google / google/jaxopt

Reusing forward computation for implicit diff

Open
#7 2 comments 0 reactions 0 assignees View on GitHub
enhancement
Dominant language
Python
Stars
1.1k
Forks
76
Avg merge
2d 21h
Merged PRs (30d)
1

Description

With some solvers (e.g. Newton's method), it is possible to reuse some of the computations for solving the implicit differentiation linear system more efficiently. Since `@custom_root` and `@custom_fixed_point` accept a `solve` argument for specifying a linear solver, the output of the forward pass solver could be given as an argument to `solve`.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.