DOCSLIB.ORG

On the Efficient Use of Givens Rotations in SVD-Based Subspace Tracking

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
On the Efficient Use of Givens Rotations in SVD-Based Subspace Tracking Signal Processing 74 (1999) 253—277 On the efficient use of Givens rotations in SVD-based subspace tracking algorithms Philippe Pango*, Benoıˆt Champagne INRS-Te& le& communications, Universite& du Que& bec, 16 place du Commerce, Verdun, Que&bec, Canada H3E 1H6 Received 28 July 1997; received in revised form 23 October 1998 Abstract In this paper, the issue of the efficient use of Givens rotations in SVD-based QR Jacobi-type subspace tracking algorithms is addressed. By relaxing the constraint of upper triangularity on the singular value matrix, we show how even fewer Givens rotations can achieve a better diagonalization and provide more accurate singular values. Then, we investigate the efficient use of Givens rotations as a vector rotation tool. The cancellation of cross-terms is presented as an efficient signal/noise separation technique which guarantees a better updating of the subspaces basis. Regarding the choice between inner and outer rotations, we properly use the permutation properties of Givens rotations to maintain the decreasing ordering of the singular values throughout the updating process and analyze the consequences on the tracking performance of QR Jacobi-type algorithms. Finally, based on the developed theory, we propose two efficient subspace tracking algorithms which outperform existing QR Jacobi-type algorithms. Comparative simulation experiments validate the concepts. ( 1999 Elsevier Science B.V. All rights reserved. Zusammenfassung In diesem Beitrag wird die effiziente Verwendung von Givens rotations in SVD-basierten subspace tracking Algorith- men vom QR Jacobi-Typ diskutiert. Wir zeigen, dass durch die Aufgabe der Einschra¨ nkung auf obere Dreiecksstruktur der Singula¨ rwertmatrix mit weniger Givens rotations eine bessere Diagonalisierung and genauere Singula¨ rwerte erzielt werden ko¨ nnen. Weiters wird die effiziente Verwendung von Givens rotations als ein vector rotation Werkzeug diskutiert. Die gegenseitige Aufhebung von Kreuztermen wird als effizientes Signal/Gerau¨ sch Trennungsverfahren pra¨ sentiert, welches ein besseres updating der Unterraumbasis garantiert. Wir demonstrieren den Einfluss der Rotations- art (innere oder au¨ ssere) auf die Reihenfolge der Singula¨ rwerte wa¨ hrend des gesamten Aktualisierungsprozesses. Weiters werden die Konsequenzen der Rotationsart auf die tracking performance diskutiert. Schliesslich werden zwei effiziente subspace tracking Algorithmen vorgeschlagen, die auf den gewonnenen Ergebnissen basieren und besser sind als existierende Algorithmen vom QR Jacobi-Typ. Vergleichende Simulationsergebnisse besta¨ tigen die vorgestellen Kon- zepte. ( 1999 Elsevier Science B.V. All rights reserved. Re´sume´ Dans cet article, nous abordons le proble`me de l’utilisation optimale des rotations de Givens dans les algorithmes de type QR Jacobi pour le suivi de sous-espaces. En enlevant la contrainte de triangularite´ supe´rieure, nous montrons * Corresponding author. Tel.: 514-761-8635; fax: 514-761-8501; e-mail: [email protected] 0165-1684/99/$ — see front matter ( 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 5 - 1 6 8 4 ( 9 8 ) 0 0 2 1 5 - 1 254 P. Pango, B. Champagne / Signal Processing 74 (1999) 253—277 comment un nombre re´duit de rotations de Givens rotations peuvent produire une meilleure diagonalisation de la matrice des valeurs singulie`res et donner des valeurs singulie`res plus pre´cises. Puis, nous analysons l’utilisation efficace des rotations de Givens en tant qu’outils de rotation de vecteurs. L’annulation des termes croise´s est pre´sente´e comme une technique efficace de se´paration des sous-espaces signal et bruit qui garanti une meilleure mise a` jour des bases des sous-espaces. A© propos du choix entre rotations internes et externes, nous exploitons ade´quatement les proprie´te´s permutatives de ces rotations pour maintenir l’ordre de´croissant des valeurs singulie`res lors de la mise a` jour, et analysons les conse´quences sur les performances de suivi des algorithmes de type QR Jacobi. Finalement, en se basant sur la the´orie de´veloppe´e, nous proposons deux algorithmes efficaces de suivi de sous-espaces dont les performances exce`dent celles des algorithmes de type QR Jacobi existant. Des simulation expe´rimentales valident les concepts pro- pose´s. ( 1999 Elsevier Science B.V. All rights reserved. Keywords: Subspace tracking; Invariant subspace updating; Singular-value decomposition; Givens rotation; Cross- terms; Direction of arrival Notation various decompositions, including the eigenvalue a angle of arrival decomposition (EVD) of the sample correlation A data matrix matrix, the QR factorization [2] or the singular- *» 1 time variation of signal subspace value decomposition (SVD) of the data matrix. In * J convergence step induced by the lth rota- the case of non-stationary signals, these decomposi- tion tions need to be updated each time new informa- c efficiency of Givens rotations (EGR) tion is provided by the most recent measurements. GH h GF Givens rotation in i—j plane with angle The quality of the updating process affects the k time index accuracy of the parameter estimates, and its low j forgetting factor complexity may allow a real-time implementation. M number of pairs of rotations at refinement These considerations have led to an intensive re- step search on the development of the so-called sub- N number of sensors space tracking algorithms. Beginning with Owsley off[ ] off-norm of its argument [18], many adaptive subspace tracking algorithms r number of sources have been developed and can be grouped in fami- p G ith singular value lies, depending on the specific technique they use: p , average noise singular value stochastic gradient [35], recursive least squares R singular value matrix [34], perturbation approach [3], to quote a few. º left singular vectors The introduction in [5] is a good source of refer- » right singular vectors ences concerning the development of subspace u electrical angle tracking algorithms. x measurement vector Generally, an SVD-based subspace tracking con- » y update vector, i.e. projection of x on sists in computing the SVD of a data matrix of yG update vector after the ith QR rotation growing dimension, defined recursively as (jA(k!1) A(k)" , (1) 1. Introduction C x&(k) D In the application of subspace methods to array where k is the discrete time index, 0(j(1 is the signal processing, projection techniques like forgetting factor, and x(k)3", is the measurement MUSIC [25], root-MUSIC [1], ESPRIT [24] or vector at time index k. The SVD of A(k) can be minimum-norm [15] are often used to estimate expressed as the signal parameters. These methods require a subspace information which can be provided by A(k)"º(k)R(k)»(k)&, (2) P. Pango, B. Champagne / Signal Processing 74 (1999) 253—277 255 where »(k)isanN;N unitary matrix, º(k)is many questions remain unanswered regarding a k;N matrix with orthonormal columns and R(k) these algorithms. Since they use exclusively Givens is an N;N diagonal matrix. Usually, the left singu- rotations throughout the updating process, one lar vectors º(k) need not be tracked to provide the must analyze the efficiency of these rotations on the subspace information. So, one is interested in track- singular values and singular vectors tracking per- +p *p *2*p , ing only the singular values , formance, and also on the accuracy of the signal along the diagonal of R(k) and the right singular parameter estimates they provide. vectors, i.e. the columns of »(k). Computing this SVD Regarding the singular values, QR Jacobi-type from scratch at each iteration is computationally algorithms usually track an approximate SVD de- expensive and time-consuming. The main issue composition (e.g. URV) in which the singular value consists then in using the information contained in matrix R(k) is given a specific structure; then, Givens the measurement x(k) to update the previous de- rotations are applied throughout the updating algo- composition obtained at time k!1. rithm in order not to destroy that specific structure. In this work, we focus our attention on a specific It might be argued that such a structural constraint family of SVD-based subspace tracking algorithms introduces limitations in the way Givens rotations which we define as QR Jacobi-type algorithms couldbeusedtoachieveabetterupdatingofthe [12,17,23]. QR Jacobi-type algorithms intensively singular values. For example, the above QR Jacobi- make use of Givens rotations in the subspace updat- type algorithms impose R(k)inEq.(2)tobeupper- ing process, the main matrix operations consisting in triangular. The efficiency of Givens rotations assum- products with Givens rotations; thus, they track the ing this specific structure (and eventually others) has subspaces with a low complexity and a rather simple not been investigated yet. Therefore, one notes that formulation. Also, Givens rotations have the ad- the above algorithms use Givens rotations in a way vantage to maintain the orthonormality of matrices, which is not proved to be optimal. assuming an infinite precision. The direct conse- Regarding the singular vectors, QR Jacobi-type quence is that QR Jacobi-type algorithms are good algorithms use Givens rotations to rotate the ap- candidates for implementation in a rather regular proximate singular vectors in the direction of the structure, e.g. using CORDIC processors [27,32]. exact ones. This re-orientation of the singular vec- While the exact SVD-based tracking of »(k)and tors is performed to update a subspace basis which R(k)requiresO(N) operations at each iteration, inputs MUSIC-like estimators and provide the de- Moonen et al. have proposed a QR Jacobi-type sired signal parameter estimates. The accuracy of algorithm with a complexity of O(N) operations per the parameter estimates is linked to the ability of update [17]. In this paper, we shall refer to Moon- the Givens rotations to rotate the proper singular en’s algorithm as MA.
