Showing papers on "Pathwidth published in 2005"

Evolutionary dynamics on graphs

Abstract: Evolutionary dynamics have been traditionally studied in the context of homogeneous or spatially extended populations1,2,3,4. Here we generalize population structure by arranging individuals on a graph. Each vertex represents an individual. The weighted edges denote reproductive rates which govern how often individuals place offspring into adjacent vertices. The homogeneous population, described by the Moran process3, is the special case of a fully connected graph with evenly weighted edges. Spatial structures are described by graphs where vertices are connected with their nearest neighbours. We also explore evolution on random and scale-free networks5,6,7. We determine the fixation probability of mutants, and characterize those graphs for which fixation behaviour is identical to that of a homogeneous population7. Furthermore, some graphs act as suppressors and others as amplifiers of selection. It is even possible to find graphs that guarantee the fixation of any advantageous mutant. We also study frequency-dependent selection and show that the outcome of evolutionary games can depend entirely on the structure of the underlying graph. Evolutionary graph theory has many fascinating applications ranging from ecology to multi-cellular organization and economics.

Subexponential parameterized algorithms on bounded-genus graphs and H-minor-free graphs

Abstract: We introduce a new framework for designing fixed-parameter algorithms with subexponential running time---2O(√k)nO(1). Our results apply to a broad family of graph problems, called bidimensional problems, which includes many domination and problems such as vertex cover, feedback vertex set, minimum maximal matching, dominating set, edge dominating set, disk dimension, and many others restricted to bounded-genus graphs (phrased as bipartite-graph problem). Furthermore, it is fairly straightforward to prove that a problem is bidimensional. In particular, our framework includes, as special cases, all previously known problems to have such subexponential algorithms. Previously, these algorithms applied to planar graphs, single-crossing-minor-free graphs, and/or map graphs; we extend these results to apply to bounded-genus graphs as well. In a parallel development of combinatorial results, we establish an upper bound on the treewidth (or branchwidth) of a bounded-genus graph that excludes some planar graph H as a minor. This bound depends linearly on the size |V(H)| of the excluded graph H and the genus g(G) of the graph G, and applies and extends the graph-minors work of Robertson and Seymour.Building on these results, we develop subexponential fixed-parameter algorithms for dominating set, vertex cover, and set cover in any class of graphs excluding a fixed graph H as a minor. In particular, this general category of graphs includes planar graphs, bounded-genus graphs, single-crossing-minor-free graphs, and any class of graphs that is closed under taking minors. Specifically, the running time is 2O(√k)nh, where h is a constant depending only on H, which is polynomial for k = O(log2n). We introduce a general approach for developing algorithms on H-minor-free graphs, based on structural results about H-minor-free graphs at the heart of Robertson and Seymour's graph-minors work. We believe this approach opens the way to further development on problems in H-minor-free graphs.

Asymptotic Enumeration of Spanning Trees

Abstract: We give new formulas for the asymptotics of the number of spanning trees of a large graph. A special case answers a question of McKay [Europ. J. Combin. 4 149–160] for regular graphs. The general answer involves a quantity for infinite graphs that we call ‘tree entropy’, which we show is a logarithm of a normalized determinant of the graph Laplacian for infinite graphs. Tree entropy is also expressed using random walks. We relate tree entropy to the metric entropy of the uniform spanning forest process on quasi-transitive amenable graphs, extending a result of Burton and Pemantle [Ann. Probab. 21 1329–1371].

