Linear constraint remover #2425
Draft
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
chriseth
reviewed
Feb 3, 2025
ast/src/parsed/mod.rs
Outdated
| @@ -738,6 +738,30 @@ impl<Ref> From<u32> for Expression<Ref> { | |||
| BigUint::from(value).into() | |||
| } | |||
| } | |||
|
|
|||
| impl<T: FieldElement> From<AlgebraicExpression<T>> for Expression<Reference> { | |||
There was a problem hiding this comment.
There is pil-analyzer/src/epressionizer.rs that does this on a large scale. Maybe call that from pilopt instead of adding this here?
There was a problem hiding this comment.
I knew we had this somewhere! Thanks! I added this just to make it compile
github-merge-queue bot
pushed a commit
that referenced
this pull request
Feb 4, 2025
Depends on #2426. With this PR, we use the annotations added in #2412 and #2426 to find the columns being generated. This allows us to detect when the folded payload is not persisted as a witness column. The background is that @Schaeff is working on #2425, which would undo persisting the folded payloads in some cases, allowing us to spend fewer witness columns per bus interaction. With this PR, this should be fine: The column type will change from witness to intermediate, which means that the bus witgen will not output any folded columns. It can be tested by changing the `materialize_folded` bool to `false`, e.g. [here](https://github.com/powdr-labs/powdr/blob/e77d3801c1decff039fd0ec6bbeb55ed734357fb/std/protocols/bus.asm#L48) and running: ``` cargo run pil test_data/asm/block_to_block.asm \ -o output -f --linker-mode bus --prove-with mock --field gl ``` This used to fail before this PR.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
what
Replace certain multilinear constraints* with intermediate polynomials in pilopt.
*with no next references, not touching public inputs
why
This comment highlights a use-case where materialising a column which is only constrained to a non-shifted multilinear polynomial is wasteful. Fixing this when we create the code is tricky, since we cannot know the degree of the constraint from pil. This is however something pilopt can do, which is what we attempt to do here
unknowns
col witness x; col witness y; x = 2 * ygets completely optimized away. One argument for is that we already removex = yentirely.- it's fine since we're doing only substitutions
- Witgen works!
x = 42andx = ywhich are both multilinear expressions. They do no yield intermediate polynomials, since in that case it is fine to "inline" them. This implementation currently makes sure all three optimizer apply to distinct cases, but maybe it would be better to have a single optimizer which handles all three cases, only introducing intermediate polynomials in the non trivial cases (42andy)