Serving the Quantitative Finance Community
Kalman Filter
For non-gaussian problems you can use a particle filter with the unscented kalman filter or the central-difference kalman filter. Neither of which require derivatives to be calculated.QuoteOriginally posted by: NYou bet it does.
-
- KackToodles
- Posts: 0
- Joined:
Kalman Filter
unscented?
Last edited by KackToodles on December 16th, 2005, 11:00 pm, edited 1 time in total.
Kalman Filter
This is what I thought because only second moments are used in updating.Could you tell me the source of the assumption?Is it because in the derivation, the loss function is defined as quadratic?
-
- Cuchulainn
- Posts: 23029
- Joined:
Kalman Filter
QuoteOriginally posted by: JamesH83This is what I thought because only second moments are used in updating.Could you tell me the source of the assumption?Is it because in the derivation, the loss function is defined as quadratic?In general, assumptions are the issue There are many answers.
Kalman Filter
Ocksendal did a good derivation of the Kalman filter in his first book. He assumes Wiener noise, of course.
Kalman Filter
or alternatively there are a couple of different derivations in ...
-
- KackToodles
- Posts: 0
- Joined:
Kalman Filter
why don't they just a simpler distribution, like a discrete binary one, to explain how it works in simple terms?