Projected gradient for multidimensional array
Open
- Dominant language
- Python
- Stars
- 1.1k
- Forks
- 76
- Avg merge
- 2d 21h
- Merged PRs (30d)
- 1
Description
Hi -- I have an array that is `mxnxn`. Each of the `m` `nxn` matrices represents a single "frame." I wish to enforce that each `nxn` matrix/frame is a doubly-stochastic matrix. For a single `nxn` matrix, I could use projected gradients with the [birkhoff](https://jaxopt.github.io/stable/_autosummary/jaxopt.projection.projection_birkhoff.html) projection. However, I would like to apply this constraint to *every* `nxn` matrix.
Does anyone know how to best implement this?
Contributor guide
Assessment
This issue has not been assessed yet.