Opportunity at Air Force Research Laboratory (AFRL)
State-of-the-Art, Massively Parallel Computational Electromagnetics
Directed Energy Directorate, RD/Engineering, Applied Sci, and Computer Science
||Kirtland Air Force Base, NM 871175776
|Greenwood, Andrew D.
Basic research in computational electromagnetics produces promising new techniques in terms of scaling and problems that can be solved. However, results for these techniques are often shown only for a few small test cases. Before the new methods can make an impact on production level simulation, more work is needed for validation and verification as well as to show that the techniques can be implemented fully in three dimensions and can scale to large numbers of processors on modern parallel architectures. Parallel scaling and algorithm implementation and performance on GPUs also needs further study. Areas of interest include, but are not necessarily limited to, fast matrix solution techniques (including robust preconditioners, both dense and sparse matrices as well as sparse-dense matrices from finite element-boundary integral techniques, and matrix compression techniques), finte-element time-domain analysis (including implicit time-domain techniques; efficient solution of mass matrices on massively parallel architectures; local terminating boundary conditions; higher order basis functions; complex material modeling; antenna feed network modeling; and the efficient, accurate addition of charged particles to the method), and finite element tearing and interconnect (FETI) techniques (including use of FETI to accelerate interprocessor communication, continuous and non-continuous mesh analysis for multiscale problems, and use of FETI for modeling finite periodic structures).
Computational electromagnetics; Parallel computing; Validation; Verification; Finite element; Fast solver; Time domain; Frequency domain;
Open to U.S. citizens
Open to Postdoctoral and Senior applicants