A simulation tool for the performance evaluation of parallel branch and bound algorithms

Parallel computation offers a challenging opportunity to speed up the time consuming enumerative procedures that are necessary to solve hard combinatorial problems. Theoretical analysis of such a parallel branch and bound algorithm is very hard and empirical analysis is not straightforward because the performance of a parallel algorithm cannot be evaluated simply by executing the algorithm on a few parallel systems. Among the difficulties encountered are the noise produced by other users on the system, the limited variation in parallelism (the number of processors in the system is strictly bounded) and the waste of resources involved: most of the time, the outcomes of all computations are already known and the only issue of interest is when these outcomes are produced.We will describe a way to simulate the execution of parallel branch and bound algorithms on arbitrary parallel systems in such a way that the memory and cpu requirements are very reasonable. The use of simulation has only minor consequences for the formulation of the algorithm.     

A random 1-011-011-01algorithm for depth first search

In this paper we present a fast parallel algorithm for constructing a depth first search tree for an undirected graph. The algorithm is anRNC algorithm, meaning that it is a probabilistic algorithm that runs in polylog time using a polynomial number of processors on aP-RAM. The run time of the algorithm isO(T _MM(n) log³ n), and the number of processors used isP _MM (n) whereT _MM(n) andP _MM(n) are the time and number of processors needed to find a minimum weight perfect matching on ann vertex graph with maximum edge weightn.This research was done while the first author was visiting the Mathematical Research Institute in Berkeley. Research supported in part by NSF grant 8120790.Supported by Air Force Grant AFOSR-85-0203A.     

On the complexity of the disjoint paths problem

In this paper we consider the disjoint paths problem. Given a graphG and a subsetS of the edge-set ofG the problem is to decide whether there exists a family of disjoint circuits inG each containing exactly one edge ofS such that every edge inS belongs to a circuit inC. By a well-known theorem of P. Seymour the edge-disjoint paths problem is polynomially solvable for Eulerian planar graphsG. We show that (assumingPNP) one can drop neither planarity nor the Eulerian condition onG without losing polynomial time solvability. We prove theNP-completeness of the planar edge-disjoint paths problem by showing theNP-completeness of the vertex disjoint paths problem for planar graphs with maximum vertex-degree three. This disproves (assumingPNP) a conjecture of A. Schrijver concerning the existence of a polynomial time algorithm for the planar vertex-disjoint paths problem. Furthermore we present a counterexample to a conjecture of A. Frank. This conjecture would have implied a polynomial algorithm for the planar edge-disjoint paths problem. Moreover we derive a complete characterization of all minorclosed classes of graphs for which the disjoint paths problem is polynomially solvable. Finally we show theNP-completeness of the half-integral relaxation of the edge-disjoint paths problem. This implies an answer to the long-standing question whether the edge-disjoint paths problem is polynomially solvable for Eulerian graphs.Supported by Sonderforschungsbereich 303 (DFG)     

A new graph triconnectivity algorithm and its parallelization

We present a new algorithm for finding the triconnected components of an undirected graph. The algorithm is based on a method of searching graphs called open ear decomposition. A parallel implementation of the algorithm on a CRCW PRAM runs inO(log² n) parallel time usingO(n+m) processors, wheren is the number of vertices andm is the number of edges in the graph.A preliminary version of this paper was presented at the19th Annual ACM Symposium on Theory of Computing, New York, NY, May 1987.Supported by NSF Grant DCR 8514961.Supported by NSF Grant ECS 8404866 and the Semiconductor Research Corporation Grant 86-12-109.     

An analysis of the greedy algorithm for the submodular set covering problem

We consider the problem: min \(\{ \mathop \Sigma \limits_{j \in s} f_j :z(S) = z(N),S \subseteqq N\} \) wherez is a nondecreasing submodular set function on a finite setN. Whenz is integer-valued andz(Ø)=0, it is shown that the value of a greedy heuristic solution never exceeds the optimal value by more than a factor \(H(\mathop {\max }\limits_j z(\{ j\} ))\) where \(H(d) = \sum\limits_{i = 1}^d {\frac{1}{i}} \) . This generalises earlier results of Dobson and others on the applications of the greedy algorithm to the integer covering problem: min {fy: Ay ≧b, y ε {0, 1}} wherea _ij ,b _i } ≧ 0 are integer, and also includes the problem of finding a minimum weight basis in a matroid.     

