A direct Schur–Fourier decomposition for the efficient solution of high‐order Poisson equations on loosely coupled parallel computers

In this paper a parallel direct Schur–Fourier decomposition (DSFD) algorithm for the direct solution of arbitrary order discrete Poisson equations on parallel computers is proposed. It is based on a combination of a Direct Schur method and a Fourier decomposition and allows to solve each Poisson equation almost to machine accuracy using only one communication episode. Thus, it is well suited for loosely coupled parallel computers, that have a high network latency compared with the CPU performance. Several three‐dimensional direct numerical simulations (DNS) of wall‐bounded turbulent incompressible flows have been carried out using the DSFD algorithm. Numerical examples illustrating the robustness and scalability of the method on a PC cluster with a conventional 100 Mbits/s network are also presented. Copyright © 2005 John Wiley & Sons, Ltd.     

A cutting plane algorithm for the General Routing Problem

The General Routing Problem (GRP) is the problem of finding a minimum cost route for a single vehicle, subject to the condition that the vehicle visits certain vertices and edges of a network. It contains the Rural Postman Problem, Chinese Postman Problem and Graphical Travelling Salesman Problem as special cases. We describe a cutting plane algorithm for the GRP based on facet-inducing inequalities and show that it is capable of providing very strong lower bounds and, in most cases, optimal solutions. Received: November 1998 / Accepted: September 2000?Published online March 22, 2001     

Greedy primal-dual algorithm for dynamic resource allocation in complex networks

In Stolyar (Queueing Systems 50 (2005) 401–457) a dynamic control strategy, called greedy primal-dual (GPD) algorithm, was introduced for the problem of maximizing queueing network utility subject to stability of the queues, and was proved to be (asymptotically) optimal. (The network utility is a concave function of the average rates at which the network generates several “commodities.”) Underlying the control problem of Stolyar (Queueing Systems 50 (2005) 401–457) is a convex optimization problem subject to a set of linear constraints. In this paper we introduce a generalized GPD algorithm, which applies to the network control problem with additional convex (possibly non-linear) constraints on the average commodity rates. The underlying optimization problem in this case is a convex problem subject to convex constraints. We prove asymptotic optimality of the generalized GPD algorithm. We illustrate key features and applications of the algorithm on simple examples. AMS Subject Classifications: 90B15 · 90C25 · 60K25 · 68M12     

Admission control with advance reservations in simple networks

In the admission control problem we are given a network and a set of connection requests, each of which is associated with a path, a time interval, a bandwidth requirement, and a weight. A feasible schedule is a set of connection requests such that at any given time, the total bandwidth requirement on every link in the network is at most 1. Our goal is to find a feasible schedule with maximum total weight.We consider the admission control problem in two simple topologies: the line and the tree. We present a 12c-approximation algorithm for the line topology, where c is the maximum number of requests on a link at some time instance. This result implies a 12c-approximation algorithm for the rectangle packing problem, where c is the maximum number of rectangles that cover simultaneously a point in the plane. We also present an O(logt)-approximation algorithm for the tree topology, where t is the size of the tree. We consider the loss minimization version of the admission control problem in which the goal is to minimize the weight of unscheduled requests. We present a c-approximation algorithm for loss minimization problem in the tree topology. This result is based on an approximation algorithm for a generalization of set cover, in which each element has a covering requirement, and each set has a covering potential. The approximation ratio of this algorithm is Δ, where Δ is the maximum number of sets that contain the same element.     

Distributed subgradient method for multi‐agent optimization with quantized communication

This paper focuses on a distributed optimization problem associated with a time‐varying multi‐agent network with quantized communication, where each agent has local access to its convex objective function, and cooperatively minimizes a sum of convex objective functions of the agents over the network. Based on subgradient methods, we propose a distributed algorithm to solve this problem under the additional constraint that agents can only communicate quantized information through the network. We consider two kinds of quantizers and analyze the quantization effects on the convergence of the algorithm. Furthermore, we provide explicit error bounds on the convergence rates that highlight the dependence on the quantization levels. Finally, some simulation results on a l₁‐regression problem are presented to demonstrate the performance of the algorithm. Copyright © 2016 John Wiley & Sons, Ltd.     

Off-line and on-line guaranteed start-up delay for Media-on-Demand with stream merging

We address the problem of designing efficient algorithms for media-on-demand in systems that use stream merging. In the stream merging model the receiving bandwidth of clients is larger than the playback bandwidth, and clients can buffer parts of the transmission to be played back later. Our goal is to minimize the required server bandwidth for a given guaranteed start-up delay and uninterrupted playback. We construct an efficient O(n) optimal off-line algorithm for a time horizon that is composed of n time slots, where the length of one slot is the guaranteed start-up delay. Our algorithm works for either clients with receiving bandwidth twice as much as the playback bandwidth or for receiving bandwidth equal to the server bandwidth, independent of the clients buffer size. We describe an on-line delay guaranteed algorithm that operates without knowledge of the time horizon size, and show that it performs asymptotically close to the optimal off-line algorithm. The on-line algorithm is simpler to implement than previously proposed on-line stream merging algorithms, and empirically performs well when the intensity of client arrivals is high.     

The retraction algorithm for factoring banded symmetric matrices

Let A be an n × n symmetric matrix of bandwidth 2m + 1. The matrix need not be positive definite. In this paper we will present an algorithm for factoring A which preserves symmetry and the band structure and limits element growth in the factorization. With this factorization one may solve a linear system with A as the coefficient matrix and determine the inertia of A, the number of positive, negative, and zero eigenvalues of A. The algorithm requires between 1/2nm² and 5/4nm² multiplications and at most (2m + 1)n locations compared to non‐symmetric Gaussian elimination which requires between nm² and 2nm² multiplications and at most (3m + 1)n locations. Our algorithm reduces A to block diagonal form with 1 × 1 and 2 × 2 blocks on the diagonal. When pivoting for stability and subsequent transformations produce non‐zero elements outside the original band, column/row transformations are used to retract the bandwidth. To decrease the operation count and the necessary storage, we use the fact that the correction outside the band is rank‐1 and invert the process, applying the transformations that would restore the bandwidth first, followed by a modified correction. This paper contains an element growth analysis and a computational comparison with LAPACKs non‐symmetric band routines and the Snap‐back code of Irony and Toledo. Copyright © 2007 John Wiley & Sons, Ltd.     

Generalized loop‐erased random walks and approximate reachability

In this paper we extend the loop‐erased random walk (LERW) to the directed hypergraph setting. We then generalize Wilson's algorithm for uniform sampling of spanning trees to directed hypergraphs. In several special cases, this algorithm perfectly samples spanning hypertrees in expected polynomial time. Our main application is to the reachability problem, also known as the directed all‐terminal network reliability problem. This classical problem is known to be # P‐complete, hence is most likely intractable (Ball and Provan, SIAM J Comput 12 (1983) 777–788). We show that in the case of bi‐directed graphs, a conjectured polynomial bound for the expected running time of the generalized Wilson algorithm implies a FPRAS for approximating reachability. Copyright © 2012 Wiley Periodicals, Inc. Random Struct. Alg., 44, 201‐223, 2014     

Lower bounds for adaptive locally decodable codes

An error‐correcting code is said to be locally decodable if a randomized algorithm can recover any single bit of a message by reading only a small number of symbols of a possibly corrupted encoding of the message. Katz and Trevisan 12 showed that any such code C : {0, 1}ⁿ → Σ^m with a decoding algorithm that makes at most q probes must satisfy m = Ω((n/log |Σ|)^q/(q?1)). They assumed that the decoding algorithm is non‐adaptive, and left open the question of proving similar bounds for adaptive decoders. We show m = Ω((n/log |Σ|)^q/(q?1)) without assuming that the decoder is nonadaptive. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2005     

Symmetries that latin squares inherit from 1‐factorizations

A 1‐factorization of a graph is a decomposition of the graph into edge disjoint perfect matchings. There is a well‐known method, which we call the ??‐construction, for building a 1‐factorization of K_n,n from a 1‐factorization of K_{n + 1}. The 1‐factorization of K_n,n can be written as a latin square of order n. The ??‐construction has been used, among other things, to make perfect 1‐factorizations, subsquare‐free latin squares, and atomic latin squares. This paper studies the relationship between the factorizations involved in the ??‐construction. In particular, we show how symmetries (automorphisms) of the starting factorization are inherited as symmetries by the end product, either as automorphisms of the factorization or as autotopies of the latin square. Suppose that the ??‐construction produces a latin square L from a 1‐factorization F of K_{n + 1}. We show that the main class of L determines the isomorphism class of F, although the converse is false. We also prove a number of restrictions on the symmetries (autotopies and paratopies) which L may possess, many of which are simple consequences of the fact that L must be symmetric (in the usual matrix sense) and idempotent. In some circumstances, these restrictions are tight enough to ensure that L has trivial autotopy group. Finally, we give a cubic time algorithm for deciding whether a main class of latin squares contains any square derived from the ??‐construction. The algorithm also detects symmetric squares and totally symmetric squares (latin squares that equal their six conjugates). © 2005 Wiley Periodicals, Inc. J Combin Designs 13: 157–172, 2005.     

