameterized analyses.

Abstract

We study the problem of finding a maximum-cardinality set of $r$-cliques in an undirected graph of fixed maximum degree $Δ$, subject to the cliques in that set being either vertex-disjoint or edge-disjoint. It is known for $r=3$ that the vertex-disjoint (edge-disjoint) problem is solvable in linear time if $Δ=3$ ($Δ=4$) but APX-hard if $Δ\geq 4$ ($Δ\geq 5$). We generalise these results to an arbitrary but fixed $r \geq 3$, and provide a complete complexity classification for both the vertex- and edge-disjoint variants in graphs of maximum degree $Δ$. Specifically, we show that the vertex-disjoint problem is solvable in linear time if $Δ< 3r/2 - 1$, solvable in polynomial time if $Δ< 5r/3 - 1$, and APX-hard if $Δ\geq \lceil 5r/3 \rceil - 1$. We also show that if $r\geq 6$ then the above implications also hold for the edge-disjoint problem. If $r \leq 5$, then the edge-disjoint problem is solvable in linear time if $Δ< 3r/2 - 1$, solvable in polynomial time if $Δ\leq 2r - 2$, and APX-hard if $Δ> 2r - 2$.

Packing $K_r$s in bounded degree graphs

TL;DR

This work provides a complete complexity classification for fixed- packing problems in bounded-degree graphs, distinguishing vertex-disjoint and edge-disjoint variants. The authors introduce thresholds on the maximum degree , showing linear-time solvability when , polynomial-time solvability when , and APX-hardness when , with the same regimes extending to EDKr for and adapting to via an additional bound . The core methodology leverages intersection graphs K_r^G}oldsymbol{ his and, when claw-free, polynomial-time maximum independent-set algorithms to obtain packing solutions; APX-hardness is established via L-reductions from MIS-3-TF and Max 2-SAT. These results yield a tight complexity landscape for fixed and advance the understanding of clique packing in bounded-degree graphs, with potential extensions to weighted variants and parameterized analyses.

Abstract

We study the problem of finding a maximum-cardinality set of -cliques in an undirected graph of fixed maximum degree , subject to the cliques in that set being either vertex-disjoint or edge-disjoint. It is known for that the vertex-disjoint (edge-disjoint) problem is solvable in linear time if () but APX-hard if (). We generalise these results to an arbitrary but fixed , and provide a complete complexity classification for both the vertex- and edge-disjoint variants in graphs of maximum degree . Specifically, we show that the vertex-disjoint problem is solvable in linear time if , solvable in polynomial time if , and APX-hard if . We also show that if then the above implications also hold for the edge-disjoint problem. If , then the edge-disjoint problem is solvable in linear time if , solvable in polynomial time if , and APX-hard if .
Paper Structure (16 sections, 26 theorems, 5 equations, 2 figures, 1 table, 1 algorithm)

This paper contains 16 sections, 26 theorems, 5 equations, 2 figures, 1 table, 1 algorithm.

Key Result

Lemma 1

For any $U_i, U_j \in K_r^G$, if $\{ U_i, U_j \} \in E_{\mathcal{K}_r^G}$ then $|U_i \cap U_j| > r/2$.

Figures (2)

  • Figure 1: The reduction from Max $2\text{SAT}_{\leq 3}$ to EDK4
  • Figure 2: The reduction from Max $2\text{SAT}_{\leq 3}$ to EDK5

Theorems & Definitions (53)

  • proof
  • Lemma 1
  • proof
  • Lemma 2
  • proof
  • Theorem 1
  • proof
  • Lemma 3
  • proof
  • Theorem 2
  • ...and 43 more