컨텐츠 시작

학술대회/행사

초록검색

제출번호(No.) 0589
분류(Section) Invited Talk
분과(Session) Numerical Analysis / Scientific Computations / Mathematics in Science and Technology (SS-14)
영문제목
(Title(Eng.))
On the asymptotic exactness of the error estimator for mixed finite element methods
저자(Author(s))
Kwang-Yeon Kim1
Kangwon National University1
초록본문(Abstract) Mixed finite element methods have been one of the popular choices for numerical approximation of
second-order elliptic equations because they fulfill local mass conservation and
allow accurate computation of the vector variable.
One of their drawbacks is the relatively large size of the discrete algebraic systems
compared with the primal methods of the same order,
which makes it inevitable to apply adaptive mesh refinement based on a posteriori error estimators
in order to keep as small number of degrees of freedom as possible in achieving a desired accuracy.
Some a posteriori error estimators often exhibit amazing accuracy
as the ratio of the estimated error to the actual error tends to unity when the mesh size goes to zero.
In such cases we say that the error estimator is asymptotically exact.
The purpose of this talk is to discuss the asymptotic exactness of the error estimator for the mixed finite element method
proposed in [Alonso, Error estimators for a mixed method, Numer. Math. 74 (1996), 385--395].
The error estimator is based on solution of local problems which are similar to those of Bank and Weiser
designed for the $P1$ conforming finite element method.
It was established that the error estimator of Bank and Weiser is asymptotically exact on uniform meshes
if the exact solution is $H^3$-regular,
where the superconvergence property of the standard nodal interpolant is crucially exploited.
By using a similar argument and the superconvergence property of the Fortin projection
for the Raviart--Thomas element, we show that the error estimator of Alonso is asymptotically exact on uniform meshes
if the vector solution is $H^2$-regular.
We also carry out some numerical experiments to validate the theoretical results.
분류기호
(MSC number(s))
65N30, 65N15
키워드(Keyword(s)) a posteriori error estimator, asymptotic exactness, superconvergence, mixed finite element method
강연 형태
(Language of Session (Talk))
English