Work

Parameterized Algorithms for GraphPartitioning Problems

Year: 2018
Type: dissertation
Abstract: In parameterized complexity, a problem instance (I, k) consists of an input I and an \nextra parameter k. The parameter k usually a positive integer indicating the size of the \nsolution or the structure of the input. A computational problem is called fixed-parameter \ntractable (FPT) if there is an algorithm for the problem with time complexity O(f(k).nc \n), \nwhere f(k) is a function dependent only on the input parameter k, n is the size of the \ninput and c is a constant. The existence of such an algorithm means that the problem \nis tractable for fixed values of the parameter. In this thesis, we provide parameterized \nalgorithms for the following NP-hard graph partitioning problems: \n(i) Matching Cut Problem: In an undirected graph, a matching cut is a partition \nof vertices into two non-empty sets such that the edges across the sets induce a matching. \nThe matching cut problem is the problem of deciding whether a given graph has \na matching cut. The Matching Cut problem is expressible in monadic second-order \nlogic (MSOL). The MSOL formulation, together with Courcelle’s theorem implies linear \ntime solvability on graphs with bounded tree-width. However, this approach leads to a \nrunning time of f(||ϕ||, t) · n, where ||ϕ|| is the length of the MSOL formula, t is the \ntree-width of the graph and n is the number of vertices of the graph. The dependency of \nf(||ϕ||, t) on ||ϕ|| can be as bad as a tower of exponentials. \nIn this thesis we give a single exponential algorithm for the Matching Cut problem \nwith tree-width alone as the parameter. The running time of the algorithm is 2O(t) \n· n. \nThis answers an open question posed by Kratsch and Le [Theoretical Computer Science, \n2016]. We also show the fixed parameter tractability of the Matching Cut problem \nwhen parameterized by neighborhood diversity or other structural parameters. \n(ii) H-Free Coloring Problems: In an undirected graph G for a fixed graph H, \nthe H-Free q-Coloring problem asks to color the vertices of the graph G using at \nmost q colors such that none of the color classes contain H as an induced subgraph. \nThat is every color class is H-free. This is a generalization of the classical q-Coloring \nproblem, which is to color the vertices of the graph using at most q colors such that no \npair of adjacent vertices are of the same color. The H-Free Chromatic Number is \nthe minimum number of colors required to H-free color the graph. \nFor a fixed q, the H-Free q-Coloring problem is expressible in monadic secondorder \nlogic (MSOL). The MSOL formulation leads to an algorithm with time complexity \nf(||ϕ||, t) · n, where ||ϕ|| is the length of the MSOL formula, t is the tree-width of the \ngraph and n is the number of vertices of the graph. \nIn this thesis we present the following explicit combinatorial algorithms for H-Free \nColoring problems: \n \n• An O(q \nO(t \nr \n) \n· n) time algorithm for the general H-Free q-Coloring problem, \nwhere r = |V (H)|. \n• An O(2t+r log t \n· n) time algorithm for Kr-Free 2-Coloring problem, where Kr is \na complete graph on r vertices. \nThe above implies an O(t \nO(t \nr \n) \n· n log t) time algorithm to compute the H-Free Chromatic \nNumber for graphs with tree-width at most t. Therefore H-Free Chromatic \nNumber is FPT with respect to tree-width. \nWe also address a variant of H-Free q-Coloring problem which we call H-(Subgraph)Free \nq-Coloring problem, which is to color the vertices of the graph such that none of the \ncolor classes contain H as a subgraph (need not be induced). \nWe present the following algorithms for H-(Subgraph)Free q-Coloring problems. \n• An O(q \nO(t \nr \n) \n· n) time algorithm for the general H-(Subgraph)Free q-Coloring \nproblem, which leads to an O(t \nO(t \nr \n) \n· n log t) time algorithm to compute the H- \n(Subgraph)Free Chromatic Number for graphs with tree-width at most t. \n• An O(2O(t \n2 \n) \n· n) time algorithm for C4-(Subgraph)Free 2-Coloring, where C4 \nis a cycle on 4 vertices. \n• An O(2O(t \nr−2 \n) \n· n) time algorithm for {Kr\\e}-(Subgraph)Free 2-Coloring, \nwhere Kr\\e is a graph obtained by removing an edge from Kr. \n• An O(2O((tr2 \n) \nr−2 \n) \n· n) time algorithm for Cr-(Subgraph)Free 2-Coloring problem, \nwhere Cr is a cycle of length r. \n(iii) Happy Coloring Problems: In a vertex-colored graph, an edge is happy if its \nendpoints have the same color. Similarly, a vertex is happy if all its incident edges are \nhappy. we consider the algorithmic aspects of the following Maximum Happy Edges \n(k-MHE) problem: given a partially k-colored graph G, find an extended full k-coloring \nof G such that the number of happy edges are maximized. When we want to maximize \nthe number of happy vertices, the problem is known as Maximum Happy Vertices \n(k-MHV). \nWe show that both k-MHE and k-MHV admit polynomial-time algorithms for trees. \nWe show that k-MHE admits a kernel of size k + `, where ` is the natural parameter, \nthe number of happy edges. We show the hardness of k-MHE and k-MHV for some \nspecial graphs such as split graphs and bipartite graphs. We show that both k-MHE \nand k-MHV are tractable for graphs with bounded tree-width and graphs with bounded \nneighborhood diversity. \nvii \nIn the last part of the thesis we present an algorithm for the Replacement Paths \nProblem which is defined as follows: Let G (|V (G)| = n and |E(G)| = m) be an undirected \ngraph with positive edge weights. Let PG(s, t) be a shortest s − t path in G. Let l be the \nnumber of edges in PG(s, t). The Edge Replacement Path problem is to compute a \nshortest s − t path in G\\{e}, for every edge e in PG(s, t). The Node Replacement \nPath problem is to compute a shortest s−t path in G\\{v}, for every vertex v in PG(s, t). \nWe present an O(TSP T (G) + m + l \n2 \n) time and O(m + l \n2 \n) space algorithm for both \nthe problems, where TSP T (G) is the asymptotic time to compute a single source shortest \npath tree in G. The proposed algorithm is simple and easy to implement.
Language: English
Open Access status: green
OpenAlex ID: W2887949973
Cites: 73
Cited by:
Related to: 9