A stable Lagrange multiplier space for stiff interface conditions within the extended finite element method

Abstract

International audienceThis paper introduces a new algorithm to define a stable Lagrange multiplier space to impose stiff interface conditions within the context of the extended finite element method. In contrast to earlier approaches. we do not work with an interior penalty formulation as, e.g. for Nitsche techniques, but impose the constraints weakly in terms of Lagrange multipliers. Roughly speaking a stable and optimal discrete Lagrange multiplier space has to satisfy two criteria: a best approximation property and a uniform inf-sup condition. Owing to the fact that the interface does not match the edges of the mesh, the choice of a good discrete Lagrange Multiplier space is not trivial. Here we propose a new algorithm for the local construction of the Lagrange Multiplier space and show that a uniform inf-sup condition is satisfied. A counterexample is also presented, i.e. the inf-sup constant depends on the mesh-size and degenerates as it tends to zero. Numerical results in two-dimensional confirm the theoretical ones. Copyrigh

A high-order immersed boundary discontinuous-Galerkin method for Poisson's equation with discontinuous coefficients and singular sources

“…For a typical electro-static problem, however, it might occur that the voltage (=poten-tial) is not controlled, but rather the total electrical charge Q on a conductor is specified, and one must calculate the corresponding potential as a (scalar) unknown-the so-called floating potential problem. In this case we modify (15) (12) holds, and such that…”

“…Moreover, the use of Lagrange multipliers delivers a non-positive system and additional degrees of freedom are introduced. There are many studies that deal with these issues and propose solutions on how to choose L h [14][15][16]; further alternative formulations such as Nitsche's method or stabilized Lagrange multipliers, respectively, have also been advocated [17][18][19][20][21][22][23][24][25]. For our purpose of benchmark testing, the classic Lagrange multiplier approach works nicely as we can control L h a-priori.…”

A XFEM Lagrange Multiplier Technique for Stefan Problems

“…Normally, this secondary flux variable is defined purely on the interface and requires the primary (temperature) and secondary (heat flux) variables to respect the inf-sup condition Babuska (1969);Brezzi (1974). Otherwise, oscillations may appear in the solution near the interface Béchet et al (2009);Ji and Dolbow (2004); Moes et al (2006).…”

A modified perturbed Lagrangian formulation for contact problems

“…methods the Vital Vertex method has been defined to fulfill the InfSup condition in the case of a Dirichlet boundary in 2D [7] and 3D [2].…”

“…To our knowledge only the Vital Vertex method, first proposed by Bechet et al [7], can satisfy the compatibility between displacements and multipliers. This method has been used for imposing Dirichlet boundary conditions in 2D and 3D [2] immersed boundaries.…”