The
nowadays Grid technologies connect the nation's computers, databases,
instruments, storage devices to provide a virtually homogeneous problem-solving
environment from the user's point of view in many field of science ? including
biochemistry, drug-research and engineering. The architectural differences
between the single computers and the Grid resources infers the rise of the
distributed, heterogeneous and dynamic application classes. Because the Grid is
inherently more complex than existing computer systems, so Grid applications
reflect some of this complexity.
The forward
2.5D frequency domain electromagnetic (FEM) modeling ? taking a model and
calculating what the observed EM responses should be - a useful tool to
investigate different effects. In practical exploration these can be considered
as influencing factors or distortion effects. The end users of geophysical EM
measurements are interested in getting depth and extension data of the
structural elements of different resistivities.
Finite
difference (FD) modeling is a common way for solving partial differential
equations. The more complicated the model is
the greater the size of the
linear system to be solved. The main
feature of a 2.5D (dimensional) problem is that the originally 3D problem is
substituted by a series of 2D ones in the spatial wave-number domain. The
numerical determination of the FEM response over 2D structure requires great
amount of computation, because after the Fourier transform of the Maxwell's
equations finite difference method is applied in the spatial wave-number domain
an a linear set of equations has to be solved for each wave-number. Independent
of the FD problem usually a lot of tasks repeated making possible to realize
the principle of parallelization.
This paper
describe the used practical techniques and solutions that the forward 2.5D FEM
modeling application are capable to exploits the computational capacity of the
SEE-GRID Grid infrastructure in the aim of reducing its overall computational
time.