The Bidimensionality Theory and Its Algorithmic Applications 1

Abstract: Our newly developing theory of bidimensional graph problems provides general techniques for designing efficient fixed-parameter algorithms and approximation algorithms for NP-hard graph problems in broad classes of graphs. This theory applies to graph problems that are bidimensional in the sense that (1) the solution value for the k × k grid graph (and similar graphs) grows with k, typically as Ω(k2), and (2) the solution value goes down when contracting edges and optionally when deleting edges. Examples of such problems include feedback vertex set, vertex cover, minimum maximal matching, face cover, a series of vertex-removal parameters, dominating set, edge dominating set, r-dominating set, connected dominating set, connected edge dominating set, connected r-dominating set, and unweighted TSP tour (a walk in the graph visiting all vertices). Bidimensional problems have many structural properties; for example, any graph embeddable in a surface of bounded genus has treewidth bounded above by the square root of the problem's solution value. These properties lead to efficient---often subexponential---fixed-parameter algorithms, as well as polynomial-time approximation schemes, for many minor-closed graph classes. One type of minor-closed graph class of particular relevance has bounded local treewidth, in the sense that the treewidth of a graph is bounded above in terms of the diameter; indeed, we show that such a bound is always at most linear. The bidimensionality theory unifies and improves several previous results. The theory is based on algorithmic and combinatorial extensions to parts of the Robertson-Seymour Graph Minor Theory, in particular initiating a parallel theory of graph contractions. The foundation of this work is the topological theory of drawings of graphs on surfaces and our results regarding the relation (the linearity) of the size of the largest grid minor in terms of treewidth in bounded-genus graphs and more generally in graphs excluding a fixed graph H as a minor. In this thesis, we also develop the algorithmic theory of vertex separators, and its relation to the embeddings of certain metric spaces. Unlike in the edge case, we show that embeddings into L1 (and even Euclidean embeddings) are insufficient, but that the additional structure provided by many embedding theorems does suffice for our purposes. We obtain an O( logn ) approximation for min-ratio vertex cuts in general graphs, based on a new semidefinite relaxation of the problem, and a tight analysis of the integrality gap which is shown to be Θ( logn ). We also prove various approximate max-flow/min-vertex-cut theorems, which in particular give a constant-factor approximation for min-ratio vertex cuts in any excluded-minor family of graphs. Previously, this was known only for planar graphs, and for general excluded-minor families the best-known ratio was O(log n). These results have a number of applications. We exhibit an O( logn ) pseudo-approximation for finding balanced vertex separators in general graphs. Furthermore, we obtain improved approximation ratios for treewidth: In any graph of treewidth k, we show how to find a tree decomposition of width at most O(k logk ), whereas previous algorithms yielded O( k log k). For graphs excluding a fixed graph as a minor, we give a constant-factor approximation for the treewidth; this via the bidimensionality theory can be used to obtain the first polynomial-time approximation schemes for problems like minimum feedback vertex set and minimum connected dominating set in such graphs. (Copies available exclusively from MIT Libraries, Rm. 14-0551, Cambridge, MA 02139-4307. Ph. 617-253-5668; Fax 617-253-1690.)

Pattern vectors from algebraic graph theory

Abstract: Graph structures have proven computationally cumbersome for pattern analysis. The reason for this is that, before graphs can be converted to pattern vectors, correspondences must be established between the nodes of structures which are potentially of different size. To overcome this problem, in this paper, we turn to the spectral decomposition of the Laplacian matrix. We show how the elements of the spectral matrix for the Laplacian can be used to construct symmetric polynomials that are permutation invariants. The coefficients of these polynomials can be used as graph features which can be encoded in a vectorial manner. We extend this representation to graphs in which there are unary attributes on the nodes and binary attributes on the edges by using the spectral decomposition of a Hermitian property matrix that can be viewed as a complex analogue of the Laplacian. To embed the graphs in a pattern space, we explore whether the vectors of invariants can be embedded in a low-dimensional space using a number of alternative strategies, including principal components analysis (PCA), multidimensional scaling (MDS), and locality preserving projection (LPP). Experimentally, we demonstrate that the embeddings result in well-defined graph clusters. Our experiments with the spectral representation involve both synthetic and real-world data. The experiments with synthetic data demonstrate that the distances between spectral feature vectors can be used to discriminate between graphs on the basis of their structure. The real-world experiments show that the method can be used to locate clusters of graphs.

Polynomial-Time Approximation Schemes for Geometric Intersection Graphs

Abstract: A disk graph is the intersection graph of a set of disks with arbitrary diameters in the plane. For the case that the disk representation is given, we present polynomial-time approximation schemes (PTASs) for the maximum weight independent set problem (selecting disjoint disks of maximum total weight) and for the minimum weight vertex cover problem in disk graphs. These are the first known PTASs for $\mathcal{NP}$-hard optimization problems on disk graphs. They are based on a novel recursive subdivision of the plane that allows applying a shifting strategy on different levels simultaneously, so that a dynamic programming approach becomes feasible. The PTASs for disk graphs represent a common generalization of previous results for planar graphs and unit disk graphs. They can be extended to intersection graphs of other "disk-like" geometric objects (such as squares or regular polygons), also in higher dimensions.

