From non-Markovian processes to stochastic real time control for tokamak plasma turbulence via artificial intelligence techniques

The mathematical approach to the magnetic plasma confinement is considered. In this direction we obtain the interpretation of possible generalization of Ruelle-Perron-Frobenius theorem and Bayesian inverse method. The aim is to use artificial neural networks for memorize the past history and on such a way the problem of controlled non-Markovian processes is transferred toward fuzzy Markovian processes. If this process is ergodic then we obtain crisp Markovian process good for application of Bayesian method to plasma confinement.