There are different types of the shape derivatives for the potential energies in Poisson equations with mixed boundary value problems, that is, derivatives with respect to the small change of boundaries and of the joint points of different boundary conditions. The systematic treatment in the shape derivatives is done using Generalized J-integral method proposed by author. In the shape derivative with respect to the joint points, the movements of singularities appear. This means that the numerical calculation is not easy in the shape derivatives with mixed boundary condition. It is shown that Generalized J-integral method is useful for the shape derivatives in theoretical and numerical studies of mixed boundary value problems.
We discuss the Fourier-Infinite-Element method, which combines the Fourier method and the infinite element method, for sloving the Poisson equation with singularities caused by the presence of a corner. In the infinite element method the underlying domain is divided into infinitely many pieces. This leads to a system of the infinitely many equations in which the matrix is of block tridiagonal form. In spite of this fact, it is the nature of these block matrices that allows the problem to be expressed by three-term inhomogeneous recurrence relations. The required solution of this problem is generated by using an algorithm based on Gaussian elimination and rapidly obtained by using fast sine transform. In this paper we also discuss the approach for estimating automatically the truncation error of the proposed algorithm.
Euler transformation is a linear sequence transformation to accelerate the convergence of alternating series. The series of weights of the transformation is extended to a continuous function which can accelerate Fourier type integrals with slowly convergent integrands. We show that the continuous function can be used to compute Fourier transforms of slowly decaying functions using FFT.
Generally, it is very difficult for a software manager to allocate optimally the financial budget to a software development project during the testing phase. Several optimal testing-effort allocation problems for the module testing phase have been discussed to minimize the total testing-effort expenditures, the total residual fault content in the system, the expected total software cost and so on. Most of them have not considered the variance of quality of each module after the testing. In this paper, two optimal testing-effort allocation problems are investigated to smooth the number of software faults for the system composed of some software modules after the module testing. We also provide numerical examples of the derived optimal policies.
We consider numerical verification methods to obtain the maximum absolute value of generalized eigenvalue problems. We present four kinds of methods and compare the performance in various situations as well as give evaluation of the advantage and disadvantage of these methods. All numerical results have been calculated by the interval arithmetic software for considering the rounding error occuring in the calculation. Finally, we will present an application to an eigenvalue problem appeared in some a priori error estimates for the finite element solution of the Stokes equations.