p3_challenger/

duplex_challenger.rs

use alloc::vec;
use alloc::vec::Vec;
use core::error::Error;
use core::fmt::{Display, Formatter};

use p3_field::{BasedVectorSpace, Field, PrimeField, PrimeField64};
use p3_monty_31::{MontyField31, MontyParameters};
use p3_symmetric::{CryptographicPermutation, Hash, MerkleCap};

use crate::{
    CanFinalizeDigest, CanObserve, CanSample, CanSampleBits, CanSampleUniformBits, FieldChallenger,
};

/// A generic duplex sponge challenger over a finite field, used for generating deterministic
/// challenges from absorbed inputs.
///
/// This structure implements a duplex sponge that alternates between:
/// - Absorbing inputs into the sponge state,
/// - Applying a cryptographic permutation over the state,
/// - Squeezing outputs from the state as challenges.
///
/// The sponge operates over a state of `WIDTH` elements, divided into:
/// - A rate of `RATE` elements (the portion exposed to input/output),
/// - A capacity of `WIDTH - RATE` elements (the hidden part ensuring security).
///
/// The challenger buffers observed inputs until the rate is full, applies the permutation,
/// and then produces challenge outputs from the permuted state. It supports:
/// - Observing single values, arrays, hashes, or nested vectors,
/// - Sampling fresh challenges as field elements or bitstrings.
#[derive(Clone, Debug)]
pub struct DuplexChallenger<F, P, const WIDTH: usize, const RATE: usize>
where
    F: Clone,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    /// The internal sponge state, consisting of `WIDTH` field elements.
    ///
    /// The first `RATE` elements form the rate section, where input values are absorbed
    /// and output values are squeezed.
    /// The remaining `WIDTH - RATE` elements form the capacity, which provides hidden
    /// entropy and security against attacks.
    pub sponge_state: [F; WIDTH],

    /// A buffer holding field elements that have been observed but not yet absorbed.
    ///
    /// Inputs added via `observe` are collected here.
    /// Once the buffer reaches `RATE` elements, the sponge performs a duplexing step:
    /// it absorbs the inputs into the state and applies the permutation.
    pub input_buffer: Vec<F>,

    /// A buffer holding field elements that have been squeezed from the sponge state.
    ///
    /// Outputs are produced by `duplexing` and stored here.
    /// Calls to `sample` or `sample_bits` pop values from this buffer.
    /// When the buffer is empty (or new inputs were absorbed), a new duplexing step is triggered.
    pub output_buffer: Vec<F>,

    /// The cryptographic permutation applied to the sponge state.
    ///
    /// This permutation must provide strong pseudorandomness and collision resistance,
    /// ensuring that squeezed outputs are indistinguishable from random and securely
    /// bound to the absorbed inputs.
    pub permutation: P,
}

impl<F, P, const WIDTH: usize, const RATE: usize> DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    pub fn new(permutation: P) -> Self
    where
        F: Default,
    {
        const {
            assert!(RATE > 0 && RATE < WIDTH);
        }
        Self {
            sponge_state: [F::default(); WIDTH],
            input_buffer: vec![],
            output_buffer: vec![],
            permutation,
        }
    }

    pub(crate) fn duplexing(&mut self) {
        assert!(self.input_buffer.len() <= RATE);

        // Overwrite the first r elements with the inputs.
        for (i, val) in self.input_buffer.drain(..).enumerate() {
            self.sponge_state[i] = val;
        }

        // Apply the permutation.
        self.permutation.permute_mut(&mut self.sponge_state);

        self.output_buffer.clear();
        self.output_buffer.extend(&self.sponge_state[..RATE]);
    }
}

impl<F, P, const WIDTH: usize, const RATE: usize> DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy + Default + PrimeField,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    /// Absorb `values.len()` elements into the rate, zero remaining rate slots, add `length_tag`
    /// to `sponge_state[RATE]`, permute, and refill `output_buffer`.
    ///
    /// Clears `input_buffer` and `output_buffer` before absorbing. Used by
    /// [`MultiField32Challenger`](crate::MultiField32Challenger) so packed scalar and native-digest
    /// absorbs share this [`DuplexChallenger`]'s sponge state without queuing through `observe`.
    pub fn absorb_rate_padded_with_tag(&mut self, values: &[F], length_tag: u8) {
        const {
            assert!(
                RATE < WIDTH,
                "RATE must be less than WIDTH for capacity length slot"
            );
        }
        assert!(values.len() <= RATE);
        self.input_buffer.clear();
        self.output_buffer.clear();
        for (i, &v) in values.iter().enumerate() {
            self.sponge_state[i] = v;
        }
        self.sponge_state[values.len()..RATE].fill(F::ZERO);
        self.sponge_state[RATE] += F::from_u8(length_tag);
        self.permutation.permute_mut(&mut self.sponge_state);
        self.output_buffer
            .extend_from_slice(&self.sponge_state[..RATE]);
    }
}

impl<F, P, const WIDTH: usize, const RATE: usize> FieldChallenger<F>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: PrimeField64,
    P: CryptographicPermutation<[F; WIDTH]>,
{
}

impl<F, P, const WIDTH: usize, const RATE: usize> CanObserve<F>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, value: F) {
        // Any buffered output is now invalid.
        self.output_buffer.clear();

        self.input_buffer.push(value);

        if self.input_buffer.len() == RATE {
            self.duplexing();
        }
    }
}

impl<F, P, const N: usize, const WIDTH: usize, const RATE: usize> CanObserve<[F; N]>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, values: [F; N]) {
        for value in values {
            self.observe(value);
        }
    }
}

impl<F, P, const N: usize, const WIDTH: usize, const RATE: usize> CanObserve<Hash<F, F, N>>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, values: Hash<F, F, N>) {
        for value in values {
            self.observe(value);
        }
    }
}

impl<F, P, const N: usize, const WIDTH: usize, const RATE: usize> CanObserve<&MerkleCap<F, [F; N]>>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, cap: &MerkleCap<F, [F; N]>) {
        for digest in cap.roots() {
            for value in digest {
                self.observe(*value);
            }
        }
    }
}

impl<F, P, const N: usize, const WIDTH: usize, const RATE: usize> CanObserve<MerkleCap<F, [F; N]>>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, cap: MerkleCap<F, [F; N]>) {
        self.observe(&cap);
    }
}

// for TrivialPcs
impl<F, P, const WIDTH: usize, const RATE: usize> CanObserve<Vec<Vec<F>>>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn observe(&mut self, valuess: Vec<Vec<F>>) {
        for values in valuess {
            for value in values {
                self.observe(value);
            }
        }
    }
}

impl<F, EF, P, const WIDTH: usize, const RATE: usize> CanSample<EF>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Field,
    EF: BasedVectorSpace<F>,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn sample(&mut self) -> EF {
        EF::from_basis_coefficients_fn(|_| {
            // If we have buffered inputs, we must perform a duplexing so that the challenge will
            // reflect them. Or if we've run out of outputs, we must perform a duplexing to get more.
            if !self.input_buffer.is_empty() || self.output_buffer.is_empty() {
                self.duplexing();
            }

            self.output_buffer
                .pop()
                .expect("Output buffer should be non-empty")
        })
    }
}

impl<F, P, const WIDTH: usize, const RATE: usize> CanSampleBits<usize>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: PrimeField64,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    /// The sampled bits are not perfectly uniform, but we can bound the error: every sequence
    /// appears with probability 1/p-close to uniform (1/2^b).
    ///
    /// Proof:
    /// We denote p = F::ORDER_U64, and b = `bits`.
    /// If X follows a uniform distribution over F, if we consider the first b bits of X, each
    /// sequence appears either with probability P1 = ⌊p / 2^b⌋ / p or P2 = (1 + ⌊p / 2^b⌋) / p.
    /// We have 1/2^b - 1/p ≤ P1, P2 ≤ 1/2^b + 1/p
    fn sample_bits(&mut self, bits: usize) -> usize {
        assert!(bits < (usize::BITS as usize));
        assert!((1 << bits) < F::ORDER_U64);
        let rand_f: F = self.sample();
        let rand_usize = rand_f.as_canonical_u64() as usize;
        rand_usize & ((1 << bits) - 1)
    }
}

