8000 `self.M`, `self.N` should be deprecated in 1D and 2D · Issue #881 · kymatio/kymatio · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

self.M, self.N should be deprecated in 1D and 2D #881

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue a 8000 nd 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

Closed
lostanlen opened this issue Jun 15, 2022 · 5 comments
Closed

self.M, self.N should be deprecated in 1D and 2D #881

lostanlen opened this issue Jun 15, 2022 · 5 comments
Assignees
Labels
2D Issue with 2D scattering code 3D Issue with 3D scattering code
Milestone

Comments

@lostanlen
Copy link
Collaborator
lostanlen commented Jun 15, 2022

for consistency with 1D #857
we should rename self.M_padded and self.N_padded to self._M_padded and self._N_padded

same for 3D, self._O_padded etc.

@lostanlen lostanlen added 2D Issue with 2D scattering code 3D Issue with 3D scattering code labels Jun 15, 2022
@lostanlen lostanlen added this to the 0.3.alpha milestone Jun 15, 2022
@lostanlen lostanlen self-assigned this Jun 15, 2022
@lostanlen lostanlen changed the title self.M and self.N should be deprecated self.M, self.N, self.O should be deprecated Jun 15, 2022
@lostanlen
Copy link
Collaborator Author
lostanlen commented Jun 18, 2022

closed by #863 and #886

@MuawizChaudhary
Copy link
Collaborator

What about 3d?

@lostanlen
Copy link
Collaborator Author

@MuawizChaudhary there is no padding in 3d, so the semantics are different.

@MuawizChaudhary
Copy link
Collaborator

the title self.O makes no sense then

@lostanlen lostanlen changed the title self.M, self.N, self.O should be deprecated self.M, self.N should be deprecated Jun 18, 2022
@lostanlen lostanlen changed the title self.M, self.N should be deprecated self.M, self.N should be deprecated in 1D and 2D Jun 18, 2022
@lostanlen
Copy link
Collaborator Author

@MuawizChaudhary the main goal of this was to allow non-power-of-two inputs in Scattering1D, in the prospect of JTFS. This goal is fulfilled now.

The longer-term goal is to allow shape=None (i.e., shape-agnostic scattering) in every dimension. This renaming helps a bit with the longer-term goal but we will need to have a debate in terms of how we name things.

  • do we keep self.N_padded, self.M_padded
  • do we join them as a single constant named self._shape_padded in every dimension
  • if so, is there a shorter name for self._shape_padded
  • how does this interact with Options for padding #859

these are all questions worth asking between v0.3 and v0.4

@lostanlen lostanlen modified the milestones: 0.3.alpha, 0.4 Mar 3, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
2D Issue with 2D scattering code 3D Issue with 3D scattering code
Projects
None yet
Development

No branches or pull requests

2 participants
0