Anisotropic <Emphasis Type="Italic">ε</Emphasis>-optimal model reduction for linear discrete time-invariant system

We consider an ε-optimal model reduction problem for a linear discrete time-invariant system, where the anisotropic norm of reduction error transfer function is used as a performance criterion. For solving the main problem, we state and solve an auxiliary problem of H ₂ ε-optimal reduction of a weighted linear discrete time system. A sufficient optimality condition defining a solution to the anisotropic ε-optimal model reduction problem has the form of a system of cross-coupled nonlinear matrix algebraic equations including a Riccati equation, four Lyapunov equations, and five special-type nonlinear equations. The proposed approach to solving the problem ensures stability of the reduced model without any additional technical assumptions. The reduced-order model approximates the steady-state behavior of the full-order system.     

Approximate Range Searching in External Memory

In this paper, we present two linear-size external memory data structures for approximate range searching. Our first structure, the BAR-B-tree, stores a set of N points in ℝ ^d and can report all points inside a query range Q by accessing O(log  _B N+ε ^γ +k _ε /B) disk blocks, where B is the disk block size, γ=1−d for convex queries and γ=−d otherwise, and k _ε is the number of points lying within a distance of ε⋅diam (Q) to the query range Q. Our second structure, the object-BAR-B-tree, is able to store objects of arbitrary shapes of constant complexity and provides similar query guarantees. In addition, both structures can support other types of range searching queries such as range aggregation and nearest-neighbor. Finally, we present I/O-efficient algorithms to build these structures.     

An Almost Space-Optimal Streaming Algorithm for Coresets in Fixed Dimensions

We present a new streaming algorithm for maintaining an ε-kernel of a point set in ℝ ^d using O((1/ε ^(d−1)/2)log (1/ε)) space. The space used by our algorithm is optimal up to a small logarithmic factor. This significantly improves (for any fixed dimension d ≥3) the best previous algorithm for this problem that uses O(1/ε ^d−(3/2)) space, presented by Agarwal and Yu. Our algorithm immediately improves the space complexity of the previous streaming algorithms for a number of fundamental geometric optimization problems in fixed dimensions, including width, minimum-volume bounding box, minimum-radius enclosing cylinder, minimum-width enclosing annulus, etc.     

Fast Dynamic Transitive Closure with Lookahead

In this paper we consider the problem of dynamic transitive closure with lookahead. We present a randomized one-sided error algorithm with updates and queries in O(n ^{ω(1,1,ε)−ε} ) time given a lookahead of n ^ε operations, where ω(1,1,ε) is the exponent of multiplication of n×n matrix by n×n ^ε matrix. For ε≤0.294 we obtain an algorithm with queries and updates in O(n ^2−ε ) time, whereas for ε=1 the time is O(n ^ω−1). This is essentially optimal as it implies an O(n ^ω ) algorithm for boolean matrix multiplication. We also consider the offline transitive closure in planar graphs. For this problem, we show an algorithm that requires O(n^\fracw2)O(n^{\frac{\omega}{2}}) time to process n^\frac12n^{\frac{1}{2}} operations. We also show a modification of these algorithms that gives faster amortized queries. Finally, we give faster algorithms for restricted type of updates, so called element updates. All of the presented algorithms are randomized with one-sided error. All our algorithms are based on dynamic algorithms with lookahead for matrix inverse, which are of independent interest.     

Large Deformation Diffeomorphic Metric Curve Mapping

We present a matching criterion for curves and integrate it into the large deformation diffeomorphic metric mapping (LDDMM) scheme for computing an optimal transformation between two curves embedded in Euclidean space ℝ ^d . Curves are first represented as vector-valued measures, which incorporate both location and the first order geometric structure of the curves. Then, a Hilbert space structure is imposed on the measures to build the norm for quantifying the closeness between two curves. We describe a discretized version of this, in which discrete sequences of points along the curve are represented by vector-valued functionals. This gives a convenient and practical way to define a matching functional for curves. We derive and implement the curve matching in the large deformation framework and demonstrate mapping results of curves in ℝ² and ℝ³. Behaviors of the curve mapping are discussed using 2D curves. The applications to shape classification is shown and experiments with 3D curves extracted from brain cortical surfaces are presented. J. Glaunès and A. Qiu contributed equally to this work.     

Approximations for discrete-time adaptive control: Construction of ε-optimal controls

We study a discrete-time, finite-horizon adaptive stochastic control problem. The unknown parameter enters the state equation linearly, but otherwise the problem is nonlinear. We construct a sequence of approximating problems with the following properties: (i) an optimal solution for each approximating problem can be obtained via dynamic programming; (ii) given ε>0, there exists an approximating problem whose optimal solution is ε-optimal for the original problem.     

Nodal Superconvergence of SDFEM for Singularly Perturbed Problems

In this paper, we analyze the streamline diffusion finite element method for one dimensional singularly perturbed convection-diffusion-reaction problems. Local error estimates on a subdomain where the solution is smooth are established. We prove that for a special group of exact solutions, the nodal error converges at a superconvergence rate of order (ln ε ⁻¹/N)^2k (or (ln N/N)^2k ) on a Shishkin mesh. Here ε is the singular perturbation parameter and 2N denotes the number of mesh elements. Numerical results illustrating the sharpness of our theoretical findings are displayed.     

A Tight Lower Bound for Computing the Diameter of a 3D Convex Polytope

The diameter of a set P of n points in ℝ ^d is the maximum Euclidean distance between any two points in P. If P is the vertex set of a 3-dimensional convex polytope, and if the combinatorial structure of this polytope is given, we prove that, in the worst case, deciding whether the diameter of P is smaller than 1 requires Ω(nlog n) time in the algebraic computation tree model. It shows that the O(nlog n) time algorithm of Ramos for computing the diameter of a point set in ℝ³ is optimal for computing the diameter of a 3-polytope. We also give a linear time reduction from Hopcroft’s problem of finding an incidence between points and lines in ℝ² to the diameter problem for a point set in ℝ⁷.     

Two-Grid Discontinuous Galerkin Method for Quasi-Linear Elliptic Problems

In this paper, we consider the symmetric interior penalty discontinuous Galerkin (SIPG) method with piecewise polynomials of degree r≥1 for a class of quasi-linear elliptic problems in Ω⊂ℝ². We propose a two-grid approximation for the SIPG method which can be thought of as a type of linearization of the nonlinear system using a solution from a coarse finite element space. With this technique, solving a quasi-linear elliptic problem on the fine finite element space is reduced into solving a linear problem on the fine finite element space and solving the quasi-linear elliptic problem on a coarse space. Convergence estimates in a broken H ¹-norm are derived to justify the efficiency of the proposed two-grid algorithm. Numerical experiments are provided to confirm our theoretical findings. As a byproduct of the technique used in the analysis, we derive the optimal pointwise error estimates of the SIPG method for the quasi-linear elliptic problems in ℝ ^d ,d=2,3 and use it to establish the convergence of the two-grid method for problems in Ω⊂ℝ³.     

Optimal External Memory Planar Point Enclosure

In this paper we study the external memory planar point enclosure problem: Given N axis-parallel rectangles in the plane, construct a data structure on disk (an index) such that all K rectangles containing a query point can be reported I/O-efficiently. This problem has important applications in e.g. spatial and temporal databases, and is dual to the important and well-studied orthogonal range searching problem. Surprisingly, despite the fact that the problem can be solved optimally in internal memory with linear space and O(log N+K) query time, we show that one cannot construct a linear sized external memory point enclosure data structure that can be used to answer a query in O(log  _B N+K/B) I/Os, where B is the disk block size. To obtain this bound, Ω(N/B ^1−ε ) disk blocks are needed for some constant ε>0. With linear space, the best obtainable query bound is O(log ₂ N+K/B) if a linear output term O(K/B) is desired. To show this we prove a general lower bound on the tradeoff between the size of the data structure and its query cost. We also develop a family of structures with matching space and query bounds. An extended abstract of this paper appeared in Proceedings of the 12th European Symposium on Algorithms (ESA’04), Bergen, Norway, September 2004, pp. 40–52. L. Arge’s research was supported in part by the National Science Foundation through RI grant EIA–9972879, CAREER grant CCR–9984099, ITR grant EIA–0112849, and U.S.-Germany Cooperative Research Program grant INT–0129182, as well as by the US Army Research Office through grant W911NF-04-01-0278, by an Ole Roemer Scholarship from the Danish National Science Research Council, a NABIIT grant from the Danish Strategic Research Council and by the Danish National Research Foundation. V. Samoladas’ research was supported in part by a grant co-funded by the European Social Fund and National Resources-EPEAEK II-PYTHAGORAS. K. Yi’s research was supported in part by the National Science Foundation through ITR grant EIA–0112849, U.S.-Germany Cooperative Research Program grant INT–0129182, and Hong Kong Direct Allocation Grant (DAG07/08).     

Analysis of Galerkin Methods for the Fully Nonlinear Monge-Ampère Equation

This paper develops and analyzes finite element Galerkin and spectral Galerkin methods for approximating viscosity solutions of the fully nonlinear Monge-Ampère equation det (D ² u ⁰)=f (>0) based on the vanishing moment method which was developed by the authors in Feng and Neilan (J. Sci. Comput. 38:74–98, 2009) and Feng (Convergence of the vanishing moment method for the Monge-Ampère equation, submitted). In this approach, the Monge-Ampère equation is approximated by the fourth order quasilinear equation −εΔ² u ^ε +det D ² u ^ε =f accompanied by appropriate boundary conditions. This new approach enables us to construct convergent Galerkin numerical methods for the fully nonlinear Monge-Ampère equation (and other fully nonlinear second order partial differential equations), a task which has been impracticable before. In this paper, we first develop some finite element and spectral Galerkin methods for approximating the solution u ^ε of the regularized problem. We then derive optimal order error estimates for the proposed numerical methods. In particular, we track explicitly the dependence of the error bounds on the parameter ε, for the error u^e-u^e_hu^{\varepsilon}-u^{\varepsilon}_{h}. Due to the strong nonlinearity of the underlying equation, the standard error estimate technique, which has been widely used for error analysis of finite element approximations of nonlinear problems, does not work here. To overcome the difficulty, we employ a fixed point technique which strongly makes use of the stability property of the linearized problem and its finite element approximations. Finally, using the Argyris finite element method as an example, we present a detailed numerical study of the rates of convergence in terms of powers of ε for the error u⁰-u_h^eu^{0}-u_{h}^{\varepsilon}, and numerically examine what is the “best” mesh size h in relation to ε in order to achieve these rates.     

On the validity of using small positive lower bounds on design variables in discrete topology optimization

It is proved that an optimal {ε, 1} ⁿ solution to a “ε-perturbed” discrete minimum weight problem with constraints on compliance, von Mises stresses and strain energy densities, is optimal, after rounding to {0, 1} ⁿ , to the corresponding “unperturbed” discrete problem, provided that the constraints in the perturbed problem are carefully defined and ε > 0 is sufficiently small.     

On signal reconstruction in white noise using dictionaries

Assume that we observe a Gaussian vector Y = Xβ + σζ, where X is a known p × n matrix with p ≥ n, β ∈ ℝ ⁿ is an unknown vector, and ζ ∈ ℝ ⁿ is a standard Gaussian white noise. The problem is to reconstruct Xβ from observations Y, provided that β is a sparse vector.     

Exact Algorithms for the Bottleneck Steiner Tree Problem

Given n points, called terminals, in the plane ℝ² and a positive integer k, the bottleneck Steiner tree problem is to find k Steiner points from ℝ² and a spanning tree on the n+k points that minimizes its longest edge length. Edge length is measured by an underlying distance function on ℝ², usually, the Euclidean or the L ₁ metric. This problem is known to be NP-hard. In this paper, we study this problem in the L _p metric for any 1≤p≤∞, and aim to find an exact algorithm which is efficient for small fixed k. We present the first fixed-parameter tractable algorithm running in f(k)⋅nlog ² n time for the L ₁ and the L _∞ metrics, and the first exact algorithm for the L _p metric for any fixed rational p with 1<p<∞ whose time complexity is f(k)⋅(n ^k +nlog n), where f(k) is a function dependent only on k. Note that prior to this paper there was no known exact algorithm even for the L ₂ metric.     

Practical Methods for Shape Fitting and Kinetic Data Structures using Coresets

The notion of ε-kernel was introduced by Agarwal et al. (J. ACM 51:606–635, 2004) to set up a unified framework for computing various extent measures of a point set P approximately. Roughly speaking, a subset Q⊆P is an ε-kernel of P if for every slab W containing Q, the expanded slab (1+ε)W contains P. They illustrated the significance of ε-kernel by showing that it yields approximation algorithms for a wide range of geometric optimization problems. We present a simpler and more practical algorithm for computing the ε-kernel of a set P of points in ℝ ^d . We demonstrate the practicality of our algorithm by showing its empirical performance on various inputs. We then describe an incremental algorithm for fitting various shapes and use the ideas of our algorithm for computing ε-kernels to analyze the performance of this algorithm. We illustrate the versatility and practicality of this technique by implementing approximation algorithms for minimum enclosing cylinder, minimum-volume bounding box, and minimum-width annulus. Finally, we show that ε-kernels can be effectively used to expedite the algorithms for maintaining extents of moving points. A preliminary version of the paper appeared in Proceedings of the 20th Annual ACM Symposium on Computational Geometry, 2004, pp. 263–272. Research by the first two authors is supported by NSF under grants CCR-00-86013, EIA-98-70724, EIA-01-31905, and CCR-02-04118, and by a grant from the US–Israel Binational Science Foundation. Research by the fourth author is supported by NSF CAREER award CCR-0237431.     

Efficiently Testing Sparse <Emphasis Type="Italic">GF</Emphasis>(2) Polynomials

We give the first algorithm that is both query-efficient and time-efficient for testing whether an unknown function f:{0,1} ⁿ →{−1,1} is an s-sparse GF(2) polynomial versus ε-far from every such polynomial. Our algorithm makes poly(s,1/ε) black-box queries to f and runs in time n⋅poly(s,1/ε). The only previous algorithm for this testing problem (Diakonikolas et al. in Proceedings of the 48th Annual Symposium on Foundations of Computer Science, FOCS, pp. 549–558, 2007) used poly(s,1/ε) queries, but had running time exponential in s and super-polynomial in 1/ε.     

Richardson extrapolation technique for singularly perturbed parabolic convection–diffusion problems

This paper deals with the study of a post-processing technique for one-dimensional singularly perturbed parabolic convection–diffusion problems exhibiting a regular boundary layer. For discretizing the time derivative, we use the classical backward-Euler method and for the spatial discretization the simple upwind scheme is used on a piecewise-uniform Shishkin mesh. We show that the use of Richardson extrapolation technique improves the ε-uniform accuracy of simple upwinding in the discrete supremum norm from O (N ⁻¹ ln N + Δt) to O (N ⁻² ln² N + Δt ²), where N is the number of mesh-intervals in the spatial direction and Δt is the step size in the temporal direction. The theoretical result is also verified computationally by applying the proposed technique on two test examples.     

Lattice constants matricial equations for conversion matrices

In this paper we proof a theorem concerning lattice constants and hence three matricial equations for conversion matricesR: if H=Δ^RT we have: i)H ³ =I; ii) H^T Σ H= Σ; iii)(DH) ² =I; where Δ,D, and ε are known when we can enumerate all non-isomorphic graphs withn vertices, we know (for Δ and ε) their edge-number and (for ε) the order of their automorphism group.     

On approximating the longest path in a graph

We consider the problem of approximating the longest path in undirected graphs. In an attempt to pin down the best achievable performance ratio of an approximation algorithm for this problem, we present both positive and negative results. First, a simple greedy algorithm is shown to find long paths in dense graphs. We then consider the problem of finding paths in graphs that are guaranteed to have extremely long paths. We devise an algorithm that finds paths of a logarithmic length in Hamiltonian graphs. This algorithm works for a much larger class of graphs (weakly Hamiltonian), where the result is the best possible. Since the hard case appears to be that of sparse graphs, we also consider sparse random graphs. Here we show that a relatively long path can be obtained, thereby partially answering an open problem of Broderet al. To explain the difficulty of obtaining better approximations, we also prove hardness results. We show that, for any ε<1, the problem of finding a path of lengthn-n ^ε in ann-vertex Hamiltonian graph isNP-hard. We then show that no polynomial-time algorithm can find a constant factor approximation to the longest-path problem unlessP=NP. We conjecture that the result can be strengthened to say that, for some constant δ>0, finding an approximation of ration ^δ is alsoNP-hard. As evidence toward this conjecture, we show that if any polynomial-time algorithm can approximate the longest path to a ratio of , for any ε>0, thenNP has a quasi-polynomial deterministic time simulation. The hardness results apply even to the special case where the input consists of bounded degree graphs. D. Karger was supported by an NSF Graduate Fellowship, NSF Grant CCR-9010517, and grants from the Mitsubishi Corporation and OTL. R. Motwani was supported by an Alfred P. Sloan Research Fellowship, an IBM Faculty Development Award, grants from Mitsubishi and OTL, NSF Grant CCR-9010517, and NSF Young Investigator Award CCR-9357849, with matching funds from IBM, the Schlumberger Foundation, the Shell Foundation, and the Xerox Corporation, G. D. S. Ramkumar was supported by a grant from the Toshiba Corporation. Communicated by M. X. Goemans.     

Piercing Translates and Homothets of a Convex Body

According to a classical result of Grünbaum, the transversal number τ(ℱ) of any family ℱ of pairwise-intersecting translates or homothets of a convex body C in ℝ ^d is bounded by a function of d. Denote by α(C) (resp. β(C)) the supremum of the ratio of the transversal number τ(ℱ) to the packing number ν(ℱ) over all finite families ℱ of translates (resp. homothets) of a convex body C in ℝ ^d . Kim et al. recently showed that α(C) is bounded by a function of d for any convex body C in ℝ ^d , and gave the first bounds on α(C) for convex bodies C in ℝ ^d and on β(C) for convex bodies C in the plane.