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
Here is a potential space and speed improvement: do not store the randomized trace polynomials but store the unrandomized trace polynomials instead.
For starters, this gives faster interpolation. The domain size is half.
To randomize the polynomial in monomial-coefficient form, select a vector of k uniformly random field elements, and add this vector to the vector of coefficients in slices [0:k] and [N:N+k]. This corresponds to adding a multiple of $X^N+1$ where the cofactor consists of k uniformly random coefficients. By construction, this term disappears on the trace domain (whose length is N.)
The next step is to not even bother storing the k randomizer coefficients for every column -- instead, derive them pseudorandomly from a random seed and the column index. You only need the full polynomial when low-degree extending; afterwards the k randomizer coefficients can be thrown away again to save space.
The text was updated successfully, but these errors were encountered:
Here is a potential space and speed improvement: do not store the randomized trace polynomials but store the unrandomized trace polynomials instead.
For starters, this gives faster interpolation. The domain size is half.
To randomize the polynomial in monomial-coefficient form, select a vector of$X^N+1$ where the cofactor consists of
k
uniformly random field elements, and add this vector to the vector of coefficients in slices[0:k]
and[N:N+k]
. This corresponds to adding a multiple ofk
uniformly random coefficients. By construction, this term disappears on the trace domain (whose length isN
.)The next step is to not even bother storing the
k
randomizer coefficients for every column -- instead, derive them pseudorandomly from a random seed and the column index. You only need the full polynomial when low-degree extending; afterwards thek
randomizer coefficients can be thrown away again to save space.The text was updated successfully, but these errors were encountered: