Iterative solution of a class of linear equations with application to reconstruction of three-dimensional object arrays†

An iterative algorithm baaed on probabilistic estimation is described for obtaining the minimum-norm solution of a very large, consistent, linear system of equations AX = g where A is an (m × n) matrix with non-negative elements, x and g are respectively (n × 1) and (m × 1) vectors with positive components.

This algorithm will find application in the reconstruction of three-dimensional object arrays from projections and in several other areas.     

Approximation of frequency responses based on Larsson's method

This paper discusses the new aspects of the formulation for structural model updates presented by Larsson and Sas at the 10th International Modal Analysis Conference held in San Diego, California in February 1991. First, a closed-form representation of the reduced impedance matrix is presented. It clarifies the issues associated with the range of the excitation frequencies used in model updates. Second, the applicable range of excitation frequencies is considered through an example where low frequency local modes are present. Third, the computational aspects of this algorithm are studied. It is shown that displacement sensitivity vectors corresponding to the virtual loads are critical in the computation of the reduced impedance matrix sensitivity. Finally, applicability of this algorithm for approximate frequency responses in structural optimization is studied. Numerical results are included to illustrate the essence of this formulation.[B1] Viscous damping matrix (N × N) - [B2] Structural damping matrix (N × N) - [E] Impedance matrix (N × N) [E] = [K+i(B ₁+B₂)–²M] - F Forcing vector used for analysis (N × 1) - G Measured force vector (N × 1) - [H] Receptance matrix (N × N) [H] = [E]^–1 - J Number of independent loading cases - [K] Stiffness matrix (N × N) - K Number of degrees of freedom where frequency responses are measured - KT Total number of responses for which measured data are available - L Number of excitation frequencies for each loading case - [M] Mass matrix (N × N) - M Number of parameters modified for system identification or design optimization - N Total number of structural system degrees of freedom in the analysis model - R Structural response - Q Number of vectors used in the modal analysis - U Frequency responses computed analytically - V Frequency responses measured in the shake tests - 0 Status prior to perturbation - model Quantities computed by an analysis model - red Indicates a reduced system obtained by isolating the measuredK degrees of freedom - test Quantities measured in the vibration tests - ~ Approximate quantity     

A Fast Spectral Subtractional Solver for Elliptic Equations

The paper presents a fast subtractional spectral algorithm for the solution of the Poisson equation and the Helmholtz equation which does not require an extension of the original domain. It takes O(N ² log N) operations, where N is the number of collocation points in each direction. The method is based on the eigenfunction expansion of the right hand side with integration and the successive solution of the corresponding homogeneous equation using Modified Fourier Method. Both the right hand side and the boundary conditions are not assumed to have any periodicity properties. This algorithm is used as a preconditioner for the iterative solution of elliptic equations with non-constant coefficients. The procedure enjoys the following properties: fast convergence and high accuracy even when the computation employs a small number of collocation points. We also apply the basic solver to the solution of the Poisson equation in complex geometries.     

Two algorithms for constructing a Delaunay triangulation

This paper provides a unified discussion of the Delaunay triangulation. Its geometric properties are reviewed and several applications are discussed. Two algorithms are presented for constructing the triangulation over a planar set ofN points. The first algorithm uses a divide-and-conquer approach. It runs inO(N logN) time, which is asymptotically optimal. The second algorithm is iterative and requiresO(N ²) time in the worst case. However, its average case performance is comparable to that of the first algorithm.This work was supported in part by the National Science Foundation under grant MCS-76-17321 and the Joint Services Electronics Program under contract DAAB-07-72-0259.     

An iterative algorithm for solving a class of matrix equations

In this paper, an iterative algorithm is presented to solve the Sylvester and Lyapunov matrix equations. By this iterative algorithm, for any initial matrix X1, a solution X＊ can be obtained within finite iteration steps in the absence of roundoff errors. Some examples illustrate that this algorithm is very efficient and better than that of [ 1 ] and [2].     

A numerical method for determining monotonicity and convergence rate in iterative learning control

In iterative learning control (ILC), a lifted system representation is often used for design and analysis to determine the convergence rate of the learning algorithm. Computation of the convergence rate in the lifted setting requires construction of large N×N matrices, where N is the number of data points in an iteration. The convergence rate computation is O(N²) and is typically limited to short iteration lengths because of computational memory constraints. As an alternative approach, the implicitly restarted Arnoldi/Lanczos method (IRLM) can be used to calculate the ILC convergence rate with calculations of O(N). In this article, we show that the convergence rate calculation using IRLM can be performed using dynamic simulations rather than matrices, thereby eliminating the need for large matrix construction. In addition to faster computation, IRLM enables the calculation of the ILC convergence rate for long iteration lengths. To illustrate generality, this method is presented for multi-input multi-output, linear time-varying discrete-time systems.     

Fast solution of large N×N matrix equations in an MIMD–SIMD Hybrid System

In this paper, we propose a new high-speed computation algorithm for solving a large N×N matrix system using the MIMD–SIMD Hybrid System. The MIMD–SIMD Hybrid System (also denoted as Hybrid System in this paper) is a new parallel architecture consisting of a combination of Cluster of Workstations (COWs) and SIMD systems working concurrently to produce an optimal parallel computation. We first introduce our prototype SIMD system and our Hybrid System setup before presenting how it can be implemented to find the unknowns in a large N×N linear matrix equation system using the Gauss–LU algorithm. This algorithm basically performs the ‘Divide and Conquer’ approach by breaking down the large N×N matrix system into a manageable 32 × 32 matrix for fast computation.     

A New and Faster Gaussian Elimination Based Fault Tolerant Systolic Linear System Solver

This paper presents a new systolic algorithm for thecompletesolution of a system ofNlinear equations in (N²/2 +O(N)) time steps using 2Nprocessing elements (PEs). It is based on a variant of the Gaussian elimination (GE) algorithm called the successive GE and is faster than any existing GE based algorithm usingO(N) PEs. We also suggest two fault tolerant schemes that tolerate up toNPE failures. The first scheme is a time redundancy based approach with no hardware overhead and 100% time overhead. This scheme can tolerate up toNPE failures. The second scheme is based on algorithm based fault tolerance (ABFT) and usesNextra PEs to tolerate up toN− 1 PE failures with very little time overhead. The number of errors that can be detected/corrected in both schemes is more than that in any existing fault tolerant systolic array.     

三维图匹配算法

目的 现有的图匹配算法大多应用于二维图像,对三维图像的特征点匹配存在匹配准确率低和计算速度慢等问题。为解决这些问题,本文将分解图匹配算法扩展应用在了三维图像上。方法 首先将需要匹配的两个三维图像的特征点作为图的节点集;再通过Delaunay三角剖分算法,将三维特征点相连,则相连得到的边就作为图的边集,从而建立有向图;然后,根据三维图像的特征点构建相应的三维有向图及其邻接矩阵;再根据有向图中的节点特征和边特征分别构建节点特征相似矩阵和边特征相似矩阵;最后根据这两个特征矩阵将节点匹配问题转化为求极值问题并求解。结果 实验表明,在手工选取特征点的情况下,本文算法对相同三维图像的特征点匹配有97.56%的平均准确率;对不同三维图像特征点匹配有76.39%的平均准确率;在三维图像有旋转的情况下,有90%以上的平均准确率;在特征点部分缺失的情况下,平均匹配准确率也能达到80%。在通过三维尺度不变特征变换（SIFT）算法得到特征点的情况下,本文算法对9个三维模型的特征点的平均匹配准确率为98.78%。结论 本文提出的基于图论的三维图像特征点匹配算法,经实验结果验证,可以取得较好的匹配效果。     

An iterative algorithm for a least squares solution of a matrix equation

In this article, we propose an iterative algorithm to compute the minimum norm least-squares solution of AXB+CYD=E, based on a matrix form of the algorithm LSQR for solving the least squares problem. We then apply this algorithm to compute the minimum norm least-squares centrosymmetric solution of min _X ‖AXB?E‖ _F . Numerical results are provided to verify the efficiency of the proposed method.     

An adaptive global reduction algorithm for wormhole-routed 2D meshes

This paper presents a global reduction algorithm for wormhole-routed 2D meshes. Well-known reduction algorithms that are optimized for short vectors have complexity O(M log N), where N = n × n is the number of nodes, and M the vector length. Algorithms suitable for long vectors have complexity O(√N + M). Previously known asymptotically optimal algorithms with complexity O(log N + M) incur inherent network contention among constituent messages. The proposed algorithm adapts to the given vector length, resulting in complexities O(M log N) for short vectors, O(log N + M) for medium-sized vectors, and O(√N + M) for sufficiently long vectors. The O(√N + M) version is preferred to the O(log N + M) version for long vectors, due to its small coefficient associated with M, the dominating factor for such vectors. The algorithm is contention-free in a synchronous environment. Under asynchronous execution models, depth contention (contention among message-passing steps) may occur. However, simulation studies show that the effect of depth contention on the actual performance is negligible.     

Parallel algorithms for finding polynomial Roots on OTIS-torus

We present two parallel algorithms for finding all the roots of an N-degree polynomial equation on an efficient model of Optoelectronic Transpose Interconnection System (OTIS), called OTIS-2D torus. The parallel algorithms are based on the iterative schemes of Durand–Kerner and Ehrlich methods. We show that the algorithm for the Durand–Kerner method requires (N ^0.75+0.5N ^0.25−1) electronic moves + 2(N ^0.5−1) OTIS moves using N processors. The parallel algorithm for Ehrlich method is shown to run in (N ^0.75+0.5N ^0.25−1) electronic moves + 2(N ^0.5−1) OTIS moves with the same number of processors. The algorithms have lower AT cost than the algorithms presented in Jana (Parallel Comput 32:301–312, 2006). The scalability of the algorithms is also discussed.     

New iterative linear matrix inequality based procedure for H2 and H∞ state feedback control of continuous‐time polytopic systems

This article provides new linear matrix inequality (LMI) sufficient conditions for a generalized robust state feedback control synthesis problem for linear continuous‐time polytopic systems. This generalized problem includes the robust stability, H₂ ‐norm, and H_∞ ‐norm problems as special cases. Using a novel general separation result, which separates the state feedback gain from the Lyapunov matrix but with the state feedback gain synthesized from the slack variable, then allows the formulation of LMI sufficient conditions for the generalized problem. Compared to existing parameterized LMI based conditions, where auxiliary scalar parameters are introduced in order to include the quadratic stability conditions (ie, assuming a constant Lyapunov matrix) as a special case, the proposed new conditions are true LMIs and contain as a particular case the optimal quadratic stability solution. Utilizing any initial solution derived by the quadratic or some existing methods as a starting solution, we propose an algorithm based on an iterative procedure, which is recursively feasible in each update, to compute a sequence of nonincreasing upper bounds for the H₂ ‐norm and H_∞ ‐norm. In addition, if no feasible initial solution can be found for some uncertain systems using any existing methods, another algorithm is presented that offers the possibility of obtaining a robust stabilizing gain. Numerical examples from the literature demonstrate that our algorithms can provide less conservative results than existing methods, and they can also find feasible solutions where all other methods fail.     

Selective sampling for approximate clustering of very large data sets

A key challenge in pattern recognition is how to scale the computational efficiency of clustering algorithms on large data sets. The extension of non‐Euclidean relational fuzzy c‐means (NERF) clustering to very large (VL = unloadable) relational data is called the extended NERF (eNERF) clustering algorithm, which comprises four phases: (i) finding distinguished features that monitor progressive sampling; (ii) progressively sampling from a N × N relational matrix R_N to obtain a n × n sample matrix R_n; (iii) clustering R_n with literal NERF; and (iv) extending the clusters in R_n to the remainder of the relational data. Previously published examples on several fairly small data sets suggest that eNERF is feasible for truly large data sets. However, it seems that phases (i) and (ii), i.e., finding R_n, are not very practical because the sample size n often turns out to be roughly 50% of n, and this over‐sampling defeats the whole purpose of eNERF. In this paper, we examine the performance of the sampling scheme of eNERF with respect to different parameters. We propose a modified sampling scheme for use with eNERF that combines simple random sampling with (parts of) the sampling procedures used by eNERF and a related algorithm sVAT (scalable visual assessment of clustering tendency). We demonstrate that our modified sampling scheme can eliminate over‐sampling of the original progressive sampling scheme, thus enabling the processing of truly VL data. Numerical experiments on a distance matrix of a set of 3,000,000 vectors drawn from a mixture of 5 bivariate normal distributions demonstrate the feasibility and effectiveness of the proposed sampling method. We also find that actually running eNERF on a data set of this size is very costly in terms of computation time. Thus, our results demonstrate that further modification of eNERF, especially the extension stage, will be needed before it is truly practical for VL data. © 2008 Wiley Periodicals, Inc.     

Efficient Concurrent L1-Minimization Solvers on GPUs

Given that the concurrent L1-minimization (L1-min) problem is often required in some real applications, we investigate how to solve it in parallel on GPUs in this paper. First, we propose a novel self-adaptive warp implementation of the matrix-vector multiplication (Ax) and a novel self-adaptive thread implementation of the matrix-vector multiplication (A^Tx), respectively, on the GPU. The vector-operation and inner-product decision trees are adopted to choose the optimal vector-operation and inner-product kernels for vectors of any size. Second, based on the above proposed kernels, the iterative shrinkage-thresholding algorithm is utilized to present two concurrent L1-min solvers from the perspective of the streams and the thread blocks on a GPU, and optimize their performance by using the new features of GPU such as the shuffle instruction and the read-only data cache. Finally, we design a concurrent L1-min solver on multiple GPUs. The experimental results have validated the high effectiveness and good performance of our proposed methods.     

Piece wise linear least–squares approximation of planar curves

Two algorithms for solving the piecewise linear least–squares approximation problem of plane curves are presented. The first is for the case when the L ₂ residual (error) norm in any segment is not to exceed a pre–assigned value. The second algorithm is for the case when the number of segments is given and a (balanced) L ₂ residual norm solution is required. The given curve is first digitized and either algorithm is then applied to the discrete points. For each segment, we obtain the upper triangular matrix R in the QR factorization of the (augmented) coefficient matrix of the resulting system of linear equations. The least–squares solutions are calculated in terms of the R (and Q) matrices. The algorithms then work in an iterative manner by updating the least–squares solutions for the segments via up dating the R matrices. The calculation requires as little computational effort as possible. Numerical results and comments are given. This, in a way, is a tutorial paper.     

Two-Pass Rearrangeability in Faulty Benes Networks

Existing fault-tolerant routing schemes for Benes networks either consider only the control line stuck-at faults, or handle the switch faults by some graceful degradation routing schemes that reconfigure the network into a smaller system with minimal loss. Now, even in the presence of a single switch fault in anN×NBenes networkB(n), (n= log₂N), noN×Npermutation can be realized in a single pass. In this paper, we attempt to characterize the switch fault sets inB(n), in the presence of which the network is always capable of realizing any arbitraryN×NpermutationPin two passes, such that any source–destination path is set up in a single pass, no recirculation is needed, but the whole set ofNsource–destination paths ofPis partitioned in two subsets and are realized in two successive passes. We propose an algorithm that will detect if the switch fault set present in aB(n), belongs to this class; if it is yes, we present another algorithm that computes the fault-tolerant routing to realize any arbitrary permutationPin two passes. This scheme enables us to makeB(n) fault-tolerant in the presence of a restricted class of multiple switch faults, without any recirculation through intermediate nodes, or any reconfiguration of the system.     

Recursive Direct Learning of Control Efforts for Trajectories with Different Magnitude Scales

Direct learning control (DLC) schemes have been developed recently to address non‐repeatable trajectory tracking problems. Unlike conventional iterative learning schemes, DLC schemes learn a set of unknown basis function vectors which can be used to generate the desired control profile of a new trajectory. DLC schemes use all available trajectory tracking information to obtain the unknown basis function vectors in a Least Squares and pointwise manner. A drawback of DLC is that the inverse matrix calculation is inevitable, which is time consuming and may result in singularities due to the batch processing nature. A Recursive Direct Learning Control method is proposed which learns the basis function vectors meanwhile overcomes the implementation difficulties in DLC schemes. The focus of this paper is on learning the control profile of trajectories with same operation period but different magnitude scales. The recursive learning method makes use of one trajectory information at a time, thus avoids the batch processing. The scheme is first developed for a class of nonlinear time varying systems and then extended to cover more general classes of nonlinear systems including robotic manipulator dynamics. Extensive simulation results on a two‐link robotic model are provided to confirm the features of the proposed algorithm.