Skip to Main content Skip to Navigation

Simulation conditionnelle de modèles isofactoriels

Xavier Emery 
Abstract : This thesis aims at developing conditional simulation algorithms for random fields with bivariate isofactorial distributions, i.e. such that there exists a basis of functions (factors) with no spatial cross-correlation. Isofactorial distributions fulfill two constraints that are often difficult to conciliate: on the one hand, they constitute a wide class of models and adapt to the description of numerous phenomena; on the other hand, their statistical inference relies only on a few parameters. The first part of the thesis highlights the limits and approximations of an "all-purpose" technique: the sequential algorithm. It consists in simulating the field locations one after the other, according to a random sequence, by estimating at each location the conditional distribution of the unknown value by either an indicator or a disjunctive kriging. In the second part, new models and algorithms are proposed in order to achieve the simulation and the conditioning to experimental values. Several structural tools are introduced and studied to enable the inference of bivariate distributions, in particular the madogram, variograms of order lower than two and indicator variograms. The concepts and methods are finally applied to a mining dataset (gold and silver accumulations in a Chilean vein-type deposit) characterized by highly skewed histograms.
Document type :
Complete list of metadata
Contributor : Ecole Mines ParisTech Connect in order to contact the contributor
Submitted on : Thursday, April 7, 2005 - 8:00:00 AM
Last modification on : Tuesday, September 29, 2015 - 10:32:44 AM


  • HAL Id : pastel-00001185, version 1



Xavier Emery. Simulation conditionnelle de modèles isofactoriels. domain_other. École Nationale Supérieure des Mines de Paris, 2004. English. ⟨pastel-00001185⟩



Record views


Files downloads