INFERENCE OF PHYSICAL MODELS USING PROBABILISTIC ENTROPY APPLIED TO THE ANALYSIS OF VARIABILITY OF RECURRENCE AND MODELING REGIONAL EXTREMES.Author:
SOLANA ORTEGA ALBERTO.
Year: 2004.
University:
COMPLUTENSE DE MADRID.
Place of defense: FACULTAD CC FÍSICA.
Place of preparation: INSTITUTO MATEMÁTICAS Y FÍSICA FUNDAMENTAL (CSIC).
Summary: It examines the current status of studies on risks associated with fonómenos extremes, highlighting the obstacles that limit to the statistical techniques that are commonly used for modelización.Se introduced a new framework of analysis based on the probability logic, which built languages plausible logic with observational data, adecudos to represent the uniqueness that comprise elmentos facts of evidencia.A from requirements of rationality and morality derives procedure inference entrópica REF, which was characterized by double extremización the relative entropy subject to ligatures fractiles of equiprobabilidad anchored at the datos.Apoyadas in the use of this procedure are proposed paths methodologies for the analysis of variability of recurrence and regionalization of extreme events, which require advance the solution of two problemas-tipo: the problem inference climb conditioned units funcionales.Por Finally, in order to illustrate these methodologies will include several examples of application, namely estudioo Probabilistic variability of the historical recurrence of the flooding river Guadelquivir in Andújar, probabilistic modeling of the accident tanker spills and causing major oil spills off the coast of Galicia and the regional analysis of the maximum annual avenues of a set of unregulated rivers of northern Spain.