Iterative approximation of statistical distributions and relation to information geometry (Q1857369)

The paper refers to the information-theoretic entropy and the geometry of parametric statistical models and defines a Riemannian surface on which dynamics may be studied and optimized. The derivation is focused on the gamma distribution as a result of its interesting features (mainly its relation with other classical distributions). A B-spline approximation to the gamma distribution is derived. The estimation of the parameters of the distribution is considered as a control problem, and simulation results are displayed. The paper refers to stochastic differential geometry.