-
-
Notifications
You must be signed in to change notification settings - Fork 985
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Normalization of Weights within SMCFilter class #3382
Labels
Comments
Our general stance is against unnecessary normalization, in the statistics field in general. The
|
Sorry, I misread. I do think it's fine to implement a |
fritzo
added
enhancement
help wanted
Issues suitable for, and inviting external contributions
labels
Jul 15, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
The weights reported by the get_emperical method in the SMCFilter class will usually not add up to one. I think it would be nice to have the option to directly get the normalized weights (maybe that should also be the default) to comply with the literature (see for example An Introduction to Sequential Monte Carlo , page 130).
The weights are actually normalized in this line. However, the result is saved in a local variable instead of updating the state weights variable. It seems like the values of the log_weights variable are only updated to be in a range between 0 and 1 using this command.
What is the reason behind not directly normalizing the weights variable?
The easiest fix for the issue would be to add an optional argument to the get_emperical function that if true normalizes the weights, e.g.:
The text was updated successfully, but these errors were encountered: