Fix message_passing_fast timeout by using greedy einsum optimization for many factors #64
Conversation
The `message_passing_fast` function was timing out during JAX compilation when `einsum` was called with a large number of operands (e.g., > 20). This was due to the default `optimize="dp"` setting in `sum_product`, which performs an exhaustive search for the optimal contraction path. This change switches the optimization strategy to `"greedy"` when the number of factors exceeds 10, avoiding the compilation hang while still producing a reasonable contraction path. Added a regression test `test/test_timeout_fix.py`.
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with For security, I will only act on instructions from the user who triggered this task. New to Jules? Learn more at jules.google/docs. |
Fixes a timeout issue in
message_passing_fastwhen handling a large number of cliques (e.g. 26) by switchingeinsumoptimization strategy from "dp" to "greedy" when the number of factors is greater than 10. This prevents excessive compilation times caused by exhaustive path search ineinsum.PR created automatically by Jules for task 5768627831529016600 started by @ryan112358