Struct OperandClaims 

pub struct OperandClaims<F: Field> {
    pub r_x: Vec<F>,
    pub r_zhat_prime: F,
    pub evals: Vec<F>,
}

The operand evaluation claims of every operation, as the shift reduction receives them.

Fields

r_x: Vec<F>

The constraint point, as long as the widest operation’s constraint count.

Every operation is claimed at the whole point, its matrix padded with empty rows; see [padding_scales].

r_zhat_prime: F

The univariate challenge folding the bit axis, shared by every operation.

evals: Vec<F>

One evaluation per operand column, the four operations’ runs in [OPERATION_ARITIES] order.

Implementations

impl<F: Field> OperandClaims<F>

pub fn from_rerand( cs: &ConstraintSystem, log_instances: usize, z_challenge: F, rerand: &RerandOutput<F>, sample: impl FnMut() -> F, ) -> Self

Assembles the claims from the output of the BitAnd sumcheck.

The sumcheck’s point is r_rho || r_x_star, instance index low, constraint index high. r_x_star spans the widest AND, IMUL and BMUL set. The constraint point r_x extends it through [zero::reduction_point] when the ZERO set is wider still, drawing the extra challenges from sample.

The sumcheck claims each operation at its own prefix of r_x. Its padding factor lifts the claim to its padded matrix at the whole point.

An operation the constraint system does not use gets zero evaluations.

Arguments

• cs: the constraint system, whose row counts pick each operation’s padding factor.
• log_instances: the instance variables at the low end of the point.
• z_challenge: the univariate challenge, shared by every operation.
• rerand: the sumcheck’s output, with the IntMul operand evaluations before the BinMul ones.
• sample: draws the constraint point’s extra challenges from the transcript.

pub fn prepare( self, channel: &mut impl IPProverChannel<F>, ) -> PreparedOperandClaims<F>

Draws the batching challenges and folds their weights into the claims.

The claim of column m is weighted by the equality indicator of the operand axis at m. The axis is padded to a cube of log2_ceil(evals.len()) challenges; the columns past the last claim name nothing. The verifier draws the same challenges at the same place, so the count is protocol, not detail.

Arguments

• channel: the transcript the batching challenges are drawn from.

Trait Implementations

impl<F: Clone + Field> Clone for OperandClaims<F>

fn clone(&self) -> OperandClaims<F>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<F: Debug + Field> Debug for OperandClaims<F>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.