zorch.pcs.ligero.config

Single-shot Ligero proof type. Co-located with the scheme like BasefoldProof in basefold/config.py — there is no shared proof.py in pcs/.

Ligero opens one committed multilinear f, laid out as a rows x cols matrix X̃ (rows = code.message_len), at a point z = (z_row, z_col). The prover sends w = X̃ · r_col in the clear (r_col = eq(z_col)); the verifier checks proximity <X[s], r_col> = encode(w)[s] on a few opened codeword rows and value <r_row, w> = y. That single sent vector is Ligero's sqrt(N) proof cost — the recursion in pcs/ligerito is exactly what removes it.

LigeroProof dataclass

One single-shot Ligero opening proof.

the sent vector X̃ · r_col, length rows (= code.message_len). Its

sqrt(N) size is Ligero's proof-cost bottleneck.

component_opening: the codeword rows opened at the sampled query positions (row is [Q, cols]); the verifier dots each with r_col for the proximity left-hand side and Merkle-checks it against the commitment.

Source code in zorch/pcs/ligero/config.py

@partial(
    frx.tree_util.register_dataclass,
    data_fields=["w", "component_opening"],
    meta_fields=[],
)
@dataclass(frozen=True)
class LigeroProof:
    """One single-shot Ligero opening proof.

    w: the sent vector `X̃ · r_col`, length `rows` (`= code.message_len`). Its
        `sqrt(N)` size is Ligero's proof-cost bottleneck.
    component_opening: the codeword rows opened at the sampled query positions
        (`row` is `[Q, cols]`); the verifier dots each with `r_col` for the
        proximity left-hand side and Merkle-checks it against the commitment.
    """

    w: Array
    component_opening: Opening