Skip to content

[QUESTION] Adjoint of Fast Fourier Transform #1138

Closed

Description

@mohsensadr

Is there a differentiable implementation of Fast Fourier Transform in Warp?

This is needed for physics-based simulation methods like particle-in-cell that solve Vlasov-Poisson type equation in fusion and particle accelerators.

Activity

  1. changed the title [-][QUESTION][/-] [+][QUESTION] Adjoint of Fast Fourier Transform[/+] on Dec 22, 2025
  2. deleted a comment from coderabbitai on Jan 7, 2026
  3. daedalus5 commented on Jan 7, 2026

    @daedalus5
    Contributor

    Currently we do not have a differentiable implementation of FFT. However this should be possible without too much effort. I'll upgrade this to a feature request.

  4. added this to the 1.12.0 milestone on Jan 15, 2026
  5. daedalus5 commented on Feb 3, 2026

    @daedalus5
    Contributor

    Hi @mohsensadr , the FFT and IFFT adjoints have been added. You can try them out using latest main. Let me know if you run into any issues.

  6. mohsensadr commented on Feb 4, 2026

    @mohsensadr
    Author

    Perfect! Thanks a lot. I will give it a try in the next days.

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

Metadata

Metadata

Assignees

Labels

feature requestRequest for something to be addedquestionThe issue author requires information

Type

No type

Projects

No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions