Large deviation theorems for Gaussian processes and their applications in information theory (Q5937964)

Various problems in information theory and mathematical statistics can be solved by investigating asymptotic behaviour of stochastic processes. The large deviation theorems are useful to study such asymptotic behaviour. The author solves some problems concerning the asymptotic behaviour of discrete time stationary Gaussian processes. The topics investigated here are error probability of string matching, error probabilities for random codings, a conditional limit theorem related to the maximum entropy principle, and error probabilities in hypothesis testing.