Second‐order performance analysis of discrete‐time queues fed by DAR(2) sources with a focus on the marginal effect of the additional traffic parameter

Discrete autoregressive process of order 1 (DAR(1)) has been used as a popular stochastic model for correlated traffic sources because it parsimoniously uses a single parameter to capture the desirable correlation structure. In contrast with DAR(1), discrete autoregressive process of order 2 (DAR(2)) uses one more parameter to provide a much richer pattern in the autocorrelation function and is able to capture slower decay rate and longer memory. To investigate how the additional traffic parameter in DAR(2) influences the queueing performance, this paper provides an analysis of the discrete‐time DAR(2)/D/1 queue. The performance measures concerned are the mean and second‐order statistics of queue size, which are both important in the queueing systems seen in telecommunication networks. Under a mild condition, these performance indices are derived in closed form that allows for efficient computing. An approximate version of these results is also developed to relax the condition and cover more general sources, and both versions serve as a simple tool set for performance evaluation. The numerical examples use this tool to demonstrate that the DAR(2) source may cause up to 30% poorer performance than DAR(1) when the traffic is heavy, bursty, and highly correlated. This indicates that the effect from slower decay rate in autocorrelation is not negligible and using the extra parameter is necessary particularly when the queue is heavily loaded with correlated traffic. Copyright © 2012 John Wiley & Sons, Ltd.     

Network delay inference from additive metrics

We use computational phylogenetic techniques to solve a central problem in inferential network monitoring. More precisely, we design a novel algorithm for multicast‐based delay inference, that is, the problem of reconstructing delay characteristics of a network from end‐to‐end delay measurements on network paths. Our inference algorithm is based on additive metric techniques used in phylogenetics. It runs in polynomial time and requires a sample of size only poly(log n). We also show how to recover the topology of the routing tree. © 2010 Wiley Periodicals, Inc. Random Struct. Alg., 2010     

Asymmetric graph coloring games

We introduce the (a,b)‐coloring game, an asymmetric version of the coloring game played by two players Alice and Bob on a finite graph, which differs from the standard version in that, in each turn, Alice colors a vertices and Bob colors b vertices. We also introduce a related game, the (a,b)‐marking game. We analyze these games and determine the (a,b)‐chromatic numbers and (a,b)‐coloring numbers for the class of forests and all values of a and b. © 2005 Wiley Periodicals, Inc. J Graph Theory 48: 169–185, 2005     

Survey propagation: An algorithm for satisfiability

We study the satisfiability of randomly generated formulas formed by M clauses of exactly K literals over N Boolean variables. For a given value of N the problem is known to be most difficult when α = M/N is close to the experimental threshold α_c separating the region where almost all formulas are SAT from the region where all formulas are UNSAT. Recent results from a statistical physics analysis suggest that the difficulty is related to the existence of a clustering phenomenon of the solutions when α is close to (but smaller than) α_c. We introduce a new type of message passing algorithm which allows to find efficiently a satisfying assignment of the variables in this difficult region. This algorithm is iterative and composed of two main parts. The first is a message‐passing procedure which generalizes the usual methods like Sum‐Product or Belief Propagation: It passes messages that may be thought of as surveys over clusters of the ordinary messages. The second part uses the detailed probabilistic information obtained from the surveys in order to fix variables and simplify the problem. Eventually, the simplified problem that remains is solved by a conventional heuristic. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2005     

Lanczos and the Riemannian SVD in information retrieval applications

Variations of the latent semantic indexing (LSI) method in information retrieval (IR) require the computation of singular subspaces associated with the k dominant singular values of a large m × n sparse matrix A, where k?min(m,n). The Riemannian SVD was recently generalized to low‐rank matrices arising in IR and shown to be an effective approach for formulating an enhanced semantic model that captures the latent term‐document structure of the data. However, in terms of storage and computation requirements, its implementation can be much improved for large‐scale applications. We discuss an efficient and reliable algorithm, called SPK‐RSVD‐LSI, as an alternative approach for deriving the enhanced semantic model. The algorithm combines the generalized Riemannian SVD and the Lanczos method with full reorthogonalization and explicit restart strategies. We demonstrate that our approach performs as well as the original low‐rank Riemannian SVD method by comparing their retrieval performance on a well‐known benchmark document collection. Copyright 2004 John Wiley & Sons, Ltd.     

