Skip to content
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

Online wrapper is not working #41

Open
lionel-rigoux opened this issue Jul 19, 2017 · 2 comments
Open

Online wrapper is not working #41

lionel-rigoux opened this issue Jul 19, 2017 · 2 comments
Assignees
Labels

Comments

@lionel-rigoux
Copy link
Member

lionel-rigoux commented Jul 19, 2017

For multiple reasons:

  • the new display uses isYout, which is not passed correctly (ie. truncated) by onlinewrapper
  • I broke the online display for hyperparameters (my bad)
  • some suffstats are not passed correctly. In particular, as only one point/state is inverted, suffstats does not return any dx infos, only dx0.

I am creating a new branch fix-online to clean up the mess I induced by the change in display.
What is the expected behaviour? All params/HP are displayed as function of time?

@lionel-rigoux
Copy link
Member Author

lionel-rigoux commented Jul 19, 2017

Ok, I think the first two issues are solved in the fix-online branch.
For the last point, I am not sure about what to do.
demo_stochasticBinomial gives an error about the suffstat dimension
demo_lorenz seems to simply fail after few iterations, but maybe that is expected from the online inversion (?)

@lionel-rigoux
Copy link
Member Author

  • Main display bugs have been solved in the development branch.
    There are still two issues remaining in the toolbox:
  1. online inversion only works with stochastic systems. For deterministic systems, the switch to ODE messes up the state dimension and thus the online recursion.

  2. Online inversion seems to fail after some iterations, at least in some demos (eg. demo_2Dlin). Everything's fine until some timepoint from which no update takes place (state and observation are stuck to the same value). The algorithm throw no errors.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

2 participants