absorb instances in folding #2198
Conversation
folding/src/instance_witness.rs
Outdated
| let extended_commitments = self.extended.iter().flat_map(|commit| { | ||
| assert_eq!(commit.elems.len(), 1); | ||
| let point = commit.elems[0]; | ||
| //this may need change if we need to support the infinity point |
There was a problem hiding this comment.
Why this case? Can we not simply use absorb_commitment, please?
There was a problem hiding this comment.
we could, but I was not sure about its implementation, and is one extra thing to duplicate in IVC, but I have no issue with changing it if you think is better.
There was a problem hiding this comment.
another option is to change the signature to allow only one field, and leave it up to the trait implementation to decompose and represent the points in the native field.
folding/src/lib.rs
Outdated
| @@ -237,6 +237,10 @@ impl<'a, CF: FoldingConfig> FoldingScheme<'a, CF> { | |||
| fq_sponge.absorb_g(&error_commitments[0].elems); | |||
| fq_sponge.absorb_g(&error_commitments[1].elems); | |||
|
|
|||
| let to_absorb = env.to_absorb(); | |||
There was a problem hiding this comment.
Can you indicate the order of the absorption in the documentation, please? We need to be consistent for the IVC circuit, and it is good to document it.
We should have a test checking the order is in line with the documentation.
There was a problem hiding this comment.
Note that the typed transcript could be used, see #2094.
The idea is that every "round" (compute_error, compute_extended, etc) sends a message type, that can be absorb.
Each function returns a type "Message", and the sponge gets to absorb a heterogeneous list of "absorbable messages".
The heterogeneous list can be implemented as Nil, Cons constructors.
But I suppose it will be for the future. For now, we can keep it simple, and write in comments the order.
|
Any update @fabrizio-m? |
folding/src/instance_witness.rs
Outdated
| @@ -82,6 +95,16 @@ impl<G: CommitmentCurve, I: Instance<G>> ExtendedInstance<G, I> { | |||
| extended: vec![], | |||
| } | |||
| } | |||
| /// wraps the inner absorb, adding its own commitments | |||
There was a problem hiding this comment.
| /// wraps the inner absorb, adding its own commitments | |
| /// Return the elements to be absorbed by the sponge | |
| /// | |
| /// The commitments to the additional columns created by quadriticization | |
| /// are appended to the existing commitments of the initial instance | |
| /// to be absorbed. The scalars are unchanged. |
Co-authored-by: Danny Willems <be.danny.willems@gmail.com>
It might be dangerous if a user is not aware that it must provide an implementation for it. By not providing it, the user will be notified at compile time.
It should not be called directly
Always grumpy...
Add comments to folding documentation from meeting
No description provided.