Regularly varying tails in a queue with discrete autoregressive arrivals of order p

We consider a discrete time single server queueing system where the service time of a customer is one slot, and the arrival process is governed by a discrete autoregressive process of order p (DAR(p)). For this queueing system, we investigate the tail behavior of the queue size and the waiting time distributions. Specifically, we show that if the stationary distribution of DAR(p) input has a tail of regular variation with index −β−1, then the stationary distributions of the queue size and the waiting time have tails of regular variation with index −β. This research was supported by the MIC (Ministry of Information and Communication), Korea, under the ITRC (Information Technology Research Center) support program supervised by the IITA (Institute of Information Technology Assessment).     

Cores in random hypergraphs and Boolean formulas

We describe a technique for determining the thresholds for the appearance of cores in random structures. We use it to determine (i) the threshold for the appearance of a k‐core in a random r‐uniform hypergraph for all r, k ≥ 2, r + k > 4, and (ii) the threshold for the pure literal rule to find a satisfying assignment for a random instance of r‐SAT, r ≥ 3. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2005     

On the number of maximal bipartite subgraphs of a graph

We show new lower and upper bounds on the maximum number of maximal induced bipartite subgraphs of graphs with n vertices. We present an infinite family of graphs having 105^n/10 ≈ 1.5926ⁿ; such subgraphs show an upper bound of O(12^n/4) = O(1.8613ⁿ) and give an algorithm that finds all maximal induced bipartite subgraphs in time within a polynomial factor of this bound. This algorithm is used in the construction of algorithms for checking k‐colorability of a graph. © 2004 Wiley Periodicals, Inc. J Graph Theory 48: 127–132, 2005     

Ordered interval routing schemes

An Interval Routing Scheme (IRS) represents the routing tables in a network in a space-efficient way by labeling each vertex with an unique integer address, and the outgoing edges at each vertex with disjoint subintervals of these addresses. An IRS that has at most k intervals per edge label is called a k-IRS. In this paper, we propose a new type of interval routing scheme, called an Ordered Interval Routing Scheme (OIRS), that uses an ordering of the outgoing edges at each vertex and allows non-disjoint intervals in the labels of those edges. We show for a number of graph classes that using an OIRS instead of an IRS reduces the size of the routing tables in the case of optimal routing, i.e., routing along shortest paths. We show that optimal routing in any k-tree is possible using an OIRS with at most 2^k−1 intervals per edge label, although the best known result for an IRS is 2^k+1 intervals per edge label. Any torus has an optimal 1-OIRS, although it may not have an optimal 1-IRS. We present similar results for the Petersen graph, k-garland graphs and a few other graphs.     

Four correlates of complex behavioral networks: Differentiation,behavior, connectivity,and compartmentalization: Carving networks at their joints

Some of the most complex networks are those that (i) have been engineered under selective pressure (either economic or evolutionary), and (ii) are capable of eliciting network‐level behaviors. Some examples are nervous systems, ant colonies, electronic circuits and computer software. Here we provide evidence that many such selected, behavioral networks are similar in at least four respects. (1) Differentiation: Nodes of different types are used in a combinatorial fashion to build network structures through local connections, and networks accommodate more structure types via increasing the number of node types in the network (i.e., increasing differentiation), not via increasing the length of structures. (2) Behavior: Structures are themselves combined globally to implement behaviors, and networks accommodate a greater behavioral repertoire via increasing the number of lower‐level behavior types (including structures), not via increasing the length of behaviors. (3) Connectivity: In order for structures in behavioral networks to combine with other structures within a fixed behavior length, the network must maintain an invariant network diameter, and this is accomplished via increasing network connectivity in larger networks. (4) Compartmentalization: Finally, for reasons of economical wiring, behavioral networks become increasingly parcellated. Special attention is given to nervous systems and computer software, but data from a variety of other behavioral selected networks are also provided, including ant colonies, electronic circuits, web sites and businesses. A general framework is introduced illuminating why behavioral selected networks share these four correlates. Because the four above features appear to apply to computer software as well as to biological networks, computer software provides a useful framework for comprehending the large‐scale function and organization of biological networks. © 2005 Wiley Periodicals, Inc. Complexity 10: 13–40, 2005