带齐次混合边界特征值问题的一种基于多尺度离散的有限元自适应算法-计算数学专业论文.docx

PAGE PAGE 11 PAGE PAGE 22 Introduction Eigenvalue problems have important physical background and wide applications. Specially, they are found in the study of vibration problem (see [1]). Therefore, finite element method for solving eigenvalue problems has attracted the attention of relevant researchers (see [2, 5, 6, 10, 12, 18]). A posteriori error estimates and adaptive methods of finite element for eigenvalue problems have been discussed(see, e.g., [20, 25, 26, 28]). Basically there are the following three ways which are more or less the same but dif- ferent in the problem solved in each iteration in the a posterior error estimator: (1) Solve the original eigenvalue problem ?(?? , ?) = ?? ?(?? , ?). The convergence and optimality of this adaptive procedure is obtained in [6]. (2) Inverse iteration type (with or without correction). The convergence has been studied in the [5, 7, 8]. (3) Rayleigh quotient iteration type (see [11, 12]). In addition, the authors in [16] discussed residual type a posteriori error estimates of finite element approximation for mixed boundary value problems. Based on the work mentioned above, adaptive finite element methods are applied to solve eigenvalue problems with homogeneous mixed boundary conditions. The purpose of this paper is to establish adaptive finite element methods for the problems by using multi- scale discretization schemes based on Rayleigh quotient iteration. In order to achieve this objective, the key is to establish a posteriori error estimate of the multi-scale discretiza- tion schemes. Based on the work of [15, 16], a posteriori error estimate of algorithms based on shifted-inverse power method is deduced. Let (??? , ??? ) be an approximate eigenpair obtained by algorithms, we prove that the reliable and efficient a posteriori error indicator ?for ??? ) is obtained by replacing ?? ? in the residual type a posteriori error indicator, where ????? is the solution of the boundary value problem with the right h