On the accuracy of finite element approximations to a class of interface problems

On the accuracy of finite element approximations to a class of interface problems
复制标题

DOI:
10.1090/mcom3051
复制
发表时间:
2015-11
期刊:
Math. Comput.
影响因子:
--
通讯作者:
J. Guzmán;M. Sánchez;M. Sarkis
J. Guzmán;M. Sánchez;M. Sarkis
中科院分区:
其他
文献类型:
--
作者:
J. Guzmán;M. Sánchez;M. Sarkis

文献摘要

被引文献

相似文献

定义跳跃为[∇u·n] =∇u·n +∇u·n,其中u = u|Ω±,n为向外垂直于Ω的单位(见图1)。同时,我们表示[u] = u - u。对于问题(1.1)已经开发了许多数值方法。其中最值得注意的可能是Peskin[11]的有限差分法(即浸入边界法)和LeVeque和Li[11]的方法(即浸入界面法;另见Mayo的方法[14,15,16])。LeVeque和Li[11]方法用于更一般的具有不连续扩散系数的问题,而Peskin[18]方法用于具有浸入边界的流体流动问题。虽然Peskin[18]的方法是用包含浸入边界弹性力Γ的力函数F来表示的,但[19]表明,它可以重新表述为一个界面问题(α = 0),其中β编码弹性力。自这两篇重要论文[18,11]以来,有许多文章扩展或改进了这些方法。特别是,这些方法的有限元版本已经出现;参见[3,9,6,2]。对于上述问题(α = 0),众所周知,Peskin[18]方法只有一阶精度,而LeVeque和Li[11]方法是二阶精度
The jump is defined as [∇u · n] = ∇u · n + ∇u ·n where u = u|Ω± and n is the unit outward pointing normal to Ω (see figure 1). Also, we denote [u] = u − u. Many numerical methods have been developed for problem (1.1). Perhaps the most notable ones are the finite difference method of Peskin [18] (i.e., immersed boundary method) and the method of LeVeque and Li [11] (i.e., the immersed interface method ; see also the method of Mayo [14, 15, 16]) .The method of LeVeque and Li [11] was developed for the more general problem with discontinuous diffusion coefficients, while the method of Peskin [18] was developed for fluid flow problems with an immersed boundary. Although the method of Peskin [18] is formulated with a force function F that incorporates the elastic force of the immersed boundary Γ, it was shown in [19] that it can be re-formulated as an interface problem (with α = 0) where β encodes the elastic force. Since the two important papers [18, 11] there have been many articles extending or improving these methods. In particular, finite element versions of these methods have appeared; see for example [3, 9, 6, 2]. For the above problem (α = 0), it is well known that the method of Peskin [18] is only first-order accurate whereas the method of LeVeque and Li [11] is second-order