feat: prove/verify and sync plonky3 #523
Conversation
| let dft = Dft::default(); | ||
| let fri_params = p3_fri::create_benchmark_fri_params(challenge_mmcs); | ||
| let mut fri_params = p3_fri::create_benchmark_fri_params(challenge_mmcs); | ||
| fri_params.log_blowup = 4; // Use a higher blowup than default for degree 9 constraints |
There was a problem hiding this comment.
Should we fix this now or is there something blocking it?
There was a problem hiding this comment.
Should we fix this now or is there something blocking it?
I don't think anything is blocking it. There are multiple functions in p3_fri to create params (create_test_fri_params, create_test_fri_params_zk, create_benchmark_fri_params..). Maybe we could add a create_miden_fri_params with the values we want if we know them all.
Here it generates:
FriParameters {
log_blowup: 4,
log_final_poly_len: 0,
num_queries: 100,
proof_of_work_bits: 16,
mmcs,
log_folding_factor: 1,
}There was a problem hiding this comment.
Makes sense, and also a configuration with blow up factor 8 for Miden VM constraints.
There was a problem hiding this comment.
Makes sense, and also a configuration with blow up factor 8 for Miden VM constraints.
Did you mean like so? 0xMiden/Plonky3#19
There was a problem hiding this comment.
Do we need log_blowup=4? I thought we only went up until 3 for our constraints, but maybe we're doing 4 for smaller proofs?
There was a problem hiding this comment.
Agreed, log_blowup=3 should be enough for us in this context
There was a problem hiding this comment.
I've updated #19 to only have a log_blow_up of 3 and one set of parameters for miden, when targetting it here I have some issues but I think just increasing the number of rows for the proof will be enough, I'll test it.
There was a problem hiding this comment.
Also will reduce the constraints degrees for a few of the tests.
|
@Leo-Besancon , should we create an issue in the P3 repo regarding missing support for aux boundary constraints? |
I've added the issue there: 0xMiden/Plonky3#17, would be great if you (or @adr1anh) could check it and add comments if needed! |
Describe your changes
This PR, based on #461, aims to:
prove/verifyin E2E tests for full workflow #512It may be used in a full E2E workflow.