|
|||||
|
|
A central limit theorem in nonlinear filtering |
|
| Abstract: | It is shown that the probability law of a diffusion process conditioned on weakly corrupted observations is asymptotically Gaussian when properly scaled. The method of proof involves Fisher information matrices and a Cramér-Rao inequality. |
| Keywords: | Fisher information asymptotically Gaussian filter low observation noise Ams 1980 subject classification: 60f05 93e11 |