1: \begin{abstract}

2: %

3: A model of Poissonian observation having a jump (change-point) in the

4: intensity function is considered.  Two cases are studied.  The first one

5: corresponds to the situation when the jump size converges to a non-zero limit,

6: while in the second one the limit is zero.  The limiting likelihood ratios in

7: these two cases are quite different.  In the first case, like in the case of a

8: fixed jump size, the normalized likelihood ratio converges to a log Poisson

9: process.  In the second case, the normalized likelihood ratio converges to a

10: log Wiener process, and so, the statistical problems of parameter estimation

11: and hypotheses testing are asymptotically equivalent in this case to the well

12: known problems of change-point estimation and testing for the model of a

13: signal in white Gaussian noise.  The properties of the maximum likelihood and

14: Bayesian estimators, as well as those of the general likelihood ratio, Wald's

15: and Bayesian tests are deduced form the convergence of normalized likelihood

16: ratios.  The convergence of the moments of the estimators is also established.

17: The obtained theoretical results are illustrated by numerical simulations.

18: %

19: \keywords{Poisson process \and non-regularity \and change-point \and limiting

20:   likelihood ratio process \and maximum likelihood estimator \and Bayesian

21:   estimators \and consistency \and limiting distribution \and efficiency \and

22:   general likelihood ratio test \and Wald's test \and Bayesian tests}

23: %

24: \subclass{62M05 \and 62M02}

25: %

26: \end{abstract}

27: