The EPH provides a way of "storing" information, under a probabilistic form : if an experiment, or a computational code, has been executed N times, it provides N results, each depending on the specific values given to the parameters. The EPH gives a way of anticipating, with some probability, a result associated with values of the parameters which have never been met before, or have been lost.

The EPH allows prevision (values in the future) or reconstruction of missing data (values in the past).

In practice, the EPH consists in a collection of densities of probability, one above each point of the configuration space. The density is a Dirac measure (certainty) if the measure has been made at that one point, otherwise, the density is less and less concentrated when we move further away from measure points.

The construction of EPH and its use is the first topic of Olga Zeydina's Ph.D. The general title of this Ph.D. is "Probabilistic Methods for Nuclear Safety". It is prepared at the University of South Britany (Vannes), under the joint supervision of Prof. Emile Le Page (Vannes) and Prof. Bernard Beauzamy (SCM SA).

All documents are in English, pdf format. Click on each link to download the corresponding report.

*****