1: \begin{abstract}

2: In this paper, we consider maximum likelihood estimations of the degree of freedom parameter $\nu$, 

3: the location parameter $\mu$  and the scatter matrix $\Sigma$ 

4: of the multivariate Student-$t$ distribution.

5: In particular, we are interested in estimating the degree of freedom parameter $\nu$ that determines 

6: the tails of the corresponding probability density function and was rarely considered in detail in the literature so far.

7: We prove that under certain assumptions a minimizer of the negative log-likelihood function exists, where we have to take special care of the case

8: $\nu \rightarrow \infty$, for which the Student-$t$ distribution approaches the Gaussian distribution.

9: As alternatives to the classical EM algorithm we propose three other algorithms which cannot be interpreted

10: as EM algorithm. For fixed $\nu$, the first algorithm is an accelerated EM algorithm known from the literature.

11: However, since we do not fix $\nu$, we cannot apply standard convergence results for the EM algorithm.

12: The other two algorithms differ from this algorithm in the iteration step for $\nu$. 

13: We show how the objective function   behaves for the different updates of $\nu$

14: and prove for all three algorithms that it  decreases in each iteration step.

15: We compare the algorithms as well as some accelerated versions 

16: by numerical simulation and apply one of them for estimating the degree of freedom parameter

17: in images corrupted by Student-$t$ noise.   

18: \end{abstract}

19: