Caching G2 points

[ggkitsas]

Caching $[\lambda*b] Z_{i,\omega_i}$ for some 2/3 weight subset of signers will benefit performance in the cases where the next block also has that same 2/3 weight subset.