Efficient vectorizable PDE solvers

The FDEM program package is a black-box solver for nonlinear systems of elliptic and parabolic PDEs. A FDM of arbitrary consistency order q on an unstructured basic FEM mesh is used. With the difference formula of order q+2 an error estimate is computed. A sophisticated algorithm has been developed to generate reliable difference formulae of order q on an unstructured grid. Completely new is the usage of the error estimate to determine under such general conditions for each node i its individual optimal order q(i) out of the orders 2, 4, 6. This order control tells us that higher order may be better or may not be better, depending on the character of the solution and the used mesh.

Two variants of modified incomplete block-matrix factorization with additive correction are proposed for the iterative solution of large linear systems of equations. Both rigorous theoretical support and numerical evidence are given displaying their efficiency when applied to discrete second order partial differential equations (PDEs), even in the case of quasi-singular problems.

Finite-Difference Frequency-Domain methods (FDFD) require solution of large linear systems of equations. These large systems are represented by matrix equations including highly sparse coefficient matrices, and they can often only be solved by using iterative methods. This paper presents an algorithm in which the matrix-equation solution approach in an iterative method is replaced by a multi-step solution process. Instead of using a coefficient matrix, the coefficients in the FDFD formulations are kept as three-dimensional arrays, and they are treated as operators. The algorithm is used together with the Bi-Conjugate Gradients Stabilized (BICGSTAB) method. This is applied to a three-dimensional FDFD method to solve for scattering from dielectric objects. It is also applied to two other FDFD methods (a single-grid and a double-grid FDFD) to solve for scattering from chiral objects. It has been shown that the presented algorithm effectively reduces the solution time and memory requirements.

The randomized Kaczmarz (RK) method is an iterative method for approximating the least-squares solution of large linear systems of equations. The standard RK method uses sequential updates, making parallel computation difficult. Here, we study a parallel version of RK where a weighted average of independent updates is used. We analyze the convergence of RK with averaging and demonstrate its performance empirically. We show that as the number of threads increases, the rate of convergence improves and the convergence horizon for inconsistent systems decreases.

Electron-impact excitation and ionization of helium is studied in the $S$-wave model. The problem is treated in full dimensionality using a time-dependent formulation of the exterior complex scaling method that does not involve the solution of large linear systems of equations. We discuss the steps that must be taken to compute stable ionization amplitudes. We present total excitation, total ionization, and single differential cross sections from the ground and $n=2$ excited states and compare our results with those obtained by others using a frozen-core model.

This paper describes and tests a parallel implementation of a factorized approximate inverse preconditioner (FSAI) to accelerate iterative linear system solvers. Such a preconditioner reveals an efficient accelerator of both Conjugate gradient and BiCGstab iterative methods in the parallel solution of large linear systems arising from the discretization of the advection-diffusion equation. The resulting message passing code allows the solution of large problems leading to a very cost-effective algorithm for the solution of large and sparse linear systems.

Computer Solution of Large Linear Systems

This article tries to use of cubic B-spline collocation method in the form of a difference method to solving system of one dimension nonlinear parabolic partial differential equations. This method is characterized by the simplicity of the finite difference method through discussion of various parabolic pde problems, but does not have complex computation of traditional cubic B-spline collocation method. The use of this method leads to a system of algebraic equations which is suitable for computer programming. The stability and convergence of this method is discussed. The obtained numerical solutions indicate that the method is reliable and yields results compatible with the exact solutions. The numerical approximate solutions to the nonlinear parabolic partial differential equations have been computed without transforming the equation and without using the linearization.

A new randomized vector algorithm for iterative solution of large linear systems

We prove existence of solutions of a new free boundary problem described by a system of degenerate parabolic equations. The problem arises in petroleum engineering and concerns fluid flows in diatomite rocks. The unknown functions represent the pressure of the fluid and a damage parameter of the porous rock. These quantities are not necessarily continuous on the free boundary, which considerably complicates the mathematical analysis.

ABSTRACTFour-dimensional variational (4D-Var) data assimilation for operational systems requires the solution of large linear systems that are poorly conditioned in general. In addition to efficient iterative solvers for linear systems, using a good preconditioner is required to guarantee an acceptable solution with a small number of iterations. We consider the assimilation of ocean observations using a weak constraints 4D-Var for the Navy Coastal Ocean Model (NCOM), based on the representer method. Two methods of preconditioning the linear system are implemented, namely the scaling of the linear system by the square root of the observations error variances, and the approximating stabilised representer matrix based on the computation of some representer functions. We evaluate their convergence using criteria such as the norm of the residuals and of the gradient of the cost function, the analysis error evaluated at the observation locations, and finally the convergence of the sequence of analyses. Results from all criteria show that the fastest convergence is achieved with the rescaling preconditioner when the conjugate gradient used to solve the linear system is equipped with a suitable inner product instead of the Euclidean.

FIDISOL is a program package for the solution of nonlinear systems of two-dimensional and three-dimensional elliptic and parabolic partial differential equations (PDEs) with nonlinear boundary conditions (BCs) on the boundaries of a rectangular domain. A finite difference method (FDM) with an arbitrary grid and arbitrary consistency order is used, these are either prescribed by the user or are self-adapted for a given relative tolerance. FIDISOL has been designed to be fully vectorizable on vector computers. In this paper we discuss several problems from the viewpoint of software development and user interface, for example, how to deliver the PDEs and BCs to FIDISOL and how to allow a flexible use by a suitable parameter list.

Accelerated generalized successive overrelaxation method for least squares problems

On the application of orthogonal polynomials to the iterative solution of linear systems of equations with indefinite or non-Hermitian matrices

The alternating group explicit (AGE) method [1] has been applied to a variety of problems involving parabolic, hyperbolic and elliptic partial differential equations. The method has been shown to be extremely powerful and flexible and to be very suitable for use on vector and parallel computers. The practical implementation of it is considered here. The new scheme is shown to possess the same convergence property with a reduced computational effort.