An adaptive Factored Sparse Approximate Inverse preconditioner for non-symmetric linear systems
This research proposal focuses on the development and application of a non-symmetric variant of FSAI (NSY-FSAI) for unsymmetric matrices. In particular, the major issue to address is the design of an autoadaptive algorithm able to build an optimal preconditioner, in terms of non-zero pattern and entries able to run on parallel computers. The developed algorithm will be tested in real world applications involving subsurface flow and geomechanical problems which are known to be challenging problems characterized by high heterogeneity and large scales. It is expected that the new preconditioner will allow for a significant reduction of the computational times usually required by the simulations and, at the same time, will foster the study of more complex problems.
This job comes from a partnership with Science Magazine and