feat: prove through a Multiframe trait #709

huitseeker · 2023-09-27T21:11:00Z

introduces a Multiframe trait generalizing over the circuit machinery produced by LEM & Lurk,
implements it for the Lurk Circuit,
converts the nova proof file to work with this MultiFrame trait, making proving generic.

Nest steps :

implementing the MultiFrame trait for LEM (see Lem integration #629 Implement NIVC Coprocessors. #677)
refactoring the test suite in src/proof/nova.rs so that it tests each expression with LEM and Lurk,
extending the same logic to SuperNova

Mutability leftovers (see lurk-lab#680) - Updated benchmarking scripts to no longer mutate the `store` variable during evaluation frames retrieval. - Reformed `get_evaluation_frames` method across several files to use immutable references to `store` instead of mutable references. Multiframe definition - Introduced several new traits - `CEKState`, `FrameLike`, `EvaluationStore` and `MultiFrameTrait` to manage complex evaluation and circuit proof operations in the `proof/mod.rs` file. - Detailed implementations of the above traits introduced in `circuit/circuit_frame.rs`, providing methods for handling evaluation frames, synthesizing data from multiple frames and more.

- uses the generic MultiFrame trait introduced in the prior commit to generically prove using nova, - For references on the latest work defining this trait, see lurk-lab#629 (and its history, incl. lurk-lab#642, lurk-lab#707) and lurk-lab#677. - make the benches use the generic trait, so LEM can bench similarly, - make the examples use the generic trait, so LEM can example similarly, - make Supernova use the generic trait, so LEM can NIVC similarly,

arthurpaulino · 2023-09-27T22:14:04Z

!benchmark

arthurpaulino

✨ 🎉
Let's just see the benchmarks to make sure that nothing weird happened.

arthurpaulino · 2023-09-27T22:25:59Z

src/proof/nova.rs

@@ -866,7 +928,7 @@ pub mod tests {
        let s = &mut Store::<Fr>::default();
        let expected = s.num(3);
        let terminal = s.get_cont_terminal();
-        test_aux::<Coproc<Fr>>(
+        test_aux::<_, _, C1<'_, _, Coproc<_>>>(


Can we abbreviate this type with a local type definition? It will probably avoid a bunch of line additions.

It would also setup an organized pattern for when we introduce the LEM tests.

I'm not quite sure what local type definition you have in mind, but I'm of course in favor of the PR re-organizing the proof tests to put LEM & Lurk side-by-side doing some heavy refactoring to do so.

The line changes made here are such that those generic methods that need annotation now bear the exact same annotation on each line, syntactically - this makes them ripe for mechanized refactoring. So I'm not too worried about that.

huitseeker#8 for a -800 lines diff

arthurpaulino · 2023-09-27T23:01:52Z

I ran the benchmarks locally and everything is okay

end2end_benchmark/end2end_go_base_nova/_10_0                                                                          
                        time:   [1.3718 s 1.3738 s 1.3756 s]
                        change: [-0.1598% +0.1216% +0.4690%] (p = 0.54 > 0.05)
                        No change in performance detected.

store_benchmark/store_go_base_bls12/_10_16                                                                           
                        time:   [152.84 µs 152.90 µs 152.96 µs]
                        change: [+2.0234% +2.1403% +2.2462%] (p = 0.00 < 0.05)
                        Performance has regressed.
store_benchmark/store_go_base_pallas/_10_16                                                                           
                        time:   [149.14 µs 149.20 µs 149.25 µs]
                        change: [+0.6970% +0.9125% +1.0992%] (p = 0.00 < 0.05)
                        Change within noise threshold.
store_benchmark/store_go_base_bls12/_10_160                                                                           
                        time:   [150.55 µs 150.58 µs 150.62 µs]
                        change: [+0.9360% +1.0404% +1.1608%] (p = 0.00 < 0.05)
                        Change within noise threshold.
store_benchmark/store_go_base_pallas/_10_160                                                                           
                        time:   [150.52 µs 150.58 µs 150.64 µs]
                        change: [+1.5555% +1.6116% +1.6773%] (p = 0.00 < 0.05)
                        Performance has regressed.

hydration_benchmark/hydration_go_base_bls12/_10_16                                                                           
                        time:   [226.64 ns 226.67 ns 226.72 ns]
                        change: [+0.1890% +0.2361% +0.2820%] (p = 0.00 < 0.05)
                        Change within noise threshold.
hydration_benchmark/hydration_go_base_pallas/_10_16                                                                           
                        time:   [227.11 ns 227.14 ns 227.17 ns]
                        change: [+0.4104% +0.4546% +0.4982%] (p = 0.00 < 0.05)
                        Change within noise threshold.
hydration_benchmark/hydration_go_base_bls12/_10_160                                                                           
                        time:   [227.14 ns 227.19 ns 227.24 ns]
                        change: [+0.4554% +0.4891% +0.5220%] (p = 0.00 < 0.05)
                        Change within noise threshold.
hydration_benchmark/hydration_go_base_pallas/_10_160                                                                           
                        time:   [226.43 ns 226.63 ns 226.99 ns]
                        change: [+0.1814% +0.2476% +0.3242%] (p = 0.00 < 0.05)
                        Change within noise threshold.

