zorch.prove

Scheme-agnostic multi-round proving driver.

fold_rounds runs a folding Round a fixed number of times, threading the (state, transcript) carry and collecting each round's message into a list — the state and message types are opaque to the driver, so any scheme reuses it: a Basefold round (univariate + commitment message), a multilinear sumcheck round, a future univariate / FFT round. It stays a Python loop: its per-round message shapes need not be round-invariant, so it is not lax.scan-shaped.

The homogeneous multilinear sumcheck specialization — one lax.scan over the hypercube variables, with the optional zorch.sumcheck register-resident marker — lives in zorch.sumcheck.prover, not here: it is multilinear-specific, where this driver is basis-agnostic.

fold_rounds

fold_rounds(
    rnd: ProverRound[Any, Any, TranscriptT],
    state: Any,
    transcript: TranscriptT,
    rounds: int,
) -> tuple[Any, TranscriptT, list[Any]]

Run rnd exactly rounds times; return (state, transcript, list[msg]).

Source code in zorch/prove.py

def fold_rounds(
    rnd: ProverRound[Any, Any, TranscriptT],
    state: Any,
    transcript: TranscriptT,
    rounds: int,
) -> tuple[Any, TranscriptT, list[Any]]:
    """Run `rnd` exactly `rounds` times; return (state, transcript, list[msg])."""
    msgs: list[Any] = []
    for _ in range(rounds):
        state, transcript, msg = rnd(state, transcript)
        msgs.append(msg)
    return state, transcript, msgs