Skip to content

Instantly share code, notes, and snippets.

@sharanry

sharanry/inverse.ipynb

Last active August 1, 2019 04:35
Show Gist options
  • Save sharanry/a4b251c59de4f12cf68d81dea8721ec4 to your computer and use it in GitHub Desktop.
Save sharanry/a4b251c59de4f12cf68d81dea8721ec4 to your computer and use it in GitHub Desktop.
Download ZIP
Display the source blob
Display the rendered blob
Raw
inverse.ipynb
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
@sharanry
Copy link
Author

sharanry commented Aug 1, 2019 •
edited
Loading

Does changing Order16 to other order level improve things?

Unfortunately no. Since both inv and logabsdetjac have problems for radial flows, there might be a problem with transformation implementation itself.

In case of radial flows, where the inverse in completely off, one main deviation I had made from the paper, was the usage of softplus to accommodate the constrained nature of the \alpha parameter. Could this be the reason? Is there a better way to accomplish non-negative constraint?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment