We consider two-level overlapping domain decomposition
preconditioners for piecewise linear finite element
approximations of elliptic PDEs with highly variable
coefficients, as they arise in practice, for example,
in the computation of flows in
heterogeneous porous media, in both the deterministic and
(Monte-Carlo simulated) stochastic
cases. We propose and analyse new robust coarsening strategies
for these problems, e.g. based on smoothed aggregation and
multiscale finite elements, and study different ways of combining
the coarse solve with the local solves (i.e. additive,
multiplicative and deflation). We will use and extend
a recently developed general theoretical framework for analysing
the resulting preconditioners to show their robustness for a variety
of model situations. In particular, our estimates prove very
precisely the previously made empirical observation
that the use of low-energy coarse spaces can
lead to robust preconditioners.
Numerical experiments will be used to illustrate the robustness and
efficiency of the proposed methods and the sharpness of our
theoretical results.
This is joint work with I.G. Graham and J. Van lent
(University of Bath) and E. Vainikko (University of Tartu, Estonia).