You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, we carry all data across AES folds via a large quantity of public inputs. This approach is not scalable and requires a messy abstraction for upstream users.
Instead, we should explore a committed data structure (merkle tree, poly commit, offline memory checking, etc) and open it at the relevant points in subsequent circuits. This will introduce a constant circuit overhead to each fold but will preserve the IVC that the current approach lacks.
Currently, we carry all data across AES folds via a large quantity of public inputs. This approach is not scalable and requires a messy abstraction for upstream users.
Instead, we should explore a committed data structure (merkle tree, poly commit, offline memory checking, etc) and open it at the relevant points in subsequent circuits. This will introduce a constant circuit overhead to each fold but will preserve the IVC that the current approach lacks.
The current approach is commented out here, but was temporarily used in the upstream circuits
The text was updated successfully, but these errors were encountered: