Optimal superconvergent one step quadratic spline collocation methods

We formulate new optimal quadratic spline collocation methods for the solution of various elliptic boundary value problems in the unit square. These methods are constructed so that the collocation equations can be solved using a matrix decomposition algorithm. The results of numerical experiments exhibit the expected optimal global accuracy as well as superconvergence phenomena. AMS subject classification (2000)  65N35, 65N22     

A Lyapunov function for a two-chemical species version of the chemotaxis model

Since the accuracy of finite element solutions of partial differential equations is generally mesh dependent, especially when solutions have singularities and discontinuities, a proper mesh generation is often important and sometimes crucial for an accurate numerical approximation of such problems. In this paper, the mesh transformation method is applied to the boundary value problems of elliptic partial differential equations, and it is proved that the method leads to the optimal finite element solutions. AMS subject classification (2000) 73C50, 65K10, 65N12, 65N30     

On the stability of meshless symmetric collocation for boundary value problems

In this paper, we study the stability of symmetric collocation methods for boundary value problems using certain positive definite kernels. We derive lower bounds on the smallest eigenvalue of the associated collocation matrix in terms of the separation distance. Comparing these bounds to the well-known error estimates shows that another trade-off appears, which is significantly worse than the one known from classical interpolation. Finally, we show how this new trade-off can be overcome as well as how the collocation matrix can be stabilized by smoothing. AMS subject classification (2000)  65N12, 65N15, 65N35     

An averaging scheme for macroscopic numerical simulation of nonconvex minimization problems

Averaging or gradient recovery techniques, which are a popular tool for improved convergence or superconvergence of finite element methods in elliptic partial differential equations, have not been recommended for nonconvex minimization problems as the energy minimization process enforces finer and finer oscillations and hence at the first glance, a smoothing step appears even counterproductive. For macroscopic quantities such as the stress field, however, this counterargument is no longer true. In fact, this paper advertises an averaging technique for a surprisingly improved convergence behavior for nonconvex minimization problems. Similar to a finite volume scheme, numerical experiments on a double-well benchmark example provide empirical evidence of superconvergence phenomena in macroscopic numerical simulations of oscillating microstructures. AMS subject classification (2000)  65K10,65N30     

Artificial time integration

Many recent algorithmic approaches involve the construction of a differential equation model for computational purposes, typically by introducing an artificial time variable. The actual computational model involves a discretization of the now time-dependent differential system, usually employing forward Euler. The resulting dynamics of such an algorithm is then a discrete dynamics, and it is expected to be “close enough” to the dynamics of the continuous system (which is typically easier to analyze) provided that small – hence many – time steps, or iterations, are taken. Indeed, recent papers in inverse problems and image processing routinely report results requiring thousands of iterations to converge. This makes one wonder if and how the computational modeling process can be improved to better reflect the actual properties sought. In this article we elaborate on several problem instances that illustrate the above observations. Algorithms may often lend themselves to a dual interpretation, in terms of a simply discretized differential equation with artificial time and in terms of a simple optimization algorithm; such a dual interpretation can be advantageous. We show how a broader computational modeling approach may possibly lead to algorithms with improved efficiency. AMS subject classification (2000)  65L05, 65M32, 65N21, 65N22, 65D18     

Three-scale finite element eigenvalue discretizations

Some three-scale finite element discretization schemes are proposed and analyzed in this paper for a class of elliptic eigenvalue problems on tensor product domains. With these schemes, the solution of an eigenvalue problem on a fine grid may be reduced to the solutions of eigenvalue problems on a relatively coarse grid and some partially mesoscopic grids, together with the solutions of linear algebraic systems on a globally mesoscopic grid and several partially fine grids. It is shown theoretically and numerically that this type of discretization schemes not only significantly reduce the number of degrees of freedom but also produce very accurate approximations. AMS subject classification (2000)  65N15, 65N25, 65N30, 65N50     

A Multigrid Scheme for Elliptic Constrained Optimal Control Problems

A multigrid scheme for the solution of constrained optimal control problems discretized by finite differences is presented. This scheme is based on a new relaxation procedure that satisfies the given constraints pointwise on the computational grid. In applications, the cases of distributed and boundary control problems with box constraints are considered. The efficient and robust computational performance of the present multigrid scheme allows to investigate bang-bang control problems.AMS Subject Classification: 49J20, 65N06, 65N12, 65N55Supported in part by the SFB 03 “Optimization and Control”     

High order generalized upwind schemes and numerical solution of singular perturbation problems

High even order generalizations of the traditional upwind method are introduced to solve second order ODE-BVPs without recasting the problem as a first order system. Both theoretical analysis and numerical comparison with central difference schemes of the same order show that these new methods may avoid typical oscillations and achieve high accuracy. Singular perturbation problems are taken into account to emphasize the main features of the proposed methods. AMS subject classification (2000)  65L10, 65L12, 65L50     

Sinc-based computations of eigenvalues of Dirac systems

The celebrated classical sampling theorem is used to compute approximate values of the eigenvalues of Dirac systems with eigenvalue parameter in the boundary conditions. We deal with problems with an eigenparameter in one or two boundary conditions. The error analysis is established considering both truncation and amplitude errors associated with the sampling theorem. We indicate the role of the amplitude error as well as other parameters in the method via illustrative examples. AMS subject classification (2000)  34L16, 65L15, 94A20     

The penetration function and its application to microscale problems

The penetration function measures the effect of the boundary data on the energy of the solution of a second order linear elliptic PDE taken over an interior subdomain. Here the coefficients of the PDE are functions of position and often represent the material properties of non homogeneous media with microstructure. The penetration function is used to assess the accuracy of global-local approaches for recovering local solution features from coarse grained solutions such as those delivered by homogenization theory. AMS subject classification (2000)  65N15, 78M40     

Adaptive application of operators in standard representation

Recently adaptive wavelet methods have been developed which can be shown to exhibit an asymptotically optimal accuracy/work balance for a wide class of variational problems including classical elliptic boundary value problems, boundary integral equations as well as certain classes of noncoercive problems such as saddle point problems. A core ingredient of these schemes is the approximate application of the involved operators in standard wavelet representation. Optimal computational complexity could be shown under the assumption that the entries in properly compressed standard representations are known or computable in average at unit cost. In this paper we propose concrete computational strategies and show under which circumstances this assumption is justified in the context of elliptic boundary value problems. Dedicated to Charles A. Micchelli on the occasion of his 60th birthday Mathematics subject classifications (2000) 41A25, 41A46, 65F99, 65N12, 65N55. This work has been supported in part by the Deutsche Forschungsgemeinschaft SFB 401, the first and third author are supported in part by the European Community's Human Potential Programme under contract HPRN-CT-202-00286 (BREAKING COMPLEXITY). The second author acknowledges the financial support provided through the European Union's Human Potential Programme, under contract HPRN-CT-2002-00285 (HASSIP) and through DFG grant DA 360/4–1.     

The chebop system for automatic solution of differential equations

In Matlab, it would be good to be able to solve a linear differential equation by typing u = L\f, where f, u, and L are representations of the right-hand side, the solution, and the differential operator with boundary conditions. Similarly it would be good to be able to exponentiate an operator with expm(L) or determine eigenvalues and eigenfunctions with eigs(L). A system is described in which such calculations are indeed possible, at least in one space dimension, based on the previously developed chebfun system in object-oriented Matlab. The algorithms involved amount to spectral collocation methods on Chebyshev grids of automatically determined resolution. AMS subject classification (2000)  65L10, 65M70, 65N35     

Fourier Analysis of Multigrid for a Model Two-Dimensional Convection-Diffusion Equation

We present a Fourier analysis of multigrid for the two-dimensional discrete convection-diffusion equation. For constant coefficient problems with grid-aligned flow and semi-periodic boundary conditions, we show that the two-grid iteration matrix can be reduced via a set of orthogonal transformations to a matrix containing individual 4×4 blocks. This enables a trivial computation of the norm of the iteration matrix demonstrating rapid convergence in the case of both small and large mesh Peclet numbers, where the streamline-diffusion discretisation is used in the latter case. We also demonstrate that these results are strongly correlated with the properties of the iteration matrix arising from Dirichlet boundary conditions. AMS subject classification (2000) 65F10, 65N22, 65N30, 65N55     

Adaptive methods for piecewise polynomial collocation for ordinary differential equations

Various adaptive methods for the solution of ordinary differential boundary value problems using piecewise polynomial collocation are considered. Five different criteria are compared using both interval subdivision and mesh redistribution. The methods are all based on choosing sub-intervals so that the criterion values have (approximately) equal values in each sub-interval. In addition to the main comparison it is shown by example that at least when accuracy is low then equidistribution may not give a unique solution. The main results that using interval size times maximum residual as criterion gives very much better results than using maximum residual itself. It is also shown that a criterion based on a global error estimate while giving very good results in some cases, is unsatisfactory in other cases. The other criteria considered are that given by De Boor and the last Chebyshev series coefficient. AMS subject classification (2000)  65L10, 65L50, 65L60     

The Method of Fundamental Solutions: A Weighted Least-Squares Approach

We investigate the Method of Fundamental Solutions (MFS) for the solution of certain elliptic boundary value problems. In particular, we study the case in which the number of collocation points exceeds the number of singularities, which leads to an over-determined linear system. In such a case, the resulting linear system is over-determined and the proposed algorithm chooses the approximate solution for which the error, when restricted to the boundary, minimizes a suitably defined discrete Sobolev norm. This is equivalent to a weighted least-squares treatment of the resulting over-determined system. We prove convergence of the method in the case of the Laplace’s equation with Dirichlet boundary data in the disk. We develop an alternative way of implementing the numerical algorithm, which avoids the inherent ill-conditioning of the MFS matrices. Finally, we present numerical experiments suggesting that introduction of Sobolev weights improves the approximation. AMS subject classification (2000) 35E05, 35J25, 65N12, 65N15, 65N35, 65T50     

Convergence Rates for an Adaptive Dual Weighted Residual Finite Element Algorithm

Basic convergence rates are established for an adaptive algorithm based on the dual weighted residual error representation,

Bivariate least squares approximation with linear constraints

In this article linear least squares problems with linear equality constraints are considered, where the data points lie on the vertices of a rectangular grid. A fast and efficient computational method for the case when the linear equality constraints can be formulated in a tensor product form is presented. Using the solution of several univariate approximation problems the solution of the bivariate approximation problem can be derived easily. AMS subject classification (2000)  65D05, 65D07, 65D10, 65F05, 65F20     

Eigenvalue problems for fractional ordinary differential equations

The eigenvalue problems are considered for the fractional ordinary differential equations with different classes of boundary conditions including the Dirichlet, Neumann, Robin boundary conditions and the periodic boundary condition. The eigenvalues and eigenfunctions are characterized in terms of the Mittag–Leffler functions. The eigenvalues of several specified boundary value problems are calculated by using MATLAB subroutine for the Mittag–Leffler functions. When the order is taken as the value 2, our results degenerate to the classical ones of the second-ordered differential equations. When the order α satisfies 1 < α < 2 the eigenvalues can be finitely many.     

Linear least squares problems with data over incomplete grids

The paper addresses bivariate surface fitting problems, where data points lie on the vertices of a rectangular grid. Efficient and stable algorithms can be found in the literature to solve such problems. If data values are missing at some grid points, there exists a computational method for finding a least squares spline by fixing appropriate values for the missing data. We extended this technique to arbitrary least squares problems as well as to linear least squares problems with linear equality constraints. Numerical examples are given to show the effectiveness of the technique presented. AMS subject classification (2000)  65D05, 65D07, 65D10, 65F05, 65F20