SNARK Light Client

[gakonst]

https://github.com/shresthagrawal/eth-pos-superlight-client/blob/master/notes/snark-nipopos.md just saw this - maybe makes sense to take a look at the code written here https://github.com/celo-org/celo-bls-snark-rs, where we build a snark over each validator set rotation which verifies a bls aggregate signature from the previous validator set on the next validator set

[shresthagrawal]

Hey @gakonst,
I had a quick look and it seems that the BLS curve used in the implementation is different than we need for ETH PoS. How difficult is it to change the curve used in the implementation? Conceptually the SNARK approach to building Superlight clients for Ethereum PoS is clear. The main missing component is the SNARK implementation for the BLS12-381 aggregate signature check. There is currently an ongoing initiative to build a Circom library for BLS12-381. Having a progressive SNARK would be a bonus.

Due to the blocker of SNARK implementation, the current implementation for the Superlight client uses interactive bisection games similar to the protocol mentioned in the Light clients for Lazy Blockchain Paper. This approach still allows syncing in Poly log complexity which is already exponentially better than the official Ethereum PoS Light Client specifications

I will go through the Celo implementation more thoroughly later.

On a side note, I am a big admirer of your work with Foundry and Paradigm. I am surprised that you reached this repo so quickly as it was open-sourced yesterday.

[shresthagrawal]

Seems that there have been separate efforts by the people from 0xPARC in collaboration with GNOSIS DAO to build the light client based on ZKSNARK
https://forum.gnosis.io/t/gip-57-should-gnosis-dao-support-research-of-a-zksnark-enabled-light-client-and-bridge/5421