Nearly-Linear Time Algorithms for Preconditioning and Solving Symmetric, Diagonally Dominant Linear Systems

SIAM Journal on Matrix Analysis and Applications (Impact Factor: 1.81). 07/2006; 35(3). DOI: 10.1137/090771430
Source: arXiv

ABSTRACT We present a randomized algorithm that, on input a symmetric, weakly
diagonally dominant n-by-n matrix A with m nonzero entries and an n-vector b,
produces a y such that $\norm{y - \pinv{A} b}_{A} \leq \epsilon \norm{\pinv{A}
b}_{A}$ in expected time $O (m \log^{c}n \log (1/\epsilon)),$ for some constant
c. By applying this algorithm inside the inverse power method, we compute
approximate Fiedler vectors in a similar amount of time. The algorithm applies
subgraph preconditioners in a recursive fashion. These preconditioners improve
upon the subgraph preconditioners first introduced by Vaidya (1990).
For any symmetric, weakly diagonally-dominant matrix A with non-positive
off-diagonal entries and $k \geq 1$, we construct in time $O (m \log^{c} n)$ a
preconditioner B of A with at most $2 (n - 1) + O ((m/k) \log^{39} n)$ nonzero
off-diagonal entries such that the finite generalized condition number
$\kappa_{f} (A,B)$ is at most k, for some other constant c.
In the special case when the nonzero structure of the matrix is planar the
corresponding linear system solver runs in expected time $ O (n \log^{2} n + n
\log n \ \log \log n \ \log (1/\epsilon))$.
We hope that our introduction of algorithms of low asymptotic complexity will
lead to the development of algorithms that are also fast in practice.

1 Bookmark
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, we propose a fast distributed solver for linear equations given by symmetric diagonally dominant M-Matrices. Our approach is based on a distributed implementation of the parallel solver of Spielman and Peng by considering a specific approximated inverse chain which can be computed efficiently in a distributed fashion. Representing the system of equations by a graph $\mathbb{G}$, the proposed distributed algorithm is capable of attaining $\epsilon$-close solutions (for arbitrary $\epsilon$) in time proportional to $n^{3}$ (number of nodes in $\mathbb{G}$), ${\alpha}$ (upper bound on the size of the R-Hop neighborhood), and $\frac{{W}_{max}}{{W}_{min}}$ (maximum and minimum weight of edges in $\mathbb{G}$).
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Complex networks can be typically broken down into groups or modules. Discovering this "community structure" is an important step in studying the large-scale structure of networks. Many algorithms have been proposed for community detection and benchmarks have been created to evaluate their performance. Typically algorithms for community detection either partition the graph (non-overlapping communities) or find node covers (overlapping communities). In this paper, we propose a particularly simple semi-supervised learning algorithm for finding out communities. In essence, given the community information of a small number of "seed nodes", the method uses random walks from the seed nodes to uncover the community information of the whole network. The algorithm runs in time $O(k \cdot m \cdot \log n)$, where $m$ is the number of edges; $n$ the number of links; and $k$ the number of communities in the network. In sparse networks with $m = O(n)$ and a constant number of communities, this running time is almost linear in the size of the network. Another important feature of our algorithm is that it can be used for either non-overlapping or overlapping communities. We test our algorithm using the LFR benchmark created by Lancichinetti, Fortunato, and Radicchi specifically for the purpose of evaluating such algorithms. Our algorithm can compete with the best of algorithms for both non-overlapping and overlapping communities as found in the comprehensive study of Lancichinetti and Fortunato.
  • [Show abstract] [Hide abstract]
    ABSTRACT: Analysis of VLSI designs and circuits often requires the construction of a small-signal equivalent impedance representation between prescribed node pairs. Examples include IR-drop calculation, electrical overstress verification, ESD protection, di/dt current rush analysis, electro-migration checks, thermal analysis and model order reduction. VLSI designs consisting of hundreds of millions (10e11) of linear circuit elements are now commonplace and thus any method which requires compute intensive calculations is difficult to apply for all the elements of the circuit. By definition, given a circuit with n elements, conventional analysis techniques have a Ω(η2) lower-bound to exhaustively enumerate all node-pairs which meet a constraint criteria. We present the first 0(nlgn + k) algorithm for answering queries of the form: ∃(x, y):((x, y) ϵ × n) ∧(Zeff (x, y) <; Z) where Zeff(x, y) is the equivalent impedance between nodes x and y, and k node-pairs meet the constraint. Calculating all node pairs for which these constraints are met is compute intensive using existing techniques, even using heuristic methods. In this paper a new technique based on method of projection using ellipsoidal norms is presented. Our proposed method employs recently discovered techniques from theoretical computer science to compute an e-approximate embedding matrix from which effective impedance of all node pairs can be estimated as easily as taking the Euclidean norm of column differences. Using computational geometry methods, existence queries can then be answered in logarithmic time. The method works for general circuits containing resistances, capacitances, and inductances.
    2014 15th International Symposium on Quality Electronic Design (ISQED); 03/2014


Available from