David Wong

ProofEvaluations is too generic

9

```rust pub struct ProofEvaluations { /// witness polynomials pub w: [Field; COLUMNS], /// permutation polynomial pub z: Field, /// permutation polynomials /// (PERMUTS-1 evaluations because the last permutation is only...

delete mina-curves

6

I believe we should be able to delete the mina-curves crate and use pallas/vesta from https://github.com/arkworks-rs/curves thanks to @mmaker for the pointer

Should we move T: Field to T: PrimeField everywhere?

3

Since kimchi is built exclusively for prime fields (and not for extension fields, for example), it might make sense to move to `PrimeField` everywhere instead of having `Field`, `FftField`, and...

[kimchi] absorbing all the values!

14

in [plonk_sponge.rs](https://github.com/o1-labs/proof-systems/blob/master/dlog/kimchi/src/plonk_sponge.rs#L38), it's too easy to forget to absorb some of the evaluation values. So here we should discuss: - is this really needed? It's not clear to me, as...

Handling optional stuff in a proof

3

I've noticed that we handle optional things in a proof not super safely. For example, in `verify` for lookups we do: ```rust let joint_combiner = if let Some(l) = &index.lookup_index...

[easy] remove dead code

6

it seems like we were using these functions (ToBytes, FromBytes) and traits (num-derive) before we moved to serde for serialization. So removing this code now.

In `utils/src/chunked_polynomial.rs`, the `evaluate_chunks` function returns a vector of field elements: ```rust impl ChunkedPolynomial { /// This function evaluates polynomial in chunks. pub fn evaluate_chunks(&self, elm: F) -> Vec {...