0xmountaintop

#962 is kind of ready. If anyone can review it will be the best!

by comparing u mean having an article? and in what aspects? some ideas come up to my mind if they r helpful: + security assumption, curve params + (benchmark) performance...

halo2 uses a different commitment scheme from the original plonk paper. and the verifier smart contract will be different

The tests seem failed. However it works good when I run "make test_unit_race" and "make test_unit" locally.

It will also be better if "LocalPartySaveData" can be abstract a bit more. Currently common.SignatureData and tss.Party are already abstract, so different type signing party can be started according to...

The tests seem failed. However it works good when I run "make test_unit_race" and "make test_unit" locally.

This is also claimed in the paper. p8 

a bit more of the context: 1. we switch to pairing bn256 instead of using pasta 2. the reported error is `virtual selectors are removed during optimization`

> the only thing necessary to reproduce the proving key is the code of the circuit. Since that's necessary for proving anyway, I don't think there's an advantage to reading...

We are working on https://github.com/appliedzkp/zkevm-circuits. We haven't had a performance number yet, but the k used in production I assume will be >=20. (The k for "evm_circuit" we are using...