A greedy algorithm for the optimal basis problem

The following problem is considered. Givenm+1 points {x _i }₀ ^m inR ⁿ which generate anm-dimensional linear manifold, construct for this manifold a maximally linearly independent basis that consists of vectors of the formx _i −x _j . This problem is present in, e.g., stable variants of the secant and interpolation methods, where it is required to approximate the Jacobian matrixf′ of a nonlinear mappingf by using values off computed atm+1 points. In this case, it is also desirable to have a combination of finite differences with maximal linear independence. As a natural measure of linear independence, we consider the hadamard condition number which is minimized to find an optimal combination ofm pairs {x _i ,x _j }. We show that the problem is not NP-hard, but can be reduced to the minimum spanning tree problem, which is solved by the greedy algorithm inO(m ²) time. The complexity of this reduction is equivalent to onem×n matrix-matrix multiplication, and according to the Coppersmith-Winograd estimate, is belowO(n ^2.376) form=n. Applications of the algorithm to interpolation methods are discussed. Part of the work was done while the author was visiting DIMACS, an NSF Science and Technology Center funded under contract STC-91-19999; DIMACS is a cooperative project of Rutgers University, Princeton University, AT&T Bell Laboratories and Bellcore, NJ, USA.     

Spanning multi-paths in hypercubes

A set A of vertices of a hypercube is called balanced if . We prove that for every natural number n there exists a natural number π₁(n) such that for every hypercube Q with dim(Q)?π₁(n) there exists a family of pairwise vertex-disjoint paths P_i between A_i and B_i for i=1,2,…,n with if and only if {A_i,B_i∣i=1,2,…,n} is a balanced set.     

A note on strong embeddings of maximal planar graphs on non-orientable sufraces

Abstract. In this paper, it is shown that for every maximal planar graph     

A primal-dual approximation algorithm for generalized steiner network problems

We present the first polynomial-time approximation algorithm for finding a minimum-cost subgraph having at least a specified number of edges in each cut. This class of problems includes, among others, the generalized Steiner network problem, also called the survivable network design problem. Ifk is the maximum cut requirement of the problem, our solution comes within a factor of 2k of optimal. Our algorithm is primal-dual and shows the importance of this technique in designing approximation algorithms.Research supported by an NSF Graduate Fellowship, DARPA contracts N00014-91-J-1698 and N00014-92-J-1799, and AT&T Bell Laboratories.Research supported in part by Air Force contract F49620-92-J-0125 and DARPA contract N00014-92-J-1799.Part of this work was done while the author was visiting AT&T Bell Laboratories and Bellcore.     

A linear-time algorithm for edge-disjoint paths in planar graphs

In this paper we discuss the problem of finding edge-disjoint paths in a planar, undirected graph such that each path connects two specified vertices on the boundary of the graph. We will focus on the “classical” case where an instance additionally fulfills the so-calledevenness-condition. The fastest algorithm for this problem known from the literature requiresO (n ^5/3(loglogn)^1/3) time, wheren denotes the number of vertices. In this paper now, we introduce a new approach to this problem, which results in anO(n) algorithm. The proof of correctness immediately yields an alternative proof of the Theorem of Okamura and Seymour, which states a necessary and sufficient condition for solvability.     

An augmenting path algorithm for linear matroid parity

Linear matroid parity generalizes matroid intersection and graph matching (and hence network flow, degree-constrained subgraphs, etc.). A polynomial algorithm was given by Lovász. This paper presents an algorithm that uses timeO(mn ³), wherem is the number of elements andn is the rank. (The time isO(mn ^2.5) using fast matrix multiplication; both bounds assume the uniform cost model). For graphic matroids the time isO(mn ²). The algorithm is based on the method of augmenting paths used in the algorithms for all subcases of the problem. First author was supported in part by the National Science Foundation under grants MCS 78-18909, MCS-8302648, and DCR-8511991. The research was done while the second author was at the University of Denver and at the University of Colorado at Boulder.     

A subexponential algorithm for the coloured tree partition problem

Given a tree of n vertices and a list of feasible colours for each vertex, the coloured tree partition problem (CTPP) consists in partitioning the tree into p vertex-disjoint subtrees of minimum total cost, and assigning to each subtree a different colour, which must be feasible for all of its vertices. The problem is strongly NP-hard on general graphs, as well as on grid and bipartite graphs. This paper deals with the previously open case of tree graphs, showing that it is strongly NP-complete to determine whether a feasible solution exists. It presents reduction, decomposition and bounding procedures to simplify the problem and an exact algorithm of complexity (with ) for the special case in which a vertex of each subtree is given.     

A fast parallel algorithm to compute the rank of a matrix over an arbitrary field

It is shown that the rank of a matrix over an arbitrary field can be computed inO(log² n) time using a polynomial number of processors. Also appeared in ACM Symposium on Theory of Computing, May 28–30, 1986 Berkeley, California. Research supported by Miller Fellowship, University of California, Berkeley.     

A las vegas rnc algorithm for maximum matching

Recently two randomized algorithms were discovered that find a maximum matching in an arbitrary graph in polylog time, when run on a parallel random access machine. Both are Monte Carlo algorithms — they have the drawback that with non-zero probability the output is a non-maximum matching. We use the min-max formula for the size of a maximum matching to convert any Monte Carlo maximum matching algorithm into a Las Vegas (error-free) one. The resulting algorithm returns (with high probability) a maximum matching and a certificate proving that the matching is indeed maximum. Research supported by DARPA grant N00039-84-C-0098 and by a US Army Research Office fellowship.     

Cycles and paths in bipartite tournaments with spanning configurations

We give necessary and sufficient conditions in terms of connectivity and factors for the existence of hamiltonian cycles and hamiltonian paths and also give sufficient conditions in terms of connectivity for the existence of cycles through any two vertices in bipartite tournaments.     

The complexity of detecting crossingfree configurations in the plane

The computational complexity of the following type of problem is studied. Given a geometric graphG=(P, S) whereP is a set of points in the Euclidean plane andS a set of straight (closed) line segments between pairs of points inP, we want to know whetherG possesses a crossingfree subgraph of a special type. We analyze the problem of detecting crossingfree spanning trees, one factors and two factors in the plane. We also consider special restrictions on the slopes and on the lengths of the edges in the subgraphs.Klaus Jansen acknowledges support by the Deutsche Forschungsgemeinschaft. Gerhard J. Woeginger acknowledges support by the Christian Doppler Laboratorium für Diskrete Optimierung.     

On the maximum cardinality search lower bound for treewidth

The Maximum Cardinality Search (MCS) algorithm visits the vertices of a graph in some order, such that at each step, an unvisited vertex that has the largest number of visited neighbours becomes visited. A maximum cardinality search ordering (MCS-ordering) of a graph is an ordering of the vertices that can be generated by the MCS algorithm. The visited degree of a vertex v in an MCS-ordering is the number of neighbours of v that are before v in the ordering. The visited degree of an MCS-ordering ψ of G is the maximum visited degree over all vertices v in ψ. The maximum visited degree over all MCS-orderings of graph G is called its maximum visited degree. Lucena [A new lower bound for tree-width using maximum cardinality search, SIAM J. Discrete Math. 16 (2003) 345-353] showed that the treewidth of a graph G is at least its maximum visited degree.We show that the maximum visited degree is of size O(logn) for planar graphs, and give examples of planar graphs G with maximum visited degree k with O(k!) vertices, for all k∈N. Given a graph G, it is NP-complete to determine if its maximum visited degree is at least k, for any fixed k?7. Also, this problem does not have a polynomial time approximation algorithm with constant ratio, unless P=NP. Variants of the problem are also shown to be NP-complete.In this paper, we also propose some heuristics for the problem, and report on an experimental analysis of them. Several tiebreakers for the MCS algorithm are proposed and evaluated. We also give heuristics that give upper bounds on the value of the maximum visited degree of a graph, which appear to give results close to optimal on many graphs from real life applications.     

Complexity of the Frobenius problem

Consider the Frobenius Problem: Given positive integersa ₁,...,a _n witha _i 2 and such that their greatest common divisor is one, find the largest natural number that is not expressible as a non-negative integer combination ofa ₁,...,a _n. In this paper we prove that the Frobenius problem is NP-hard, under Turing reductions.