1 Dependence in Statistics with applications

My work makes use of the dependence of stochastic processes, for which I am one of the world-leaders, for statistical inference. Randomness is an easy way to model phenomena for which it looks impossible to propose a deterministic model; think of the stock-exchanges, the quality control in the industry, meteorological forecasts or biological problems in computational genomics. Dependence between data clearly occurs in many time varying problems as well as in questions where geographic questions arise. Dependence definitely interacts with the random behaviors. The project is aimed at developing real statistical applications of dependence. The Short and Long Range Dependence (resp. SRD and LRD) will be considered in both the discrete and continuous-time cases, and also in the case of random fields (geographic dependence). Note that asymptotic results in case of SRD look similar to independent situations up to some constants. At the contrary LRD yields new normalizations. Valuable models should be introduced in order to fit the properties of current data: integer valued, continuous time or point process data will be considered. The identification of the models is an essential question; tests of goodness-of-fit are also needed to check their validity. The level of reliability of the procedures may be calculated from empirical methods. The previous problems give a special importance to the resampling techniques. This allows quantifying the quality of the calibration procedures. Applications will b