Preconditioning Methods for Local Discontinuous Galerkin Discretizations
Guido Kanschat
Abstract
Guido Kanschat
Abstract
A multilevel interior penalty method is used as an efficient preconditioner for the Schur complement of the local discontinuous Galerkin (LDG) discretization of a Poisson problem. The method is then used in a block-triangular preconditioner of the LDG saddle point system. The block preconditioner is of the same efficiency as the Schur complement version. Finally, the block preconditioner is extended to the discretization of the Stokes problem by the LDG method. Again, the preconditioned saddle point problem can be solved in about as many steps as the Schur complement. The influence of several parameters on the performance of these methods is investigated.
OpenAlex reports 32 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
A multilevel interior penalty method is used as an efficient preconditioner for the Schur complement of the local discontinuous Galerkin (LDG) discretization of a Poisson problem. The method is then used in a block-triangular preconditioner of the LDG saddle point system. The block preconditioner is of the same efficiency as the Schur complement version. Finally, the block preconditioner is extended to the discretization of the Stokes problem by the LDG method. Again, the preconditioned saddle point problem can be solved in about as many steps as the Schur complement. The influence of several parameters on the performance of these methods is investigated.
Key concepts: Preconditioner, Schur complement, Mathematics, Saddle point, Discontinuous Galerkin method, Discretization, Block (permutation group theory), Applied mathematics