refactor: add precommitted polynomial reductions by RadNi · Pull Request #17 · LayerZero-Research/jolt

RadNi · 2026-05-21T01:28:38Z

Generated stack PR from amir/bytecode-commitment-merged.

Depends on spec PR: a16z#1565

Stack position: 02
Base branch: amir/bytecode-stack/01-stage6-split

Owned paths:

jolt-core/src/zkvm/claim_reductions/precommitted.rs
jolt-core/src/zkvm/claim_reductions/advice.rs
jolt-core/src/zkvm/claim_reductions/mod.rs
jolt-core/src/poly/commitment/dory/
jolt-core/src/poly/one_hot_polynomial.rs
jolt-core/src/poly/opening_proof.rs
jolt-core/src/poly/rlc_polynomial.rs
jolt-core/src/zkvm/prover.rs
jolt-core/src/zkvm/transpilable_verifier.rs
jolt-core/src/zkvm/verifier.rs

This PR is expected to be updated manually when amir/bytecode-commitment-merged is resliced.

github-actions · 2026-05-21T01:28:50Z

Warning

This PR has more than 500 changed lines and does not include a spec.

Large features and architectural changes benefit from a spec-driven workflow.
See CONTRIBUTING.md for details on how to create a spec.

If this PR is a bug fix, refactor, or doesn't warrant a spec, feel free to ignore this message.

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit e2c9276. Configure here.}

cursor · 2026-05-21T18:33:18Z

+            match DoryGlobals::get_layout() {
+                DoryLayout::AddressMajor => OpeningPoint::<BIG_ENDIAN, F>::new(
+                    [r_cycle_stage6.as_slice(), r_address_stage7.as_slice()].concat(),
+                ),


AddressMajor opening order mismatch

High Severity

AddressMajor Stage-8 now constructs opening_point as [r_cycle || r_address], while earlier reduction claims (for RA polynomials) are still cached at [r_address || r_cycle]. With reorder_opening_point_for_layout removed from DoryCommitmentScheme, those claims are combined against a different point, so batched opening claims no longer correspond to committed polynomial evaluations.

Additional Locations (2)

jolt-core/src/zkvm/verifier.rs#L345-L348

jolt-core/src/poly/commitment/dory/commitment_scheme.rs#L208-L216

^{Reviewed by Cursor Bugbot for commit e2c9276. Configure here.}

moodlezoup · 2026-05-22T12:10:10Z

 };

 pub type JoltFieldWrapper = ArkFr;
+type DenseTier1Setup = (


I don't think we need to name this type

moodlezoup · 2026-05-22T12:23:52Z

+        candidates
+    }
+
+    fn stage8_opening_point(&self) -> Result<OpeningPoint<BIG_ENDIAN, F>, ProofVerifyError> {


this is almost identical to the stage8_opening_point in prover.rs. We should DRY into one function

…rifier Adopt embedded-main Dory scheduling with shared precommitted claim-reduction plumbing so stage 6/8 can handle dominant precommitted contexts consistently in both zk and non-zk flows. Made-with: Cursor (cherry picked from commit d7b160b) Co-authored-by: Cursor <cursoragent@cursor.com>

Keep the precommitted scale when rebuilding advice claim-reduction state from bound coefficients so GPU cycle-phase handoff remains consistent across dummy rounds. (cherry picked from commit e58c0df)

Compute the verifier advice equality evaluation in the same precommitted opening order used by the prover, while keeping the cycle-phase skip scale separate from the full address-phase scale. Avoid reading Dory main_t when the precommitted schedule does not need it, so unit tests that do not initialize Dory globals can still construct the schedule. (cherry picked from commit f19c179)

Co-authored-by: Cursor <cursoragent@cursor.com>

…tack Co-authored-by: Cursor <cursoragent@cursor.com>

Co-authored-by: Cursor <cursoragent@cursor.com>

Use the cycle/address ordering expected by the active Dory layout when the main trace domain anchors Stage 8, so precommitted advice openings verify in AddressMajor mode. Co-authored-by: Cursor <cursoragent@cursor.com>

Keep the PR 02 Stage 8 verifier fix byte-for-byte aligned with the final merged branch while preserving the AddressMajor advice proof fix. Co-authored-by: Cursor <cursoragent@cursor.com>

Use local type inference for the dense AddressMajor setup tuple to avoid carrying an intermediate-only alias. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

Move the common Stage 8 Dory opening-point selection into the commitment scheme module so prover and verifier use the same layout logic. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

Cache real cycle-phase opening points for precommitted advice reductions so verifier phase transitions reconstruct the address phase from accumulator state. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

Co-authored-by: Cursor <cursoragent@cursor.com>

github-actions Bot added the no-spec PR has no spec file label May 21, 2026

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from d2b422b to b561ac0 Compare May 21, 2026 01:44

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 5386e59 to 4b2423f Compare May 21, 2026 01:44

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from b561ac0 to e528eda Compare May 21, 2026 01:51

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 4b2423f to 3de7f26 Compare May 21, 2026 01:51

RadNi changed the title ~~Add precommitted advice reductions~~ [02/05] Add precommitted advice reductions May 21, 2026

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 3de7f26 to 5217190 Compare May 21, 2026 01:56

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from e528eda to be768fc Compare May 21, 2026 15:37

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 5217190 to 9e4300e Compare May 21, 2026 15:37

RadNi changed the title ~~[02/05] Add precommitted advice reductions~~ [02/05] Add precommitted reductions May 21, 2026

RadNi changed the title ~~[02/05] Add precommitted reductions~~ [02/05] Add precommitted polynomials reductions May 21, 2026

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from be768fc to 4624ea6 Compare May 21, 2026 18:02

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 9e4300e to e2c9276 Compare May 21, 2026 18:02

RadNi marked this pull request as ready for review May 21, 2026 18:10

cursor Bot reviewed May 21, 2026

View reviewed changes

RadNi changed the title ~~[02/05] Add precommitted polynomials reductions~~ refactor: add precommitted polynomial reductions May 21, 2026

moodlezoup reviewed May 22, 2026

View reviewed changes

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch 2 times, most recently from d340832 to 6c29ef5 Compare May 22, 2026 17:30

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from 0eddf1c to 3e44c13 Compare May 22, 2026 17:30

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from a5ed82a to a6bfc73 Compare May 22, 2026 19:53

RadNi force-pushed the amir/bytecode-stack/01-stage6-split branch from 3e44c13 to 4175c7a Compare May 25, 2026 18:46

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from 0f357e1 to 5ca9b21 Compare May 25, 2026 18:46

RadNi and others added 6 commits May 25, 2026 20:36

fix(claim-reduction): preserve advice handoff scale

07b55f5

Keep the precommitted scale when rebuilding advice claim-reduction state from bound coefficients so GPU cycle-phase handoff remains consistent across dummy rounds. (cherry picked from commit e58c0df)

fix(zkvm): align precommitted advice with stage6 split

57cba2a

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): update transpilable verifier for precommitted advice

0bc845e

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): transition precommitted advice in transpilable verifier

41bff73

Co-authored-by: Cursor <cursoragent@cursor.com>

RadNi and others added 15 commits May 25, 2026 20:36

style: format precommitted advice branch

589fe61

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): satisfy strict clippy for precommitted advice

d1c27ec

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): remove unused advice output constraint helper

d641b5d

Co-authored-by: Cursor <cursoragent@cursor.com>

refactor(zkvm): align precommitted advice reduction shape

5b4c166

Co-authored-by: Cursor <cursoragent@cursor.com>

refactor(dory): move address-major layout cleanup into precommitted s…

8353ada

…tack Co-authored-by: Cursor <cursoragent@cursor.com>

refactor(dory): align commitment scheme before committed programs

b3c62bb

Co-authored-by: Cursor <cursoragent@cursor.com>

chore(dory): remove unused opening hint accessors

a169a45

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(dory): derive stage8 opening point by layout

b388d1a

Use the cycle/address ordering expected by the active Dory layout when the main trace domain anchors Stage 8, so precommitted advice openings verify in AddressMajor mode. Co-authored-by: Cursor <cursoragent@cursor.com>

fix(dory): match final stage8 verifier error

98dfdd0

Keep the PR 02 Stage 8 verifier fix byte-for-byte aligned with the final merged branch while preserving the AddressMajor advice proof fix. Co-authored-by: Cursor <cursoragent@cursor.com>

chore(dory): remove dense tier setup alias

9b37aef

Use local type inference for the dense AddressMajor setup tuple to avoid carrying an intermediate-only alias. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

refactor(dory): share stage8 opening point selection

eeb2613

Move the common Stage 8 Dory opening-point selection into the commitment scheme module so prover and verifier use the same layout logic. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): preserve precommitted cycle openings

2fea03c

Cache real cycle-phase opening points for precommitted advice reductions so verifier phase transitions reconstruct the address phase from accumulator state. Made-with: Cursor Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): gate final opening helper behind prover

0a70e07

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): share final opening helper

b5832c0

Co-authored-by: Cursor <cursoragent@cursor.com>

fix(zkvm): drop stale opening point imports

c41f679

Co-authored-by: Cursor <cursoragent@cursor.com>

RadNi force-pushed the amir/bytecode-stack/02-precommitted branch from e0ed5d6 to c41f679 Compare May 26, 2026 04:02

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

refactor: add precommitted polynomial reductions#17

refactor: add precommitted polynomial reductions#17
RadNi wants to merge 21 commits into
amir/bytecode-stack/01-stage6-splitfrom
amir/bytecode-stack/02-precommitted

RadNi commented May 21, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented May 21, 2026

Uh oh!

cursor Bot left a comment

Uh oh!

cursor Bot May 21, 2026

Uh oh!

moodlezoup May 22, 2026

Uh oh!

moodlezoup May 22, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

RadNi commented May 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented May 21, 2026

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot May 21, 2026

Choose a reason for hiding this comment

AddressMajor opening order mismatch

Uh oh!

moodlezoup May 22, 2026

Choose a reason for hiding this comment

Uh oh!

moodlezoup May 22, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

RadNi commented May 21, 2026 •

edited

Loading