La méthode {D.S.I.}: Optimisation, Implémentation et Applications

Richard Cognot. ( 1996 )
INPL

Abstract

3D modelling in geology is a growing research topic aimed at the representation, in a form that can be transmitted easily from one geologist to another, of the complexity of the subsurface geologie structure. Classical C.A.D. methods have proven to be inadequate to the complexity of data encountered in geology, so for that reason Professeur Mallet introduced a new modelling approach in the frame of the GÜCADproject; based on a powerfull 3D interpolation method: Dicrete Smooth Interpolation (D.S.I.). The work presented in this document was performed in the frame of this project, and consisted in characterizing this method to tune its use in the GÜCAD software. In particular we focused our attention on the convergence problems that could arise, and on numerical convergence acceleration techniques. We also defined a completely general implemetation framework, defining the interpolation scheme as a relation between an "interpolation support" object and a "controller" object. This model will be known as the "Atomic-Controller" framework. We will then apply this framework to the resolution of common problems in geophysics.

Download / Links

BibTeX Reference

@PHDTHESIS{Cognot96,
    author = { Cognot, Richard },
     title = { La méthode {D.S.I.}: Optimisation, Implémentation et Applications },
   chapter = { 0 },
      year = { 1996 },
    school = { INPL },
  abstract = { 3D modelling in geology is a growing research topic aimed at the representation, in a
form that can be transmitted easily from one geologist to another, of the complexity of
the subsurface geologie structure. Classical C.A.D. methods have proven to be inadequate
to the complexity of data encountered in geology, so for that reason Professeur Mallet
introduced a new modelling approach in the frame of the GÜCADproject; based on a
powerfull 3D interpolation method: Dicrete Smooth Interpolation (D.S.I.).
The work presented in this document was performed in the frame of this project,
and consisted in characterizing this method to tune its use in the GÜCAD software. In
particular we focused our attention on the convergence problems that could arise, and
on numerical convergence acceleration techniques. We also defined a completely general
implemetation framework, defining the interpolation scheme as a relation between an
"interpolation support" object and a "controller" object. This model will be known as
the "Atomic-Controller" framework. We will then apply this framework to the resolution
of common problems in geophysics. }
}