1: \begin{abstract}

2: A stationary Gaussian process is said to be long-range dependent (resp.

3: anti-persistent) if its spectral density $f(\lambda)$ can be written

4: as $f(\lambda)=|\lambda|^{-2d}g(|\lambda|)$, where $0<d<1/2$ (resp.

5: $-1/2<d<0$), and $g$ is continuous. We propose a novel Bayesian

6: nonparametric approach for the estimation of the spectral density

7: of such processes. Within this approach, we prove posterior consistency

8: for both $d$ and $g$, under appropriate conditions on the prior

9: distribution. We establish the rate of convergence for a general class

10: of priors, and apply our results to the family of fractionally exponential

11: priors. Our approach is based on the true likelihood function, and

12: does not resort to Whittle's approximation, which is not valid in

13: a long memory set-up. 

14: 

15: Key-words: Bayesian nonparametric;

16: consistency;

17: FEXP priors;

18: Gaussian long memory processes;

19: rates of convergence

20: \end{abstract}

21: