Source-linked AI summary
Odd Cycle Transversal on $H$-free graphs
Esther Galby, Paloma T. de Lima, Andrea Munaro, Amir Nikabadi
TL;DR
The paper addresses the remaining complexity cases for Odd Cycle Transversal and related list-coloring optimization on H-free graphs. It constructs H-amiable families and uses them in a quasi-polynomial-time algorithm on kP4-free graphs. Together with prior results, this gives a complete quasi-polynomial-time-versus-NP-hard dichotomy for the two-color problems.
Problem
The complexity classification for Odd Cycle Transversal and related coloring problems on H-free graphs remained incomplete, with open cases including H induced by kP4.
Source: §1 · pp. 2–3Method
The paper constructs quasi-polynomial-size H-amiable families for kH-free graphs and uses a P4-amiable family with a two-layer auxiliary graph to solve the two-color problem.
Source: §1.1 · p. 4, §1.2 · p. 6Results
Both two-color problems are polynomial-time solvable for H induced by kP3 or P5+kP1, quasi-polynomial-time solvable in the remaining cases where H is induced by kP4, and NP-hard otherwise.
Source: §1.1 · p. 4Takeaways & Limitations
The dichotomy resolves the open problem of Agrawal, Lima, Lokshtanov, Saurabh, and Sharma for Odd Cycle Transversal on H-free graphs.
Source: §5 · p. 14Whether the quasi-polynomial algorithms on kP4-free graphs can be improved to polynomial time remains open.
Source: §5 · p. 14
Abstract
from arXiv · showhide
\textsc{Odd Cycle Transversal} is a classic $\mathsf{NP}$-hard graph optimization problem asking for a minimum-weight set of vertices whose deletion makes the input graph bipartite, or equivalently, a maximum-weight induced bipartite subgraph. We show that \textsc{Odd Cycle Transversal} is quasi-polynomial-time solvable on $kP_4$-free graphs, for every fixed $k \in \mathbb{N}$. In fact, we provide an $n^{O_k(\log n)}$-time algorithm for the more general \textsc{Max-Weight List $2$-Colorable Induced Subgraph}, where the notation $O_{k}(\cdot)$ hides factors depending on $k$. Paired with known results from the literature, this allows us to obtain a complete complexity dichotomy for these two problems on $H$-free graphs into cases solvable in quasi-polynomial time and cases which are $\mathsf{NP}$-hard, in particular resolving an open problem of Agrawal, Lima, Lokshtanov, Saurabh, and Sharma [SODA 2024]. Our algorithms are based on a new structural tool that may be of independent interest. We introduce the notion of $H$-amiable family and show that, for every fixed graph $H$ without isolated vertices and every fixed $k\ge2$, every $kH$-free graph admits an $H$-amiable family of quasi-polynomial size that can be constructed in quasi-polynomial time. Besides yielding the aforementioned algorithms, this result gives, for every fixed connected graph $H$ and every fixed $k\ge2$, a reduction from \textsc{Max-Weight Independent Set} on $kH$-free graphs to the same problem on $H$-free graphs with $n^{O_{H,k}(\log n)}$ overhead. In this setting, it improves the $n^{O_{H,k}(\log^3 n)}$ overhead obtained by specializing the general reduction of Gartland and Lokshtanov [FOCS 2020].
1 Introduction
The paper advances the complexity classification of induced-subgraph coloring problems, proving quasi-polynomial-time algorithms on kP4-free graphs and completing the H-free dichotomy for the two-color case. Its structural tool, H-amiable families, also yields a sharper reduction for Max-Weight Independent Set on kH-free graphs.
1 Introduction
- Earlier results established polynomial-time cases for induced subgraphs of kP3 or P5+kP1 and NP-hardness when H contains P6 or P5+P2; for r ≥5, hardness also holds when H contains P4+P2.For r ≥5, these results already give a complete polynomial-time-versus-NP-hard dichotomy; the remaining open cases for r ∈{2, 3, 4} involved H induced by kP4.Source: §1 · p. 3
1.1 Our results
- For every fixed k, Max-Weight List 2-Colorable Induced Subgraph has an n^O_k(log n)-time algorithm on kP4-free graphs, and Odd Cycle Transversal follows as a special case.The running-time exponent hides factors depending on k.Source: §1.1 · p. 3
- For both problems on H-free graphs, they are polynomial-time solvable when H is an induced subgraph of kP3 or P5 + kP1 for some k ≥ 1, quasi-polynomial-time solvable in the remaining cases where H is an induced subgraph of kP4 for some k ≥ 1, and NP-hard otherwise.This resolves [1, Problem 4, p. 5288].Source: §1.1 · p. 4
- Theorem 3 classifies both two-color problems as polynomial-time solvable for H induced by kP3 or P5+kP1, quasi-polynomial-time solvable for the remaining H induced by kP4, and NP-hard otherwise.The hardness for H that is not a linear forest follows from the result of Chiarelli, Hartinger, Johnson, Milanič, and Paulusma [4].Source: §1.1 · p. 4
- For every fixed H without isolated vertices and fixed k ≥2, each n-vertex kH-free graph has a constructible H-amiable family of size n^O_H,k(log n).The family is a key ingredient in the proof of the two-color algorithm.Source: §1.1 · p. 4
- For connected H and fixed k ≥2, an oracle for Max-Weight Independent Set on H-free graphs gives an n^O_H,k(log n)-overhead reduction from kH-free graphs.This improves the n^O_H(log^3 n) overhead of the general reduction in [12].Source: §1.1 · p. 5
1.2 Technical overview
- A branching procedure selects a maximum-score vertex, splitting independent sets into those avoiding it and those containing it; the second branch removes a constant fraction of induced H-copies.The resulting recurrence bounds the family construction by quasi-polynomial size and time, while the absence of isolated vertices in H preserves H-freeness when the selected vertex is added.Source: §1.2 · p. 6
- A two-layer auxiliary graph encodes feasible list colorings as stable sets and is perfect when both input subsets induce P4-free graphs.The algorithm enumerates quasi-polynomially many subset pairs using the P4-amiable family, solves maximum-weight stable set on each perfect graph, and returns the largest solution.Source: §1.2 · pp. 6–7
2 Preliminaries
The preliminaries fix graph and set notation used throughout. They define induced subgraphs, neighborhoods, and the standard adjacency relations between vertex sets.
- The paper denotes the positive integers by N and writes [n] for {1, …, n}.Source: §2 · p. 7
- For a vertex set X, G[X] is the subgraph induced by X; X may also denote that induced subgraph.Source: §2 · p. 7
- N_G(v) and N_G[v] denote the open and closed neighborhoods of v, respectively.Source: §2 · p. 7
- Vertex sets are complete to one another when all cross-pairs are edges, and anticomplete when none are.Source: §2 · p. 7
- An isomorphism preserves adjacency and non-adjacency, and is written φ: G → G′.Source: §2 · p. 7
3 Building H-amiable families
The section constructs a small H-amiable family for each kH-free graph by recursively branching on a vertex selected using its score. The family can be generated in quasi-polynomial time and supports the subsequent algorithmic reductions.
- Every n-vertex kH-free graph admits an H-amiable family of size n^O_H,k(log n), constructible in time n^O_H,k(log n).Source: §3 · p. 7
- The recursion chooses a maximum-score vertex; its score counts H-copies meeting its closed neighborhood, and the selected vertex has score at least µ(F)/((k−1)h).Source: §3 · pp. 7–8
- An H-amiable family consists of H-free induced subgraphs and contains a member covering every independent set of the input graph.Source: §3 · p. 8
- The recursion has two branches, on F−v_F and F−N_F[v_F], with v_F added to the selected set on the second branch.Source: §3 · p. 9
- The distinct outputs at the recursion leaves are precisely the members of the constructed family, and lazy construction with deduplication achieves the stated quasi-polynomial running time.Source: §3 · p. 9
4 The proof of Theorem 1
The proof reduces list 2-colorable induced-subgraph optimization to stable-set computations on perfect auxiliary graphs. Enumerating a quasi-polynomial family of containers then yields an optimal solution on every fixed-kP4-free graph.
4 The proof of Theorem 1
- Stable sets of the two-layer auxiliary graph correspond weight-preservingly to feasible list-colored induced subgraphs whose color classes lie in the chosen containers.With both containers equal to V(G), maximizing stable-set weight solves the original optimization problem.Source: §4 · p. 10
- When both containers induce P4-free graphs, the two-layer list auxiliary graph is perfect.This permits polynomial-time maximum-weight stable-set computation on each auxiliary graph.Source: §4 · p. 11
- The perfectness proof excludes odd holes by bounding layer-segment lengths and excludes long odd antiholes because each layer contributes at most three vertices.With no odd holes or odd antiholes, the Strong Perfect Graph Theorem gives perfectness.Source: §4 · pp. 11–12
- For every fixed k, Max-Weight List 2-Colorable Induced Subgraph is solvable optimally on kP4-free graphs in quasi-polynomial time.The stated running time is n^{O_k(log n)}.Source: §4 · p. 13
- The algorithm enumerates ordered pairs from a quasi-polynomial-size P4-amiable family and solves a maximum-weight stable-set problem for each pair.The family covers every stable set, and its members induce P4-free graphs, making the auxiliary graphs perfect.Source: §4 · p. 13
- The enumerated pair containing the optimum coloring’s two stable color classes produces a feasible candidate of at least optimal weight, so the best candidate is optimal.The stable-set and coloring correspondence preserves weights.Source: §4 · pp. 10, 13
5 Concluding remarks
The quasi-polynomial-time result for kP4-free graphs, combined with known results, resolves the stated Odd Cycle Transversal dichotomy problem. The conclusion leaves polynomial-time solvability and further dichotomies as open directions.
5 Concluding remarks
- The result for kP4-free graphs, combined with known literature results, resolves the open quasi-polynomial-time-versus-NP-hard dichotomy for Odd Cycle Transversal on H-free graphs.The paper identifies this as resolving an open problem posed by Agrawal, Lima, Lokshtanov, Saurabh, and Sharma.Source: §5 · p. 14
- Whether the algorithms for Odd Cycle Transversal and Max-Weight List 2-Colorable Induced Subgraph can be made polynomial-time on kP4-free graphs remains open.The paper connects a positive answer to completing the P-versus-NP-hard dichotomy for both problems on H-free graphs.Source: §5 · p. 14
- Similar dichotomies for Max-Weight List r-Colorable Induced Subgraph remain to be established for r in {3, 4}.The paper also notes that no quasi-polynomial-time algorithm for List 4-Coloring on kP4-free graphs is known to its authors.Source: §5 · p. 14
- Extending the method to bounded induced-P4-packing treewidth would require a local perfectness argument or another way to solve the residual instances.The current lemma requires both auxiliary-graph layers to be globally P4-free.Source: §5 · p. 14