Coarse space orthogonalization for indefinite linear systems of equations arising in geometrically nonlinear elasticity (Q1294597)

The authors consider the problem of finding solutions of indefinite linear systems arising in the course of Newton's iterations for problems of geometrically nonlinear elasticity by the Krylov subspace method. Often, the convergence of these methods slows down considerably due to ill-conditioning or even indefiniteness of the matrix near the simple limit or bifurcation points. It is found that orthogonalization against a low-dimensional coarse finite element space makes convergence satisfactory in all parameter ranges. This approach is combined with the usual hierarchical and multilevel preconditions and implementation issues in some details. Finally, some numerical results are presented.