Load more

Recommended publications
  • 2 Homework Solutions 18.335 " Fall 2004
    2 Homework Solutions 18.335 - Fall 2004 2.1 Count the number of ‡oating point operations required to compute the QR decomposition of an m-by-n matrix using (a) Householder re‡ectors (b) Givens rotations. 2 (a) See Trefethen p. 74-75. Answer: 2mn2 n3 ‡ops. 3 (b) Following the same procedure as in part (a) we get the same ‘volume’, 1 1 namely mn2 n3: The only di¤erence we have here comes from the 2 6 number of ‡opsrequired for calculating the Givens matrix. This operation requires 6 ‡ops (instead of 4 for the Householder re‡ectors) and hence in total we need 3mn2 n3 ‡ops. 2.2 Trefethen 5.4 Let the SVD of A = UV : Denote with vi the columns of V , ui the columns of U and i the singular values of A: We want to …nd x = (x1; x2) and such that: 0 A x x 1 = 1 A 0 x2 x2 This gives Ax2 = x1 and Ax1 = x2: Multiplying the 1st equation with A 2 and substitution of the 2nd equation gives AAx2 = x2: From this we may conclude that x2 is a left singular vector of A: The same can be done to see that x1 is a right singular vector of A: From this the 2m eigenvectors are found to be: 1 v x = i ; i = 1:::m p2 ui corresponding to the eigenvalues = i. Therefore we get the eigenvalue decomposition: 1 0 A 1 VV 0 1 VV = A 0 p2 U U 0 p2 U U 3 2.3 If A = R + uv, where R is upper triangular matrix and u and v are (column) vectors, describe an algorithm to compute the QR decomposition of A in (n2) time.
    [Show full text]
  • A Fast Algorithm for the Recursive Calculation of Dominant Singular Subspaces N
    View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Elsevier - Publisher Connector Journal of Computational and Applied Mathematics 218 (2008) 238–246 www.elsevier.com/locate/cam A fast algorithm for the recursive calculation of dominant singular subspaces N. Mastronardia,1, M. Van Barelb,∗,2, R. Vandebrilb,2 aIstituto per le Applicazioni del Calcolo, CNR, via Amendola122/D, 70126, Bari, Italy bDepartment of Computer Science, Katholieke Universiteit Leuven, Celestijnenlaan 200A, 3001 Leuven, Belgium Received 26 September 2006 Abstract In many engineering applications it is required to compute the dominant subspace of a matrix A of dimension m × n, with m?n. Often the matrix A is produced incrementally, so all the columns are not available simultaneously. This problem arises, e.g., in image processing, where each column of the matrix A represents an image of a given sequence leading to a singular value decomposition-based compression [S. Chandrasekaran, B.S. Manjunath, Y.F. Wang, J. Winkeler, H. Zhang, An eigenspace update algorithm for image analysis, Graphical Models and Image Process. 59 (5) (1997) 321–332]. Furthermore, the so-called proper orthogonal decomposition approximation uses the left dominant subspace of a matrix A where a column consists of a time instance of the solution of an evolution equation, e.g., the flow field from a fluid dynamics simulation. Since these flow fields tend to be very large, only a small number can be stored efficiently during the simulation, and therefore an incremental approach is useful [P. Van Dooren, Gramian based model reduction of large-scale dynamical systems, in: Numerical Analysis 1999, Chapman & Hall, CRC Press, London, Boca Raton, FL, 2000, pp.
    [Show full text]
  • Numerical Linear Algebra Revised February 15, 2010 4.1 the LU
    Numerical Linear Algebra Revised February 15, 2010 4.1 The LU Decomposition The Elementary Matrices and badgauss In the previous chapters we talked a bit about the solving systems of the form Lx = b and Ux = b where L is lower triangular and U is upper triangular. In the exercises for Chapter 2 you were asked to write a program x=lusolver(L,U,b) which solves LUx = b using forsub and backsub. We now address the problem of representing a matrix A as a product of a lower triangular L and an upper triangular U: Recall our old friend badgauss. function B=badgauss(A) m=size(A,1); B=A; for i=1:m-1 for j=i+1:m a=-B(j,i)/B(i,i); B(j,:)=a*B(i,:)+B(j,:); end end The heart of badgauss is the elementary row operation of type 3: B(j,:)=a*B(i,:)+B(j,:); where a=-B(j,i)/B(i,i); Note also that the index j is greater than i since the loop is for j=i+1:m As we know from linear algebra, an elementary opreration of type 3 can be viewed as matrix multipli- cation EA where E is an elementary matrix of type 3. E looks just like the identity matrix, except that E(j; i) = a where j; i and a are as in the MATLAB code above. In particular, E is a lower triangular matrix, moreover the entries on the diagonal are 1's. We call such a matrix unit lower triangular.
    [Show full text]
  • Massively Parallel Poisson and QR Factorization Solvers
    Computers Math. Applic. Vol. 31, No. 4/5, pp. 19-26, 1996 Pergamon Copyright~)1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0898-1221/96 $15.00 + 0.00 0898-122 ! (95)00212-X Massively Parallel Poisson and QR Factorization Solvers M. LUCK£ Institute for Control Theory and Robotics, Slovak Academy of Sciences DdbravskA cesta 9, 842 37 Bratislava, Slovak Republik utrrluck@savba, sk M. VAJTERSIC Institute of Informatics, Slovak Academy of Sciences DdbravskA cesta 9, 840 00 Bratislava, P.O. Box 56, Slovak Republic kaifmava©savba, sk E. VIKTORINOVA Institute for Control Theoryand Robotics, Slovak Academy of Sciences DdbravskA cesta 9, 842 37 Bratislava, Slovak Republik utrrevka@savba, sk Abstract--The paper brings a massively parallel Poisson solver for rectangle domain and parallel algorithms for computation of QR factorization of a dense matrix A by means of Householder re- flections and Givens rotations. The computer model under consideration is a SIMD mesh-connected toroidal n x n processor array. The Dirichlet problem is replaced by its finite-difference analog on an M x N (M + 1, N are powers of two) grid. The algorithm is composed of parallel fast sine transform and cyclic odd-even reduction blocks and runs in a fully parallel fashion. Its computational complexity is O(MN log L/n2), where L = max(M + 1, N). A parallel proposal of QI~ factorization by the Householder method zeros all subdiagonal elements in each column and updates all elements of the given submatrix in parallel. For the second method with Givens rotations, the parallel scheme of the Sameh and Kuck was chosen where the disjoint rotations can be computed simultaneously.
    [Show full text]
  • The Sparse Matrix Transform for Covariance Estimation and Analysis of High Dimensional Signals
    The Sparse Matrix Transform for Covariance Estimation and Analysis of High Dimensional Signals Guangzhi Cao*, Member, IEEE, Leonardo R. Bachega, and Charles A. Bouman, Fellow, IEEE Abstract Covariance estimation for high dimensional signals is a classically difficult problem in statistical signal analysis and machine learning. In this paper, we propose a maximum likelihood (ML) approach to covariance estimation, which employs a novel non-linear sparsity constraint. More specifically, the covariance is constrained to have an eigen decomposition which can be represented as a sparse matrix transform (SMT). The SMT is formed by a product of pairwise coordinate rotations known as Givens rotations. Using this framework, the covariance can be efficiently estimated using greedy optimization of the log-likelihood function, and the number of Givens rotations can be efficiently computed using a cross- validation procedure. The resulting estimator is generally positive definite and well-conditioned, even when the sample size is limited. Experiments on a combination of simulated data, standard hyperspectral data, and face image sets show that the SMT-based covariance estimates are consistently more accurate than both traditional shrinkage estimates and recently proposed graphical lasso estimates for a variety of different classes and sample sizes. An important property of the new covariance estimate is that it naturally yields a fast implementation of the estimated eigen-transformation using the SMT representation. In fact, the SMT can be viewed as a generalization of the classical fast Fourier transform (FFT) in that it uses “butterflies” to represent an orthonormal transform. However, unlike the FFT, the SMT can be used for fast eigen-signal analysis of general non-stationary signals.
    [Show full text]
  • Chapter 3 LEAST SQUARES PROBLEMS
    Chapter 3 LEAST SQUARES PROBLEMS E B D C A the sea F One application is geodesy & surveying. Let z = elevation, and suppose we have the mea- surements: zA ≈ 1., zB ≈ 2., zC ≈ 3., zB − zA ≈ 1., zC − zB ≈ 2., zC − zA ≈ 1. 71 This is overdetermined and inconsistent: 0 1 0 1 100 1. B C 0 1 B . C B 010C z B 2 C B C A B . C B 001C @ z A ≈ B 3 C . B − C B B . C B 110C z B 1 C @ 0 −11A C @ 2. A −101 1. Another application is fitting a curve to given data: y y=ax+b x 2 3 2 3 x1 1 y1 6 7 6 7 6 x2 1 7 a 6 y2 7 6 . 7 ≈ 6 . 7 . 4 . 5 b 4 . 5 xn 1 yn More generally Am×nxn ≈ bm where A is known exactly, m ≥ n,andb is subject to inde- pendent random errors of equal variance (which can be achieved by scaling the equations). Gauss (1821) proved that the “best” solution x minimizes kb − Axk2, i.e., the sum of the squares of the residual components. Hence, we might write Ax ≈2 b. Even if the 2-norm is inappropriate, it is the easiest to work with. 3.1 The Normal Equations 3.2 QR Factorization 3.3 Householder Reflections 3.4 Givens Rotations 3.5 Gram-Schmidt Orthogonalization 3.6 Singular Value Decomposition 72 3.1 The Normal Equations Recall the inner product xTy for x, y ∈ Rm. (What is the geometric interpretation of the 3 m T inner product in R ?) A sequence of vectors x1,x2,...,xn in R is orthogonal if xi xj =0 T if and only if i =6 j and orthonormal if xi xj = δij.Twosubspaces S, T are orthogonal if T T x ∈ S, y ∈ T ⇒ x y =0.IfX =[x1,x2,...,xn], then orthonormal means X X = I.
    [Show full text]
  • SVD TR1690.Pdf
    Computer Sciences Department Computing the Singular Value Decomposition of 3 x 3 matrices with minimal branching and elementary floating point operations Aleka McAdams Andrew Selle Rasmus Tamstorf Joseph Teran Eftychios Sifakis Technical Report #1690 May 2011 Computing the Singular Value Decomposition of 3 × 3 matrices with minimal branching and elementary floating point operations Aleka McAdams1,2 Andrew Selle1 Rasmus Tamstorf1 Joseph Teran2,1 Eftychios Sifakis3,1 1 Walt Disney Animation Studios 2 University of California, Los Angeles 3 University of Wisconsin, Madison Abstract The algorithm first determines the factor V by computing the eige- nanalysis of the matrix AT A = VΣ2VT which is symmetric A numerical method for the computation of the Singular Value De- and positive semi-definite. This is accomplished via a modified composition of 3 × 3 matrices is presented. The proposed method- Jacobi iteration where the Jacobi factors are approximated using ology robustly handles rank-deficient matrices and guarantees or- inexpensive, elementary arithmetic operations as described in sec- thonormality of the computed rotational factors. The algorithm is tion 2. Since the symmetric eigenanalysis also produces Σ2, and tailored to the characteristics of SIMD or vector processors. In par- consequently Σ itself, the remaining factor U can theoretically be ticular, it does not require any explicit branching beyond simple obtained as U = AVΣ−1; however this process is not applica- conditional assignments (as in the C++ ternary operator ?:, or the ble when A (and as a result, Σ) is singular, and can also lead to SSE4.1 instruction VBLENDPS), enabling trivial data-level paral- substantial loss of orthogonality in U for an ill-conditioned, yet lelism for any number of operations.
    [Show full text]
  • Discontinuous Plane Rotations and the Symmetric Eigenvalue Problem
    December 4, 2000 Discontinuous Plane Rotations and the Symmetric Eigenvalue Problem Edward Anderson Lockheed Martin Services Inc. [email protected] Elementary plane rotations are one of the building blocks of numerical linear algebra and are employed in reducing matrices to condensed form for eigenvalue computations and during the QR algorithm. Unfortunately, their implementation in standard packages such as EISPACK, the BLAS and LAPACK lack the continuity of their mathematical formulation, which makes results from software that use them sensitive to perturbations. Test cases illustrating this problem will be presented, and reparations to the standard software proposed. 1.0 Introduction Unitary transformations are frequently used in numerical linear algebra software to reduce dense matrices to bidiagonal, tridiagonal, or Hessenberg form as a preliminary step towards finding the eigenvalues or singular values. Elementary plane rotation matrices (also called Givens rotations) are used to selectively reduce values to zero, such as when reducing a band matrix to tridiagonal form, or during the QR algorithm when attempting to deflate a tridiagonal matrix. The mathematical properties of rota- tions were studied by Wilkinson [9] and are generally not questioned. However, Wilkinson based his analysis on an equation for computing a rotation matrix that is different from those in common use in the BLAS [7] and LAPACK [2]. While the classical formula is continuous in the angle of rotation, the BLAS and LAPACK algorithms have lines of discontinuity that can lead to abrupt changes in the direction of computed eigenvectors. This makes comparing results difficult and proving backward stability impossible. It is easy to restore continuity in the generation of Givens rotations even while scaling to avoid overflow and underflow, and we show how in this report.
    [Show full text]
  • CORDIC-Based LMMSE Equalizer for Software Defined Radio
    CORDIC-Based LMMSE Equalizer for Software Defined Radio Murugappan SENTHILVELAN † §, Javier HORMIGO ‡, Joon Hwa CHUN €, Mihai SIMA *, Daniel IANCU †¶, Michael SCHULTE §, John GLOSSNER ¥ †Optimum Semiconductor Technologies, Inc., Tarrytown, NY, U.S.A. ‡ University of Malaga, Department of Computer Architecture, Malaga, Spain. € Intelligent Automation Inc., Rockville, MD, U.S.A. * University of Victoria, Department of Electrical and Computer Engineering, Victoria, BC, Canada. § University of Wisconsin-Madison, Department of Electrical and Computer Engineering, Madison, WI, U.S.A. ¶ Tampere University of Technology, Korkeakoulunkatu 1, FIN-33720 Tampere, Finland. ¥ Sandbridge Technologies, Inc., Tarrytown, NY, U.S.A. e-mail: [email protected], [email protected], [email protected] [email protected], [email protected], [email protected], [email protected] Abstract — In Code Division Multiple Access (CDMA) interference, and with synchronized transmission of blocks systems, the orthogonality of the spreading codes used to over the channel, multiple users can be multiplexed over the achieve multiple access over a channel is severely degraded same channel by using a different spreading code for each due to multi-path interference. Expensive equalization user. Even though signals from multiple users interfere with techniques are needed to recover the transmitted signal. each other in the channel, due to the orthogonal spreading The Linear Minimum Mean Square Error (LMMSE) codes, each user can extract the transmission intended for equalizer is a sub-optimal equalizer that is a good them from the received signal by despreading it using their compromise between computational complexity and spreading code. communication system performance. It uses While the spreading codes are perfectly orthogonal to computationally-intensive matrix inversion operations to each other, they are not orthogonal to time-shifted versions of perform equalization.
    [Show full text]
  • On Computing Givens Rotations Reliably and Efficiently
    On Computing Givens Rotations Reliably and Efficiently DAVID BINDEL, JAMES DEMMEL, and WILLIAM KAHAN University of California, Berkeley and OSNI MARQUES Lawrence Berkeley National Laboratory We consider the efficient and accurate computation of Givens rotations.p When f and g arep positive real numbers,p this simply amounts to computing the values of c = f / f 2 + g 2, s = g/ f 2 + g 2, and r = f 2 + g 2. This apparently trivial computation merits closer consideration for the follow- ing three reasons. First, while the definitions of c, s and r seem obvious in the case of two nonnegative arguments f and g, there is enough freedom of choice when one or more of f and g are negative, zero or complex that LAPACK auxiliary routines SLARTG, CLARTG, SLARGV and CLARGV can compute rather different values of c, s and r for mathematically identical values of f and g.To eliminate this unnecessary ambiguity, the BLAS Technical Forum chose a single consistent defini- tion of Givens rotations that we will justify here. Second, computing accurate values of c, s and r as efficiently as possible and reliably despite over/underflow is surprisingly complicated. For complex Givens rotations, the most efficient formulas require only one real square root and one real divide (as well as several much cheaper additions and multiplications), but a reliable implementation using only working precision has a number of cases. On a Sun Ultra-10, the new implementation is slightly faster than the previous LAPACK implementation in the most common case, and 2.7 to 4.6 times faster than the corresponding vendor, reference or ATLAS routines.
    [Show full text]
  • The Holy Grail of Adam's Ale Locating Aquifers Through Geostatistic Modeling
    The Holy Grail of Adam's Ale Locating Aquifers through Geostatistic Modeling Team #65 April 7, 2010 New Mexico Supercomputing Challenge Final Report Los Alamos High School Team Members: Gabriel Montoya Rachel Robey Orli Shlachter Orion Staples Teacher Sponsor: Lee Goodwin Mentors: Robert Robey Thomas Robey Contents 1 Introduction 6 1.1 Problem Statement . 6 1.2 Objective . 6 1.3 Backround . 6 1.3.1 Geostatistics . 6 1.3.2 Aquifers . 7 2 Mathematical Models 8 2.1 Semi-variogram . 8 2.2 Interpolation . 9 2.2.1 Inverse Distance Weighting . 10 2.2.2 Kriging . 10 2.3 Sampling from Gaussian Distribution . 12 3 Computational Model 13 3.1 Semi-variogram . 13 3.2 Solution of Kriging Equations . 14 3.3 Anisotropy . 17 3.4 Sampling from Gaussian Distribution . 18 3.5 Multiple Points and Runs . 18 3.6 Optimization . 19 3.6.1 Variation on Random Iteration Algorithm . 19 3.6.2 Transferring Matrix Solver to the GPU . 20 4 Code 23 4.1 Overview and Structure . 23 4.2 WxWidget Windowing . 24 4.3 Computational . 26 4.4 OpenGL Graphics . 26 4.5 wxPlotCtrl Graphing . 28 5 Results 29 5.1 Case Study . 29 5.1.1 Data . 29 5.1.2 Semi-variogram . 31 5.1.3 Mixed Success of Interpolated Fields . 32 6 Conclusions 33 6.1 Current Status . 34 7 Teamwork 34 8 Recommendations 35 2 A References 35 A.1 Bibliography . 35 A.2 Software/Tools . 36 A.3 Acknowledgments . 36 B Glossary 37 C User Guide 37 3 List of Figures 1 Diagram of an aquifer[8].
    [Show full text]
  • Restructuring the QR Algorithm for High-Performance Application of Givens Rotations
    Restructuring the QR Algorithm for High-Performance Application of Givens Rotations FLAME Working Note #60 Field G. Van Zee Robert A. van de Geijn Department of Computer Science The University of Texas at Austin Austin, TX 78712 Gregorio Quintana-Ort´ı Departamento de Ingenier´ıay Ciencia de Computadores Universidad Jaume I 12.071, Castell´on, Spain October 18, 2011 Abstract We show how the QR algorithm can be restructured so that it becomes rich in operations that can achieve near-peak performance on a modern processor. The key is a novel, cache-friendly algorithm for applying multiple sets of Givens rotations, which is then implemented with optimizations that (1) leverage vector instruction units to increase floating-point throughput, and (2) fuse multiple rotations to decrease the total number of memory operations. We demonstrate the merits of this new QR algorithm for computing the Hermitian (symmetric) eigenvalue decomposition and singular value decomposition of dense matrices when all eigenvectors/singular vectors are computed. The approach yields vastly im- proved performance relative to the traditional QR algorithm and is competitive with two commonly used alternatives—Cuppen’s Divide and Conquer algorithm and the Method of Multiple Relatively Robust Representations—while inheriting the more modest O(n) workspace requirements of the original QR algorithm. Since the computations performed by the restructured algorithm remain essentially identical to those performed by the original method, robust numerical properties are preserved. 1 Introduction The QR algorithm is taught in a typical graduate-level numerical linear algebra course, and despite being among the most accurate methods for solving eigenvalue and singular value problems, it is not used much in practice because its performance is not competitive [44, 19, 40, 12].
    [Show full text]