feat: add PLONK in-circuit verifier #880
Conversation
|
📦 📦 📦 📦 📦 ❌ 📦 📦 📦 ❌ ❌ 📦 ❌ ❌ 📦 ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ 🚧 📦 ❌ 📦 📦 📦 |
|
📦 📦 📦 📦 📦 ❌ 📦 📦 📦 ❌ ❌ 📦 ❌ ❌ 📦 ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ 🚧 📦 ❌ 📦 📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
6ecc547
to
42a4d5c
Compare
|
📦 📦 📦 📦 📦 ❌ 📦 ❌ 📦 📦 ❌ ❌ 📦 🚧 📦 ❌ 📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
42a4d5c
to
54c2c35
Compare
|
📦 |
|
📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
f58117b
to
7dba1bd
Compare
|
📦 |
|
📦 📦 ❌ ❌ ❌ ❌ ❌ ❌ ❌ 📦 ❌ |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
08e1a34
to
93ef51f
Compare
|
📦 ❌ |
|
📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
1592414
to
1667207
Compare
|
📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
1667207
to
0ac8f1c
Compare
|
📦 ❌ 📦 ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ ❌ 📦 📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
0ac8f1c
to
4e6454d
Compare
|
📦 📦 |
|
📦 |
|
📦 |
|
📦 |
ivokub
force-pushed
the
feat/plonk-verifier
branch
from
6bebdca
to
5f8a060
Compare
|
Ready from my side. @ThomasPiellard - if you can confirm that also good from your side then I think it is good for review. I assigned @yelhousni for review as we both have worked on it :) |
ivokub
changed the title
|
Added MSM tests for emulated and 2-chain curves. This revealed another bug :) I think the PR is done from my side. We could merge from my side after approving review. I think we can address additional optimisations in different PRs, to keep number of different open PRs small and avoid needing to rebase/merge. Some other discussed optimisations:
Ping @yelhousni for review. |
Alright on my end. Let's merge this and address optimizations in other PRs. I gave the Wesolowski VDF approach a try to compute efficiently
|
|
Thanks for the review and co-authoring the PR! I gathered the ideas for improvements in #935. |
commit 6c05ea4 Author: Ivo Kubjas <ivo.kubjas@consensys.net> Date: Tue Nov 28 16:24:47 2023 +0100 perf: use G2 precomputed lines for Miller loop (#930) * feat: add lazy line eval for Miller loop * chore: go mod * fix: DoublePairFixed order * refactor: remove fixed Q specialized methods * chore: serialize lines for KZG key * chore: go generate * docs: add init docs * feat: add fixed KZG verification key init * test: add constant and fixed VK test cases * test: use fixed init * feat: add fixed Groth16 verification * fix: unused import * refactor: merge last manual iteration commit a99d198 Author: Ivo Kubjas <ivo.kubjas@consensys.net> Date: Tue Nov 28 16:20:44 2023 +0100 feat: add PLONK in-circuit verifier (#880) * test: add recursion hash tests * fix: accumulate MSM result * refactor: take emulated element for additional data * fix: handled infinity point in native multi scalar exp * fix: use only nbBits when creating scalar * feat: add PLONK verifier * feat: PlaceholderVerifyingKey takes the vk as argument * feat: f -> scalarApi * feat: addition of computeIthLagrangeAtZeta * feat: bsb commitments are added to pi * refactor: PlaceholderProof takes the proof as argument * fix: compute ith lagrange ok, hashToField failing * fix: native short hash output size * feat: add bw6 * docs: add package documentation * refactor: describe error in panic * refactor: init curve and pairing implicitly * refactor: remove comments * docs: add package examples * feat: add all supported witness assignments * test: add MSM test * fix: remove todo panic * feat: add option shortcuts * fix: include hash to field in shortcut option * feat: use only CCS for placeholder proof and verifyingkey * chore: typos and cleanup * docs: add KZG package documentation --------- Co-authored-by: Thomas Piellard <thomas.piellard@consensys.net> commit 62b52ea Merge: ec07217 97156f3 Author: Youssef El Housni <youssef.elhousni@consensys.net> Date: Fri Nov 24 10:44:33 2023 -0500 Merge pull request #933 from Consensys/perf/karabina-cycloSq Perf: variant of the Karabina cyclotomic squaring commit 97156f3 Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Fri Nov 24 10:27:00 2023 -0500 refactor: apply PR review suggestions commit f52c4cb Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Thu Nov 23 01:50:41 2023 -0500 perf(bls12-377): implement a variant of Karabina cyclo square commit d7e8d78 Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Wed Nov 22 23:28:26 2023 -0500 perf(bw6): implement a variant of Karabina cyclo square commit ec07217 Merge: 3aa2559 5479586 Author: Youssef El Housni <youssef.elhousni@consensys.net> Date: Wed Nov 22 18:16:46 2023 -0500 Merge pull request #931 from Consensys/perf/bw6-finalExp Perf: optimize addition chains in BW6-761 final exponentiation commit 5479586 Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Wed Nov 22 13:07:50 2023 -0500 perf(bw6/finalExp): replace Add(x,x) by MulConst(x,2) commit 65cd6ee Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Tue Nov 21 21:39:55 2023 -0500 fix(linter): ineffectual assignment commit d948c7c Author: Youssef El Housni <youssef.housni21@gmail.com> Date: Tue Nov 21 21:27:02 2023 -0500 perf(bw6/finalExp): optimize addition chains commit 3aa2559 Author: Gautam Botrel <gautam.botrel@gmail.com> Date: Mon Nov 20 14:03:52 2023 -0600 feat: if we don't compress we don't need the dict (#929)
Description
This PR adds in-circuit verifier supporting commitments for both 2-chain and emulated computations.
Fixes #847
Type of change
How has this been tested?
TestBLS12InBW6WoCommitTestBW6InBN254WoCommitTestBLS12InBW6CommitTestBW6InBN254CommitHow has this been benchmarked?
Checklist:
golangci-lintdoes not output errors locally