Newton‐Krylov iterative matrix representative spectrum

The concept of a representative spectrum is introduced in the context of Newton‐Krylov methods. This concept allows a better understanding of convergence rate accelerating techniques for Krylov‐subspace iterative methods in the context of CFD applications of the Newton‐Krylov approach to iteratively solve sets of non‐linear equations. The dependence of the representative spectrum on several parameters such as mesh, Mach number or discretization techniques is studied and analyzed. © 2005 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2006     

Accuracy of multivalue methods during change of time‐step size: Adams‐Moulton

Multivalue methods are a class of time‐stepping procedures for numerical solution of differential equations that progress to a new time level using the approximate solution for the function of interest and its derivatives at a single time level. The methods differ from multistep procedures, which make use of solutions to the differential equation at multiple time levels to advance to the new time level. Multistep methods are difficult to employ when a change in time‐step is desired, because the standard formulas (e.g., Adams‐Moulton or Gear) must be modified to accommodate the change. Multivalue methods seem to possess the desirable feature that the time‐step may be changed arbitrarily as one proceeds, since the solution proceeds from a single time level. However, in practice, changes in the time‐step introduce lower order errors or alter the coefficient in the truncation error term. Here, the multivalue Adams‐Moulton method is presented based on a general interpolation procedure. Modifications required to retain the high‐order accuracy of these methods during a change in time‐step are developed. Additionally, a formula for the unknown initial derivatives is presented. Finally, two examples are provided to illustrate the potential merit of the modification to the standard multivalue methods. © 2000 John Wiley & Sons, Inc. Numer Methods Partials Differential Eq 16: 312–326, 2000     

Compress‐and‐restart block Krylov subspace methods for Sylvester matrix equations

Block Krylov subspace methods (KSMs) comprise building blocks in many state‐of‐the‐art solvers for large‐scale matrix equations as they arise, for example, from the discretization of partial differential equations. While extended and rational block Krylov subspace methods provide a major reduction in iteration counts over polynomial block KSMs, they also require reliable solvers for the coefficient matrices, and these solvers are often iterative methods themselves. It is not hard to devise scenarios in which the available memory, and consequently the dimension of the Krylov subspace, is limited. In such scenarios for linear systems and eigenvalue problems, restarting is a well‐explored technique for mitigating memory constraints. In this work, such restarting techniques are applied to polynomial KSMs for matrix equations with a compression step to control the growing rank of the residual. An error analysis is also performed, leading to heuristics for dynamically adjusting the basis size in each restart cycle. A panel of numerical experiments demonstrates the effectiveness of the new method with respect to extended block KSMs.     

Block truncated-Newton methods for parallel optimization

Truncated-Newton methods are a class of optimization methods suitable for large scale problems. At each iteration, a search direction is obtained by approximately solving the Newton equations using an iterative method. In this way, matrix costs and second-derivative calculations are avoided, hence removing the major drawbacks of Newton's method. In this form, the algorithms are well-suited for vectorization. Further improvements in performance are sought by using block iterative methods for computing the search direction. In particular, conjugate-gradient-type methods are considered. Computational experience on a hypercube computer is reported, indicating that on some problems the improvements in performance can be better than that attributable to parallelism alone.Partially supported by Air Force Office of Scientific Research grant AFOSR-85-0222.Partially supported by National Science Foundation grant ECS-8709795, co-funded by the U.S. Air Force Office of Scientific Research.     

Modified two‐step hybrid methods for the numerical integration of oscillatory problems

The construction of modified two‐step hybrid methods for the numerical solution of second‐order initial value problems with periodic or oscillatory behavior is considered. The coefficients of the new methods depend on the frequency of each problem so that the harmonic oscillator is integrated exactly. Numerical experiments indicate that the new methods are more efficient than existing methods with constant or variable coefficients. Copyright © 2017 John Wiley & Sons, Ltd.     

