openvm_stark_backend::prover::stacked_reduction

Trait StackedReductionProver

pub trait StackedReductionProver<'a, PB: ProverBackend, PD> {
    // Required methods
    fn new(
        device: &'a PD,
        stacked_per_commit: Vec<&'a PB::PcsData>,
        need_rot_per_commit: Vec<Vec<bool>>,
        r: &[PB::Challenge],
        lambda: PB::Challenge,
    ) -> Self;
    fn batch_sumcheck_uni_round0_poly(
        &mut self,
    ) -> UnivariatePoly<PB::Challenge>;
    fn fold_ple_evals(&mut self, u_0: PB::Challenge);
    fn batch_sumcheck_poly_eval(
        &mut self,
        round: usize,
        u_prev: PB::Challenge,
    ) -> [PB::Challenge; 2];
    fn fold_mle_evals(&mut self, round: usize, u_round: PB::Challenge);
    fn into_stacked_openings(self) -> Vec<Vec<PB::Challenge>>;
}

Expand description

Helper trait for proving the reduction of column opening claims and column rotation opening claims to opening claims of column polynomials of the stacked matrix.

Returns the reduction proof and the random vector u of length 1 + n_stack.

Required Methods§

Source

fn new( device: &'a PD, stacked_per_commit: Vec<&'a PB::PcsData>, need_rot_per_commit: Vec<Vec<bool>>, r: &[PB::Challenge], lambda: PB::Challenge, ) -> Self

We only provide a view to the stacked PcsData per commitment because the WHIR prover will still use the PLE evaluations of the stacked matrices later. The order of stacked_per_commit is common_main, preprocessed for trace_idx=0 (if any), cached_0 for trace_idx=0, ..., preprocessed for trace_idx=1 (if any), ....

The lambda is the batching randomness for the batch sumcheck.

Source