feat(ivc): impl cyclefold zero step #404
Conversation
cyphersnake
force-pushed
the
369-sfc-10
branch
from
ddd13b6 to
ae71038
Compare
|
A lot of staff will changed in future PR, not spent a lot of time with IVC::new |
| initial_secondary_trace: FoldablePlonkTrace<C2>, | ||
| pub support_ck: CommitmentKey<CSup>, | ||
| pub support_S: PlonkStructure<CSup::ScalarExt>, | ||
| pub support_initial_trace: FoldablePlonkTrace<CSup>, |
There was a problem hiding this comment.
what are difference between FoldablePlonkTrace and PlonkTrace? Can we just use one type of trace?
There was a problem hiding this comment.
This is a NewType pattern for checking conditions
Using NonZero as an example: using this type implies that it is not zero and you don't need to use a check everywhere.
Specifically here it saved me hours of debugging or hundreds of lines like assert!(trace.instances.len() == MARKERS_LEN).
Well and this has been in the code base for a some time
| @@ -20,11 +20,11 @@ pub const RATE: usize = T - 1; | |||
| pub const R_F: usize = 10; | |||
| pub const R_P: usize = 10; | |||
|
|
|||
| /// Safety: because 32 != 0 | |||
| pub const DEFAULT_LIMB_WIDTH: NonZeroUsize = unsafe { NonZeroUsize::new_unchecked(32) }; | |||
| /// Safety: because 64 != 0 | |||
There was a problem hiding this comment.
what is the reason to increase both LIMB_WIDTH and COUNT_LIMIT here?
There was a problem hiding this comment.
This simply reduced the number of rows, the search for optimal values is left to the benchmark stage
| @@ -430,12 +449,55 @@ impl<F: PrimeField> SangriaAccumulatorInstance<F> { | |||
|
|
|||
| pub type SangriaCrossTermCommits<F> = Vec<(AssignedValue<F>, AssignedValue<F>)>; | |||
|
|
|||
| pub struct PairedIncoming<F: PrimeField> { | |||
There was a problem hiding this comment.
NBD: add a short comment to describe what is it, e.g. something like
// instance from support circuit
There was a problem hiding this comment.
Ok, will add at the end of PR-train ✍️
| @@ -408,7 +576,121 @@ impl<const ARITY: usize, F: PrimeField, RO: ROTrait<F>> AbsorbInRO<F, RO> for In | |||
|
|
|||
| impl<const ARITY: usize, F: PrimeField> Input<ARITY, F> { | |||
| pub(super) fn get_without_witness(&self) -> Self { | |||
There was a problem hiding this comment.
Just to double confirm, this is used for initial input at step zero?
There was a problem hiding this comment.
Yes, in pp.new
Motivation
Part of #373 and testing of #369
Overview
The first part of the implementation of the zero step cyclefold. Split because of size of changes
Temporarily commented out part of cyclefold step folding circuit concerning pairing check