Comparison of iterative and direct solvers for 3D CSEM modeling

Electromagnetic (EM) modeling in frequency domain requires solving large linear systems of equations. This can be accomplished using direct or iterative techniques. We compare the performance of the direct MUMPS and an iterative QMR solver for controlled-source EM modeling on distributed computing platforms. We first establish a conductivity value for air that lets the iterative solver converge while maintaining accuracy of subsurface EM fields, and then demonstrate the accuracy of the direct and iterative solutions. We consider the influence of model size, the number of right hand sides, parallelization approaches, and hardware architecture on performance. Clearly, iterative solutions are significantly more efficient than direct ones for small numbers of right hand sides. Nevertheless, we find that for simulations of moderatesized models on a modern, moderate-sized cluster, the point at which iterative solver runtime exceeds that of the direct solver is reached for relatively small numbers of right hand sides.

[1]  Jean Virieux,et al.  3-D Parallel Frequency-domain Visco-acoustic Wave Modelling based on a Hybrid Direct/Iterative Solver , 2011 .

[2]  Uri M. Ascher,et al.  Fast Finite Volume Simulation of 3D Electromagnetic Problems with Highly Discontinuous Coefficients , 2000, SIAM J. Sci. Comput..

[3]  M. Zhdanov,et al.  Integral equation method for 3D modeling of electromagnetic fields in complex structures with inhomogeneous background conductivity , 2006 .

[4]  M. Benzi.  Preconditioning techniques for large linear systems: a survey , 2002 .

[5]  Azzam Haidar,et al.  Three‐dimensional parallel frequency‐domain visco‐acoustic wave modelling based on a hybrid direct/iterative solver , 2011 .

[6]  Patrick Amestoy,et al.  Hybrid scheduling for the parallel solution of linear systems , 2006, Parallel Comput..

[7]  J. T. Smith.  Conservative modeling of 3-D electromagnetic fields, Part II: Biconjugate gradient solution and an accelerator , 1996 .

[8]  Victor M. Calo,et al.  A survey on direct solvers for Galerkin methods , 2012 .

[9]  Jianlin Xia,et al.  On 3D modeling of seismic wave propagation via a structured parallel multifrontal direct Helmholtz solver , 2011 .

[10]  W. A. Mulder,et al.  A multigrid solver for 3D electromagnetic diffusion , 2006 .

[11]  R. Hiptmair.  Multigrid Method for Maxwell's Equations , 1998 .

[12]  Christoph Schwarzbach,et al.  Stability of finite element solutions to Maxwell's equations in frequency domain , 2009 .

[13]  Rita Streich,et al.  3D finite-difference frequency-domain modeling of controlled-source electromagnetic data: Direct solution and optimization for high accuracy , 2009 .

[14]  Nicholas I. M. Gould,et al.  A numerical evaluation of sparse direct solvers for the solution of large sparse symmetric linear systems of equations , 2007, TOMS.

[15]  S. Operto,et al.  3D finite-difference frequency-domain modeling of visco-acoustic wave propagation using a massively parallel direct solver: A feasibility study , 2007 .

[16]  Christoph Schwarzbach,et al.  Three-dimensional adaptive higher order finite element simulation for geo-electromagnetics—a marine CSEM example , 2011 .

[17]  Chester J. Weiss,et al.  Mapping thin resistors and hydrocarbons with marine EM methods, Part II -Modeling and analysis in 3D , 2006 .

[18]  G. Newman,et al.  Frequency‐domain modelling of airborne electromagnetic responses using staggered finite differences , 1995 .

[19]  Michael Commer,et al.  Three-dimensional controlled-source electromagnetic and magnetotelluric joint inversion , 2009 .

[20]  Douglas W. Oldenburg,et al.  Forward Modelling And Inversion of Multi-Source TEM Data , 2008 .