Researches on Properties of Generalized Normalization Coefficient in NML Distribution

In coding theory the minimum description length principle plays a very important role. And the most important conception in minimum description length is stochastic complexity. In order to compute the stochastic complexity of some certain normalized maximum likelihood distribution, it is important to know the properties of C_n, which denotes the normalization coefficient in the universal normalized maximum likelihood distribution. In this paper, an integral expression of C_n in the Gama model class is proposed by the generation function. In this way we can solve some problems by this method. And we also define a generalized normalization coefficient C_n,k and present some properties about it