1: \begin{abstract}

2: Stationary ergodic processes with finite alphabets are estimated by

3: finite memory processes from a sample, an $n$-length realization of the

4: process, where the memory depth of the estimator process is also

5: estimated from the sample using penalized maximum likelihood (PML).

6: Under some assumptions on the continuity rate and the assumption of

7: non-nullness, a rate of convergence in $\bar{d}$-distance is obtained,

8: with explicit constants. The result requires an analysis of the

9: divergence of PML Markov order estimators for not necessarily finite

10: memory processes. This divergence problem is investigated in more

11: generality for three information criteria: the Bayesian information

12: criterion with generalized penalty term yielding the PML, and the

13: normalized maximum likelihood and the Krichevsky--Trofimov code

14: lengths. Lower and upper bounds on the estimated order are obtained.

15: The notion of consistent Markov order estimation is generalized for

16: infinite memory processes using the concept of oracle order estimates,

17: and generalized consistency of the PML Markov order estimator is

18: presented.

19: \end{abstract}