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
For simplicity we now use a unique alpha (subrelation batching challenge) per instance but will not raise it up to NUM_SUBRELATION - 1 power . This is sowe don't blow up the degree of combiner and compilation time. The plan is to generate one alpha per subrelation so batching those only takes the degree of the combiner from dk to (d + k)*d
The text was updated successfully, but these errors were encountered:
This PR introduces the Protogalaxy decider, now having a complete proof
of concept of the folding protocol for Ultra instances. To enable this,
sumcheck has been modified to work with the same power polynomial as in
protogalaxy paper and can be initialised with a custom `target_sum`.
Moreover, we now use different batching challenges for each subrelation
in sumcheck, essential to ensure a reasonable degree of the combiner
polynomial.
Additional: Minor bug fixed in the computation of \vec{\beta*}.
ResolvesAztecProtocol/barretenberg#772.
ResolvesAztecProtocol/barretenberg#691.
This PR introduces the Protogalaxy decider, now having a complete proof
of concept of the folding protocol for Ultra instances. To enable this,
sumcheck has been modified to work with the same power polynomial as in
protogalaxy paper and can be initialised with a custom `target_sum`.
Moreover, we now use different batching challenges for each subrelation
in sumcheck, essential to ensure a reasonable degree of the combiner
polynomial.
Additional: Minor bug fixed in the computation of \vec{\beta*}.
Resolves#772.
Resolves#691.
)
This PR introduces the Protogalaxy decider, now having a complete proof
of concept of the folding protocol for Ultra instances. To enable this,
sumcheck has been modified to work with the same power polynomial as in
protogalaxy paper and can be initialised with a custom `target_sum`.
Moreover, we now use different batching challenges for each subrelation
in sumcheck, essential to ensure a reasonable degree of the combiner
polynomial.
Additional: Minor bug fixed in the computation of \vec{\beta*}.
ResolvesAztecProtocol/barretenberg#772.
ResolvesAztecProtocol/barretenberg#691.
For simplicity we now use a unique
alpha
(subrelation batching challenge) per instance but will not raise it up toNUM_SUBRELATION - 1
power . This is sowe don't blow up the degree of combiner and compilation time. The plan is to generate one alpha per subrelation so batching those only takes the degree of the combiner fromdk
to(d + k)*d
The text was updated successfully, but these errors were encountered: