PARALLEL SUBSPACE ITERATION METHOD FOR THE SPARSE SYMMETRIC EIGENVALUE PROBLEM

2006 ◽  
Vol 05 (04) ◽  
pp. 801-818 ◽  
Author(s):  
RICHARD LOMBARDINI ◽  
BILL POIRIER
Keyword(s):  
Eigenvalue Problem ◽  
Iteration Method ◽  
Degrees Of Freedom ◽  
Sparse Matrices ◽  
Model Systems ◽  
Symmetric Matrices ◽  
Six Degrees Of Freedom ◽  
Subspace Iteration ◽  
The Matrix ◽  
Parallel Iterative

A new parallel iterative algorithm for the diagonalization of real sparse symmetric matrices is introduced, which uses a modified subspace iteration method. A novel feature is the preprocessing of the matrix prior to iteration, which allows for a natural parallelization resulting in a great speedup and scalability of the method with respect to the number of compute nodes. The method is applied to Hamiltonian matrices of model systems up to six degrees of freedom, represented in a truncated Weyl–Heisenberg wavelet (or "weylet") basis developed by one of the authors (Poirier). It is shown to accurately determine many thousands of eigenvalues for sparse matrices of the order N ≈ 106, though much larger matrices may also be considered.

scholarly journals A Combination of Modal Synthesis and Subspace Iteration for an Efficient Algorithm for Modal Analysis within a FE-Code

2003 ◽  
Vol 10 (1) ◽  
pp. 27-35 ◽  
Author(s):  
M.W. Zehn
Keyword(s):  
Eigenvalue Problem ◽  
Modal Analysis ◽  
Efficient Algorithm ◽  
Degrees Of Freedom ◽  
Synthesis Method ◽  
Problem Solver ◽  
Synthesis Methods ◽  
Combined Method ◽  
Subspace Iteration ◽  
Modal Synthesis

Various well-known modal synthesis methods exist in the literature, which are all based upon certain assumptions for the relation of generalised modal co-ordinates with internal modal co-ordinates. If employed in a dynamical FE substructure/superelement technique the generalised modal co-ordinates are represented by the master degrees of freedom (DOF) of the master nodes of the substructure. To conduct FE modal analysis the modal synthesis method can be integrated to reduce the number of necessary master nodes or to ease the process of defining additional master points within the structure. The paper presents such a combined method, which can be integrated very efficiently and seamless into a special subspace eigenvalue problem solver with no need to alter the FE system matrices within the FE code. Accordingly, the merits of using the new algorithm are the easy implementation into a FE code, the less effort to carry out modal synthesis, and the versatility in dealing with superelements. The paper presents examples to illustrate the proper work of the algorithm proposed.

Eigenvalue Problem for Uncertain Systems

1991 ◽  
Vol 44 (11S) ◽  
pp. S89-S95 ◽  
Author(s):  
Mircea Grigoriu
Keyword(s):  
Eigenvalue Problem ◽  
Perturbation Method ◽  
Random Process ◽  
Uncertain Systems ◽  
Stochastic Matrix ◽  
Symmetric Matrices ◽  
Zero Crossings ◽  
Level Zero ◽  
The Matrix ◽  
The Relationship

Methods are developed for calculating probabilistic characteristics of the eigenvalues of stochastic symmetric matrices. The methods are based on the relationship between the elements of a matrix and its eigenvalues, perturbation method, bounds on eigenvalues, and zero-crossings of the characteristic polynomial. It is shown that the polynomial characteristic of a stochastic matrix can be viewed as a random process whose crossings of level zero define the eigenvalues of the matrix. The proposed methods of analysis are demonstrated by examples from dynamics and elasticity.

Advances in Classical Subspace Iteration Method for Eigenvalue Problem

2007 ◽  
pp. 194-205
Author(s):  
P. Chen ◽  
Q. C. Zhao ◽  
W. B. Peng ◽  
Y. Q. Chen ◽  
M. W. Yuan
Keyword(s):  
Eigenvalue Problem ◽  
Iteration Method ◽  
Subspace Iteration

scholarly journals The Extended Hamiltonian Algorithm for the Solution of the Algebraic Riccati Equation

2014 ◽  
Vol 2014 ◽  
pp. 1-8 ◽  
Author(s):  
Zhikun Luo ◽  
Huafei Sun ◽  
Xiaomin Duan
Keyword(s):  
Riccati Equation ◽  
Iteration Method ◽  
Learning Algorithm ◽  
Positive Definite ◽  
Gradient Algorithm ◽  
Algebraic Riccati Equation ◽  
Symmetric Matrices ◽  
Algebraic Riccati Equations ◽  
Subspace Iteration ◽  
Hamiltonian Algorithm

We use a second-order learning algorithm for numerically solving a class of the algebraic Riccati equations. Specifically, the extended Hamiltonian algorithm based on manifold of positive definite symmetric matrices is provided. Furthermore, this algorithm is compared with the Euclidean gradient algorithm, the Riemannian gradient algorithm, and the new subspace iteration method. Simulation examples show that the convergence speed of the extended Hamiltonian algorithm is the fastest one among these algorithms.

Vibration of a Spring-Supported Body

1965 ◽  
Vol 7 (2) ◽  
pp. 185-192 ◽  
Author(s):  
P. Grootenhuis ◽  
D. J. Ewins
Keyword(s):  
Rigid Body ◽  
Degrees Of Freedom ◽  
Equations Of Motion ◽  
Vertical Axis ◽  
The Body ◽  
Stiffness Ratio ◽  
Six Degrees Of Freedom ◽  
Centre Of Gravity ◽  
Longitudinal Stiffness ◽  
The Matrix

The equations of motion for a rigid body supported on four springs are derived for the general case of the centre-of-gravity being anywhere within the body and allowing for the sideways as well as the longitudinal stiffnesses of the springs. This constitutes a six-degrees-of-freedom case with three degrees of asymmetry. Coupling between motions in all directions occurs even when the centre-of-gravity is at the geometric centre with the exception then of vertical oscillations and rotation about the vertical axis. Any number of additional springs can be allowed for by adding terms to the expression for the potential energy stored in the springs. Allowance is made in the expression for kinetic energy for the products of inertia which arise with an offset centre-of-gravity. The real case is simulated for purposes of analysis by replacing the rigid body by a rectangular box with a light framework and all the mass concentrated at the eight corners. The matrix solution is changed into dimensionless parameters and the effect of an offset centre-of-gravity upon the eigenvalues and eigenvectors studied. Only the proportions of the box and the stiffness ratio between sideways to longitudinal stiffness of the springs remain as factors. The numerical example given is for proportions of height to width to length of 3/4/5 and for a stiffness ratio of 5. Small amounts of offset of the centre-of-gravity from the geometric centre do not alter the dynamic behaviour of the system much but displacing the total mass towards either a lower or an upper corner has marked effects. Some of the natural frequencies associated with motion in rotation when the system is symmetric become less than the frequencies connected with motion in translation for the centre-of-gravity being close to a corner connected to a spring. A large region free from any natural frequency arises when the centre-of-gravity is moved towards a corner furthest removed from the plane containing the springs. The asymptotic conditions for the position of the centre-of-gravity are also considered.

Excitonic Coupled-cluster Theory: Part I, General Formalism

2017 ◽  
Author(s):  
Yuhong Liu ◽  
Anthony Dutoi
Keyword(s):  
Electron Correlation ◽  
Degrees Of Freedom ◽  
Model Systems ◽  
Coupled Cluster ◽  
Effective Hamiltonian ◽  
Coupled Cluster Theory ◽  
Cluster Theory ◽  
Companion Article ◽  
High Level ◽  
Fragment Interaction

<div> <div>A shortcoming of presently available fragment-based methods is that electron correlation (if included) is described at the level of individual electrons, resulting in many redundant evaluations of the electronic relaxations associated with any given fluctuation. A generalized variant of coupled-cluster (CC) theory is described, wherein the degrees of freedom are fluctuations of fragments between internally correlated states. The effects of intra-fragment correlation on the inter-fragment interaction is pre-computed and permanently folded into the effective Hamiltonian. This article provides a high-level description of the CC variant, establishing some useful notation, and it demonstrates the advantage of the proposed paradigm numerically on model systems. A companion article shows that the electronic Hamiltonian of real systems may always be cast in the form demanded. This framework opens a promising path to build finely tunable systematically improvable methods to capture precise properties of systems interacting with a large number of other systems. </div> </div>

Excitonic Coupled-cluster Theory: Part I, General Formalism

2017 ◽  
Author(s):  
Yuhong Liu ◽  
Anthony Dutoi
Keyword(s):  
Electron Correlation ◽  
Degrees Of Freedom ◽  
Model Systems ◽  
Coupled Cluster ◽  
Effective Hamiltonian ◽  
Coupled Cluster Theory ◽  
Cluster Theory ◽  
Companion Article ◽  
High Level ◽  
Fragment Interaction

<div> <div>A shortcoming of presently available fragment-based methods is that electron correlation (if included) is described at the level of individual electrons, resulting in many redundant evaluations of the electronic relaxations associated with any given fluctuation. A generalized variant of coupled-cluster (CC) theory is described, wherein the degrees of freedom are fluctuations of fragments between internally correlated states. The effects of intra-fragment correlation on the inter-fragment interaction is pre-computed and permanently folded into the effective Hamiltonian. This article provides a high-level description of the CC variant, establishing some useful notation, and it demonstrates the advantage of the proposed paradigm numerically on model systems. A companion article shows that the electronic Hamiltonian of real systems may always be cast in the form demanded. This framework opens a promising path to build finely tunable systematically improvable methods to capture precise properties of systems interacting with a large number of other systems. </div> </div>

Sign in / Sign up

Export Citation Format

Share Document