Ability to set seed, number of rounds and exponent in MIMC hash

[shreyas-londhe]

Referencing to https://github.com/ConsenSys/gnark-crypto/blob/master/ecc/bn254/fr/mimc/mimc.go

There should be a way to set mimcNbRounds, seed and exponent (which is currently set to 5).
I find that this modularity would help us get more control over the hash.

I was using the snark to verify on ethereum and the mimc implementation on solidity was using a different seed and exponent value, so I had to change it manually in the go modules which I think is not a great way to do things.

If I get an idea of how I should implement this issue, I would love to make a PR to solve this.

Thanks!

[ThomasPiellard]

Hi @shryasss , where does the 7 come from? Normally for BN254, the smallest permutation is x -> x^5. Ok for a PR to change the number of rounds, etc. But one advantage of hard coding the exponent is that the permutation is slightly more efficient because we don't need to do a double&add algo but rather we can directly write the shortest addition chain.

[shreyas-londhe]

So you are saying that having exponent 5 is efficient? but I have seen some libraries for solidity which use the mimc7 so I thought that should also be possible using gnark.

Even iden3 has a mimc7 implementation.

you can check it here - https://github.com/iden3/go-iden3-crypto/blob/master/mimc7/mimc7.go

[gbotrel]

@shryasss parametrizing MiMC could make sense. If you want to take a shot at it, let me know. I think it's a good opportunity to experiment with generic. Recently in gnark we did:
https://github.com/ConsenSys/gnark/blob/develop/std/math/emulated/params.go

Which enables a clean usage across in-circuit / out-of-circuit:

secp256k1, err := emulated.NewField[emulated.Secp256k1](api)

Which would translate similarly to MiMC instantiation.

[shreyas-londhe]

Hey @gbotrel! I would love to take this up and make a PR, please let me know on how to proceed on this.