关于变分伴随的扩散方程系数反问题的数值反 演算法.pdfVIP

Advances in Applied Mathematics 应用数学进展, 2018, 7(4), 466-475 Published Online April 2018 in Hans. /journal/aam /10.12677/aam.2018.74057 An Inversion Algorithm for Inverse Coefficient Problems in the Diffusion Equation Based on Variational Adjoint Method Zhendong Wang, Gongsheng Li, Yingmei Wang School of Mathematics and Statistics, Shandong University of Technology, Zibo Shandong th rd th Received: Apr. 12 , 2018; accepted: Apr. 23 , 2018; published: Apr. 30 , 2018 Abstract A numerical inversion algorithm is proposed based on the variational adjoint method for solving the backward problem and inverse source problem in the diffusion equation. With the help of the adjoint problem, a variational identity connecting the known data with the unknown is derived with which a linear system for the unknown is set forth by suitably choosing the solution of the adjoint equation. Numerical solutions to the inverse problems are obtained by solving the linear system with Tikhonov regularization, and numerical examples are presented to demonstrate the effectiveness of the inversion algorithm. Keywords Diffusion Equation, Inverse Problem, Adjoint Method, Variational Identity, Numerical Inversion 基于变分伴随的扩散方程系数反问题的数值反 演算法 王桢东，李功胜，王迎美 山东理工大学数学与统计学院，山东 淄博 收稿日期：2018年4月12 日；录用日期：2018年4月23 日；发布日期：2018年4月30 日 文章引用: 王桢东, 李功胜, 王迎美. 基于变分伴随的扩散方程系数反问题的数值反演算法[J]. 应用数学进展, 2018, 7(4): 466-475. DOI: 10.12677/aam.2018.74057 王桢东 等 摘 要 对于扩散方程的逆时问题与确定时间–空间依赖源项的反问题，提出一种基于变分伴随方法的数值反演 算法。借助正问题的伴随问题,构建联系已知数据与未知量的变分恒等式，通过适当选择伴随方程的解进 而得到关于未知量的线性方程组，结合正则化方法，获得了反问题的数值解。几个算例结果表明了这种 方法的有效性。 关键词 扩散