Locally s‐distance transitive graphs

We give a unified approach to analyzing, for each positive integer s, a class of finite connected graphs that contains all the distance transitive graphs as well as the locally s‐arc transitive graphs of diameter at least s. A graph is in the class if it is connected and if, for each vertex v, the subgroup of automorphisms fixing v acts transitively on the set of vertices at distance i from v, for each i from 1 to s. We prove that this class is closed under forming normal quotients. Several graphs in the class are designated as degenerate, and a nondegenerate graph in the class is called basic if all its nontrivial normal quotients are degenerate. We prove that, for s≥2, a nondegenerate, nonbasic graph in the class is either a complete multipartite graph or a normal cover of a basic graph. We prove further that, apart from the complete bipartite graphs, each basic graph admits a faithful quasiprimitive action on each of its (1 or 2) vertex‐orbits or a biquasiprimitive action. These results invite detailed additional analysis of the basic graphs using the theory of quasiprimitive permutation groups. © 2011 Wiley Periodicals, Inc. J Graph Theory 69:176‐197, 2012     

On the convergence of basic iterative methods for convection–diffusion equations

In this paper we analyze convergence of basic iterative Jacobi and Gauss–Seidel type methods for solving linear systems which result from finite element or finite volume discretization of convection–diffusion equations on unstructured meshes. In general the resulting stiffness matrices are neither M‐matrices nor satisfy a diagonal dominance criterion. We introduce two newmatrix classes and analyse the convergence of the Jacobi and Gauss–Seidel methods for matrices from these classes. A new convergence result for the Jacobi method is proved and negative results for the Gauss–Seidel method are obtained. For a few well‐known discretization methods it is shown that the resulting stiffness matrices fall into the new matrix classes. Copyright © 1999 John Wiley & Sons, Ltd.     

Two‐step Newton‐type methods for solving inverse eigenvalue problems

In this paper, we apply the two‐step Newton method to solve inverse eigenvalue problems, including exact Newton, Newton‐like, and inexact Newton‐like versions. Our results show that both two‐step Newton and two‐step Newton‐like methods converge cubically, and the two‐step inexact Newton‐like method is super quadratically convergent. Numerical implementations demonstrate the effectiveness of new algorithms.     

Operator trigonometry of iterative methods

A new and general approach to the understanding and analysis of widely used iterative methods for the numerical solution of the equation Ax = b is presented. This class of algorithms, which includes CGN, GMRES. ORTHOMIN, BCG, CGS, and others of current importance, utilizes residual norm minimizing procedures, such as those often found under the general names Galerkin method, Arnoldi method, Lanczos method, and so on. The view here is different: the needed error minimizations are seen trigonometrically. © 1997 John Wiley & Sons, Ltd.     

High‐order split‐step theta methods for non‐autonomous stochastic differential equations with non‐globally Lipschitz continuous coefficients

In this paper, we first propose the so‐called improved split‐step theta methods for non‐autonomous stochastic differential equations driven by non‐commutative noise. Then, we prove that the improved split‐step theta method is convergent with strong order of one for stochastic differential equations with the drift coefficient satisfying a superlinearly growing condition and a one‐sided Lipschitz continuous condition. Finally, the obtained results are verified by numerical experiments. Copyright © 2016 John Wiley & Sons, Ltd.     

On the convergence of general stationary iterative methods for range‐Hermitian singular linear systems

General stationary iterative methods with a singular matrix M for solving range‐Hermitian singular linear systems are presented, some convergence conditions and the representation of the solution are also given. It can be verified that the general Ortega–Plemmons theorem and Keller theorem for the singular matrix M still hold. Furthermore, the singular matrix M can act as a good preconditioner for solving range‐Hermitian linear systems. Numerical results have demonstrated the effectiveness of the general stationary iterations and the singular preconditioner M. Copyright © 2009 John Wiley & Sons, Ltd.     

Recent computational developments in Krylov subspace methods for linear systems

Many advances in the development of Krylov subspace methods for the iterative solution of linear systems during the last decade and a half are reviewed. These new developments include different versions of restarted, augmented, deflated, flexible, nested, and inexact methods. Also reviewed are methods specifically tailored to systems with special properties such as special forms of symmetry and those depending on one or more parameters. Copyright © 2006 John Wiley & Sons, Ltd.     

Acceleration procedures for matrix iterative methods

In this paper, several procedures for accelerating the convergence of an iterative method for solving a system of linear equations are proposed. They are based on projections and are closely related to the corresponding iterative projection methods for linear systems.     

Third-order iterative methods under Kantorovich conditions

We study a class of third-order iterative methods for nonlinear equations on Banach spaces. A characterization of the convergence under Kantorovich type conditions and optimal estimates of the error are found. Though, in general, these methods are not very extended due to their computational costs, we will show some examples in which they are competitive and even cheaper than other simpler methods. We center our analysis in both, analytic and computational, aspects.     

On t‐designs and s‐resolvable t‐designs from hyperovals

Hyperovals in projective planes turn out to have a link with t‐designs. Motivated by an unpublished work of Lonz and Vanstone, we present a construction for t‐designs and s‐resolvable t‐designs from hyperovals in projective planes of order $2^n$. We prove that the construction works for $t\le 5$. In particular, for $t=5$ the construction yields a family of 5‐ $\left(2^n+2,8,70\left(2^{n?2}?1\right)\right)$ designs. For $t=4$ numerous infinite families of 4‐designs on $2^n+2$ points with block size $2k$ can be constructed for any $k\ge 4$. The construction assumes the existence of a 4‐ $\left(2^{n?1}+1,k,\lambda \right)$ design, called the indexing design, including the complete 4‐ $\left(2^{n?1}+1,k,\left(\genfrac{}{}{0ex}{}{2^{n?1}?3}{k?4}\right)\right)$ design. Moreover, we prove that if the indexing design is s‐resolvable, then so is the constructed design. As a result, many of the constructed designs are s‐resolvable for $s=2,3$. We include a short discussion on the simplicity or non‐simplicity of the designs from hyperovals.     

Consistency of iterative operator‐splitting methods: Theory and applications

In this article, we describe a different operator‐splitting method for decoupling complex equations with multidimensional and multiphysical processes for applications for porous media and phase‐transitions. We introduce different operator‐splitting methods with respect to their usability and applicability in computer codes. The error‐analysis for the iterative operator‐splitting methods is discussed. Numerical examples are presented. © 2009 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2010     

Minimum time‐step criteria for the Galerkin finite element methods applied to one‐dimensional parabolic partial differential equations

The finite element method has been well established for numerically solving parabolic partial differential equations (PDEs). Also it is well known that a too large time step should not be chosen in order to obtain a stable and accurate numerical solution. In this article, accuracy analysis shows that a too small time step should not be chosen either for some time‐stepping schemes. Otherwise, the accuracy of the numerical solution cannot be improved or can even be worsened in some cases. Furthermore, the so‐called minimum time step criteria are established for the Crank‐Nicolson scheme, the Galerkin‐time scheme, and the backward‐difference scheme used in the temporal discretization. For the forward‐difference scheme, no minimum time step exists as far as the accuracy is concerned. In the accuracy analysis, no specific initial and boundary conditions are invoked so that such established criteria can be applied to the parabolic PDEs subject to any initial and boundary conditions. These minimum time step criteria are verified in a series of numerical experiments for a one‐dimensional transient field problem with a known analytical solution. The minimum time step criteria developed in this study are useful for choosing appropriate time steps in numerical simulations of practical engineering problems. © 2005 Wiley Periodicals, Inc. Numer Methods Partial Differential Eq, 2006     

Monotone iterative methods for impulsive hyperbolic differential-functional equations

Theorems on impulsive hyperbolic differential-functional inequalities are considered. Comparison results and a uniqueness criterion are obtained. A method of approximation of the solutions of impulsive hyperbolic differential-functional equations by means of solutions of the associated linear problems is established. The difference between the exact and the approximate solutions is estimated.     

Efficient Krylov subspace methods for uncertainty quantification in large Bayesian linear inverse problems

Uncertainty quantification for linear inverse problems remains a challenging task, especially for problems with a very large number of unknown parameters (e.g., dynamic inverse problems) and for problems where computation of the square root and inverse of the prior covariance matrix are not feasible. This work exploits Krylov subspace methods to develop and analyze new techniques for large‐scale uncertainty quantification in inverse problems. In this work, we assume that generalized Golub‐Kahan‐based methods have been used to compute an estimate of the solution, and we describe efficient methods to explore the posterior distribution. In particular, we use the generalized Golub‐Kahan bidiagonalization to derive an approximation of the posterior covariance matrix, and we provide theoretical results that quantify the accuracy of the approximate posterior covariance matrix and of the resulting posterior distribution. Then, we describe efficient methods that use the approximation to compute measures of uncertainty, including the Kullback‐Liebler divergence. We present two methods that use the preconditioned Lanczos algorithm to efficiently generate samples from the posterior distribution. Numerical examples from dynamic photoacoustic tomography demonstrate the effectiveness of the described approaches.     

Superattracting cycles for some Newton type iterative methods

The paper is devoted to the analysis of certain dynamical properties of a family of iterative Newton type methods used to find roots of non-linear equations. We present a procedure for constructing polynomials in such a way that superattracting cycles of any prescribed length occur when these iterative methods are applied. This paper completes the study begun in Amat, Bermúclez, Busquier, et al., (2009).