scientific article; zbMATH DE number 3968340

From MaRDI portal
Publication:3735478





[For the entire collection see Zbl 0585.00014.] The idea of the Bayesian approach is to consider the function to be minimized as a sample observation of a stochastic function. We can therefore define a search procedure which minimizes the average deviation of the observation from the minimum. This differs from classical numerical analysis, where the maximum deviation is usually considered. The average deviation can be defined mathematically if the probability distribution is fixed. Since the main objections to the Bayesian approach are doubts about the existence and nature of a priori distributions, a system of simple and natural assumptions are introduced which provide for the existence of a unique family of finite-dimensional probability density functions and define the a priori distributions on the space of functions to be minimized.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3735478)