In probability and statistics, the Dirichlet distribution (after Peter Gustav Lejeune Dirichlet) is a family of continuous multivariate probability distributions parameterized by a vector {\displaystyle {\boldsymbol {\alpha }}} of positive reals. It is a multivariate generalization of the beta distribution. Dirichlet distributions are very often used as prior distributions in Bayesian statistics, and in fact the Dirichlet distribution is the conjugate prior of the categorical distribution and multinomial distribution.