Universität Rostock, 2021
Abstract: We investigate the mode of a distribution defined on a function space, e.g. the space of integrable functions. We give a definition of the mode using small ball probabilities. We use entropy methods, e.g. finite covers, to define an estimator of the mode and to deduce its asymptotic behaviour. We show strong consistency and continue to derive the optimal rate of convergence over a class of distributions whose modes are contained in a totally bounded subset of the function space.
doctoral thesis free access