Fixed-parameter algorithms for (k, r)-center in planar graphs and map graphs

Abstract: The (k, r)-center problem asks whether an input graph G has ≤k vertices (called centers) such that every vertex of G is within distance ≤r from some center. In this article, we prove that the (k, r)-center problem, parameterized by k and R, is fixed-parameter tractable (FPT) on planar graphs, i.e., it admits an algorithm of complexity f(k, r)nO(1) where the function f is independent of n. In particular, we show that f(k,r) = 2O(r log r) √k, where the exponent of the exponential term grows sublinearly in the number of centers. Moreover, we prove that the same type of FPT algorithms can be designed for the more general class of map graphs introduced by Chen, Grigni, and Papadimitriou. Our results combine dynamic-programming algorithms for graphs of small branchwidth and a graph-theoretic result bounding this parameter in terms of k and r. Finally, a byproduct of our algorithm is the existence of a PTAS for the r-domination problem in both planar graphs and map graphs.Our approach builds on the seminal results of Robertson and Seymour on Graph Minors, and as a result is much more powerful than the previous machinery of Alber et al. for exponential speedup on planar graphs. To demonstrate the versatility of our results, we show how our algorithms can be extended to general parameters that are “large” on grids. In addition, our use of branchwidth instead of the usual treewidth allows us to obtain much faster algorithms, and requires more complicated dynamic programming than the standard leaf/introduce/forget/join structure of nice tree decompositions. Our results are also unique in that they apply to classes of graphs that are not minor-closed, namely, constant powers of planar graphs and map graphs.

Fast deterministic distributed maximal independent set computation on growth-bounded graphs

Abstract: The distributed complexity of computing a maximal independent set in a graph is of both practical and theoretical importance While there exists an elegant O(log n) time randomized algorithm for general graphs [20], no deterministic polylogarithmic algorithm is known In this paper, we study the problem in graphs with bounded growth, an important family of graphs which includes the well-known unit disk graph and many variants thereof Particularly, we propose a deterministic algorithm that computes a maximal independent set in time O(log Δ· log*n) in graphs with bounded growth, where n and Δ denote the number of nodes and the maximal degree in G, respectively

On Domination in Graphs

Abstract: For a finite undirected graph G on n vertices two continuous optimization problems taken over the n-dimensional cube are presented and it is proved that their optimum values equal the domination number γ of G. An efficient approximation method is developed and known upper bounds on γ are slightly improved.

Combining Graph Laplacians for Semi--Supervised Learning

Abstract: A foundational problem in semi-supervised learning is the construction of a graph underlying the data. We propose to use a method which optimally combines a number of differently constructed graphs. For each of these graphs we associate a basic graph kernel. We then compute an optimal combined kernel. This kernel solves an extended regularization problem which requires a joint minimization over both the data and the set of graph kernels. We present encouraging results on different OCR tasks where the optimal combined kernel is computed from graphs constructed with a variety of distances functions and the 'k' in nearest neighbors.

Self-organizing maps for learning the edit costs in graph matching

Abstract: Although graph matching and graph edit distance computation have become areas of intensive research recently, the automatic inference of the cost of edit operations has remained an open problem. In the present paper, we address the issue of learning graph edit distance cost functions for numerically labeled graphs from a corpus of sample graphs. We propose a system of self-organizing maps (SOMs) that represent the distance measuring spaces of node and edge labels. Our learning process is based on the concept of self-organization. It adapts the edit costs in such a way that the similarity of graphs from the same class is increased, whereas the similarity of graphs from different classes decreases. The learning procedure is demonstrated on two different applications involving line drawing graphs and graphs representing diatoms, respectively.

Graphs excluding a fixed minor have grids as large as treewidth, with combinatorial and algorithmic applications through bidimensionality

Abstract: We prove that any H-minor-free graph, for a fixed graph H, of treewidth ω has an Ω(ω) × Ω(ω) grid graph as a minor. Thus grid minors suffice to certify that H-minor-free graphs have large treewidth, up to constant factors. This strong relationship was previously known for the special cases of planar graphs and bounded-genus graphs, and is known not to hold for general graphs. The approach of this paper can be viewed more generally as a framework for extending combinatorial results on planar graphs to hold on H-minor-free graphs for any fixed H. Our result has many combinatorial con-sequences on bidimensionality theory, parameter-treewidth bounds, separator theorems, and bounded local treewidth; each of these combinatorial results has several algorithmic consequences including subexponential fixed-parameter algorithms and approximation algorithms.

Power domination problem in graphs

Abstract: To monitor an electric power system by placing as few phase measurement units (PMUs) as possible is closely related to the famous vertex cover problem and domination problem in graph theory. A set S is a power dominating set (PDS) of a graph G=(V,E), if every vertex and every edge in the system is observed following the observation rules of power system monitoring. The minimum cardinality of a PDS of a graph G is the power domination number γp(G). We show that the problem of finding the power domination number for split graphs, a subclass of chordal graphs, is NP-complete. In addition, we present a linear time algorithm for finding γp(G) of an interval graph G, if the interval ordering of the graph is provided, and show that the algorithm with O(nlog n) time complexity, is asymptotically optimal, if the interval ordering is not given, where n is the number of intervals. We also show that the same results hold for the class of proper circular-arc graphs.

Learning closed sets of labeled graphs for chemical applications

Abstract: Similarity of graphs with labeled vertices and edges is naturally defined in terms of maximal common subgraphs. To avoid computation overload, a parameterized technique for approximation of graphs and their similarity is used. A lattice-based method of binarizing labeled graphs that respects the similarity operation on graph sets is proposed. This method allows one to compute graph similarity by means of algorithms for computing closed sets. Results of several computer experiments in predicting biological activity of chemical compounds that employ the proposed technique testify in favour of graph approximations as compared to complete graph representations: gaining in efficiency one (almost) does not lose in accuracy.

Graphs of bounded rank-width

Abstract: We define rank-width of graphs to investigate clique-width. Rank-width is a complexity measure of decomposing a graph in a kind of tree-structure, called a rank-decomposition. We show that graphs have bounded rank-width if and only if they have bounded clique-width. It is unknown how to recognize graphs of clique-width at most k for fixed k > 3 in polynomial time. However, we find an algorithm recognizing graphs of rank-width at most k, by combining following three ingredients. First, we construct a polynomial-time algorithm, for fixed k , that confirms rank-width is larger than k or outputs a rank-decomposition of width at most f (k) for some function f. It was known that many hard graph problems have polynomial-time algorithms for graphs of bounded clique-width, however, requiring a given decomposition corresponding to clique-width (k-expression ); we remove this requirement. Second, we define graph vertex-minors which generalizes matroid minors, and prove that if {G1, G2,…} is an infinite sequence of graphs of bounded rank-width, then there exist i < j such that Gi is isomorphic to a vertex-minor of Gj. Consequently there is a finite list Ck of graphs such that a graph has rank-width at most k if and only if none of its vertex-minors are isomorphic to a graph in Ck . Finally we construct, for fixed graph H, a modulo-2 counting monadic second-order logic formula expressing a graph contains a vertex-minor isomorphic to H. It is known that such logic formulas are solvable in linear time on graphs of bounded clique-width if the k-expression is given as an input. Another open problem in the area of clique-width is Seese's conjecture; if a set of graphs have an algorithm to answer whether a given monadic second-order logic formula is true for all graphs in the set, then it has bounded rank-width. We prove a weaker statement; if the algorithm answers for all modulo-2 counting monadic second-order logic formulas, then the set has bounded rank-width.

Posets and planar graphs

Abstract: Usually dimension should be an integer valued parameter. We introduce a refined version of dimension for graphs, which can assume a value [t - 1 ↕ t], thought to be between t - 1 and t. We have the following two results: (a) a graph is outerplanar if and only if its dimension is at most [2↕3]. This characterization of outerplanar graphs is closely related to the celebrated result of W. Schnyder [16] who proved that a graph is planar if and only if its dimension is at most 3. (b) The largest n for which the dimension of the complete graph Kn is at most [t - 1 ↕ t] is the number of antichains in the lattice of all subsets of a set of size t - 2. Accordingly, the refined dimension problem for complete graphs is equivalent to the classical combinatorial problem known as Dedekind's problem. This result extends work of Hosten and Morris [14]. The main results are enriched by background material, which links to a line of research in extremal graph theory, which was stimulated by a problem posed by G. Agnarsson: Find the maximum number of edges in a graph on n nodes with dimension at most t. © 2005 Wiley Periodicals, Inc. J Graph Theory 49: 273284, 2005

Exponential Speedup of Fixed-Parameter Algorithms for Classes of Graphs Excluding Single-Crossing Graphs as Minors

Abstract: We present a fixed-parameter algorithm that constructively solves the $k$-dominating set problem on any class of graphs excluding a single-crossing graph (a graph that can be drawn in the plane with at most one crossing) as a minor in $O(4^{9.55\sqrt{k}}n^{O(1)})$ time. Examples of such graph classes are the $K_{3,3}$-minor-free graphs and the $K_{5}$-minor-free graphs. As a consequence, we extend our results to several other problems such as vertex cover, edge dominating set, independent set, clique-transversal set, kernels in digraphs, feedback vertex set, and a collection of vertex-removal problems. Our work generalizes and extends the recent results of exponential speedup in designing fixed-parameter algorithms on planar graphs due to Alber et al. to other (nonplanar) classes of graphs.