Closed Sword-Smith closed 2 months ago
How about adding BFieldCodec
support for Polynomial
instead? That seems like the more general approach to me.
How about adding
BFieldCodec
support forPolynomial
instead? That seems like the more general approach to me.
That might very well be a more robust solution, yes.
Superseeded by 7367c677a53487825d7afc101066a16a83ae7227
The last FRI polynomial is not added to the proof stream in a normalized manner, making the degree-check (in Triton VM/TASM) of the last round harder than it should be. Currently, we do:
Which adds a list of coefficients to the proof stream which will have trailing zeros. A potential solution is to change the above-function to:
This, combined with a test that the last polynomial is always shared in its normalized form, should solve the problem. The alternative is to write a degree-checker in Triton VM assembly but that seems needlessly complicated and will also lead to larger-than-necessary proofs.