Expand description
The BitAnd sumcheck, batched with the multiplication reductions’ operand-column MLE-checks.
After the univariate skip, the BitAnd claim g(z) is the MLE-check
g(z) = sum_x eq(x, r_x) * (A(z, x) * B(z, x) - C(z, x))at the zerocheck point r_x. Each operand column of an IntMul or BinMul reduction carries one
claim per bit at that reduction’s own point rho. The Lagrange weights at z fold them to one
MLE-check per column, alpha = sum_x eq(x, rho) * Z(z, x), with no transcript traffic.
One sumcheck in plain form proves all of them, batched by powers of one challenge theta,
BitAnd first. It runs over the longest summand’s variables. A shorter summand is zero-padded on
its high variables, so its point is a prefix of the unified point r*.
The round polynomials have degree 3: BitAnd’s composition has degree 2, and its equality indicator adds one.
Structs§
- Operand
Claims - One reduction’s per-bit evaluation claims on its operand columns, at that reduction’s point.
- Rerand
Output - The evaluation claims the batched sumcheck reduces to, all at prefixes of one point.
Constants§
- SUMCHECK_
DEGREE - Degree of the batched round polynomials: the BitAnd summand’s prime degree 2 plus its indicator.
Functions§
- verify
- Verifies the BitAnd sumcheck batched with the operand-column MLE-checks.