Title :
High precision differentiation of FEM approximate solutions
Author_Institution :
Dept. of Electr. & Comput. Eng., McMaster Univ., Hamilton, Ont., Canada
Abstract :
This paper presents the high precision differentiation method based on Green´s second identity. The technique is compared to several recent methods based on local smoothing and superconvergent patch recovery (SPR). The methodology is extended to 3D problems described by scalar Poisson equation, using the sphere as a base domain for extraction of derivatives. Analytic verification and error sensitivity analysis is performed. The alternative approach employing fundamental solutions to the Dirichlet problem in place of Green´s functions is also outlined. The technique is suited to postprocessing of finite element solutions, or may be applied to other numerical approximate solutions.
Keywords :
Green´s function methods; boundary-value problems; differentiation; error analysis; finite element analysis; 3D problems; Dirichlet problem; FEM approximate solutions; Green´s functions; Green´s second identity; analytic verification; error sensitivity analysis; finite element solutions; fundamental solutions; high precision differentiation; local smoothing; numerical approximate solutions; postprocessing; scalar Poisson equation; superconvergent patch recovery; Design automation; Displays; Error analysis; Finite element methods; Graphics; Integral equations; Performance analysis; Poisson equations; Sensitivity analysis; Smoothing methods;
Conference_Titel :
Antennas and Propagation Society International Symposium, 1997. IEEE., 1997 Digest
Conference_Location :
Montreal, Quebec, Canada
Print_ISBN :
0-7803-4178-3
DOI :
10.1109/APS.1997.631811