eval_benchmark/eval_go_base_bls12/_10_16                                                                           
                        time:   [474.64 µs 474.77 µs 474.90 µs]
                        change: [+0.5484% +0.6288% +0.7118%] (p = 0.00 < 0.05)
                        Change within noise threshold.
eval_benchmark/eval_go_base_pallas/_10_16                                                                           
                        time:   [475.62 µs 475.78 µs 475.97 µs]
                        change: [+0.8392% +0.9057% +0.9782%] (p = 0.00 < 0.05)
                        Change within noise threshold.

eval_benchmark/eval_go_base_bls12/_10_160                                                                            
                        time:   [4.5522 ms 4.5547 ms 4.5575 ms]
                        change: [+0.1824% +0.8828% +2.0558%] (p = 0.08 > 0.05)
                        No change in performance detected.

eval_benchmark/eval_go_base_pallas/_10_160                                                                            
                        time:   [4.5658 ms 4.5680 ms 4.5703 ms]
                        change: [+0.8843% +0.9426% +0.9984%] (p = 0.00 < 0.05)
                        Change within noise threshold.

prove_benchmark/prove_go_base_nova/_10_0                                                                           
                        time:   [1.3575 s 1.3643 s 1.3683 s]
                        change: [-0.8152% -0.0892% +0.7712%] (p = 0.85 > 0.05)
                        No change in performance detected.

prove_compressed_benchmark/prove_compressed_go_base_nova/_10_0                                                                          
                        time:   [5.3678 s 5.3903 s 5.4138 s]
                        change: [-0.2555% +0.1745% +0.6378%] (p = 0.48 > 0.05)
                        No change in performance detected.

verify_benchmark/verify_go_base_nova/_10_0                                                                          
                        time:   [184.19 ms 185.63 ms 187.87 ms]
                        change: [+1.3051% +2.5493% +4.2089%] (p = 0.00 < 0.05)
                        Performance has regressed.

verify_compressed_benchmark/verify_compressed_go_base_nova/_10_0                                                                          
                        time:   [186.25 ms 187.16 ms 188.40 ms]
                        change: [-0.4886% +0.2702% +0.9845%] (p = 0.50 > 0.05)
                        No change in performance detected.

use a type alias to simplify test calls

huitseeker · 2023-09-28T15:36:26Z

This will be blocked on @porcuquine 's review because of the removal of the &mut modifier on generate_frames (which was actually permitted since #680 => bypassing to unblock.

huitseeker requested review from porcuquine and a team as code owners September 27, 2023 21:11

huitseeker force-pushed the multiframe_trait branch from 7cb9ad2 to b239e95 Compare September 27, 2023 21:12

huitseeker requested a review from a team as a code owner September 27, 2023 21:12

huitseeker force-pushed the multiframe_trait branch 5 times, most recently from 223a363 to 90e1dfd Compare September 27, 2023 21:36

huitseeker force-pushed the multiframe_trait branch from 90e1dfd to 6b5f11f Compare September 27, 2023 21:41

arthurpaulino previously approved these changes Sep 27, 2023

View reviewed changes

arthurpaulino and others added 2 commits September 28, 2023 10:23

use a type alias to simplify test calls

5a962c4

Merge pull request #8 from arthurpaulino/multiframe_trait_shortened

890f956

use a type alias to simplify test calls

huitseeker dismissed arthurpaulino’s stale review via 890f956 September 28, 2023 13:33

arthurpaulino approved these changes Sep 28, 2023

View reviewed changes

gabriel-barrett approved these changes Sep 28, 2023

View reviewed changes

huitseeker merged commit aa27d0f into lurk-lab:master Sep 28, 2023
6 checks passed

huitseeker mentioned this pull request Oct 7, 2023

On duplication in LEM integration #639

Closed

This was referenced Jan 3, 2024

[LUR-20] Remove MultiframeTrait #1014

Closed

[LUR-21] Remove the MultiFrameTrait trait #1015

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: prove through a Multiframe trait #709

feat: prove through a Multiframe trait #709

huitseeker commented Sep 27, 2023 •

edited

Loading

arthurpaulino commented Sep 27, 2023

arthurpaulino left a comment

arthurpaulino Sep 27, 2023

huitseeker Sep 28, 2023

arthurpaulino Sep 28, 2023

arthurpaulino commented Sep 27, 2023

huitseeker commented Sep 28, 2023

feat: prove through a Multiframe trait #709

feat: prove through a Multiframe trait #709

Conversation

huitseeker commented Sep 27, 2023 • edited Loading

Nest steps :

arthurpaulino commented Sep 27, 2023

arthurpaulino left a comment

Choose a reason for hiding this comment

arthurpaulino Sep 27, 2023

Choose a reason for hiding this comment

huitseeker Sep 28, 2023

Choose a reason for hiding this comment

arthurpaulino Sep 28, 2023

Choose a reason for hiding this comment

arthurpaulino commented Sep 27, 2023

huitseeker commented Sep 28, 2023

huitseeker commented Sep 27, 2023 •

edited

Loading