/// Trait for fields that support uniform bit sampling optimizations
pub trait UniformSamplingField {
    /// Maximum number of bits we can sample at negligible (~1/field prime) probability of
    /// triggering an error / requiring a resample.
    const MAX_SINGLE_SAMPLE_BITS: usize;
    /// An array storing the largest value `m_k` for each `k` in [0, 31], such that `m_k`
    /// is a multiple of `2^k` and less than P. `m_k` is defined as:
    ///
    /// \( m_k = ⌊P / 2^k⌋ · 2^k \)
    ///
    /// This is used as a rejection sampling threshold (or error trigger), when sampling
    /// random bits from uniformly sampled field elements. As long as we sample up to the `k`
    /// least significant bits in the range [0, m_k), we sample from exactly `m_k` elements. As
    /// `m_k` is divisible by 2^k, each of the least significant `k` bits has exactly the same
    /// number of zeroes and ones, leading to a uniform sampling.
    const SAMPLING_BITS_M: [u64; 64];
}

// Provide a blanket implementation for Monty31 fields here, which forwards the
// implementation of the variables to the generic argument `<Field>Parameter`,
// for which we implement the trait (KoalaBear, BabyBear).
impl<MP> UniformSamplingField for MontyField31<MP>
where
    MP: UniformSamplingField + MontyParameters,
{
    const MAX_SINGLE_SAMPLE_BITS: usize = MP::MAX_SINGLE_SAMPLE_BITS;
    const SAMPLING_BITS_M: [u64; 64] = MP::SAMPLING_BITS_M;
}

// Set of different strategies we currently support for sampling
// Implementations for each are below.
/// A zero-sized struct representing the "resample" strategy.
pub(super) struct ResampleOnRejection;
/// A zero-sized struct representing the "error" strategy.
pub(super) struct ErrorOnRejection;

/// Custom error raised when resampling is required for uniform bits but disabled
/// via `ErrorOnRejection` strategy.
#[derive(Debug)]
pub struct ResamplingError {
    /// The sampled value
    value: u64,
    /// The target value we need to be smaller than
    m: u64,
}

impl Display for ResamplingError {
    fn fmt(&self, f: &mut Formatter<'_>) -> core::fmt::Result {
        write!(
            f,
            "Encountered value {0}, which requires resampling for uniform bits as it not smaller than {1}. But resampling is not enabled.",
            self.value, self.m
        )
    }
}

impl Error for ResamplingError {}

/// A trait that defines a strategy for handling out-of-range samples.
pub(super) trait BitSamplingStrategy<F, P, const W: usize, const R: usize>
where
    F: PrimeField64,
    P: CryptographicPermutation<[F; W]>,
{
    /// Whether to error instead of resampling when a drawn value is too large.
    const ERROR_ON_REJECTION: bool;

    #[inline]
    fn sample_value(
        challenger: &mut DuplexChallenger<F, P, W, R>,
        m: u64,
    ) -> Result<F, ResamplingError> {
        let mut result: F = challenger.sample();
        if Self::ERROR_ON_REJECTION {
            if result.as_canonical_u64() >= m {
                return Err(ResamplingError {
                    value: result.as_canonical_u64(),
                    m,
                });
            }
        } else {
            while result.as_canonical_u64() >= m {
                result = challenger.sample();
            }
        }
        Ok(result)
    }
}

/// Implement rejection sampling
impl<F, P, const W: usize, const R: usize> BitSamplingStrategy<F, P, W, R> for ResampleOnRejection
where
    F: PrimeField64,
    P: CryptographicPermutation<[F; W]>,
{
    const ERROR_ON_REJECTION: bool = false;
}

/// Implement erroring on a required rejection
impl<F, P, const W: usize, const R: usize> BitSamplingStrategy<F, P, W, R> for ErrorOnRejection
where
    F: PrimeField64,
    P: CryptographicPermutation<[F; W]>,
{
    const ERROR_ON_REJECTION: bool = true;
}

impl<F, P, const WIDTH: usize, const RATE: usize> DuplexChallenger<F, P, WIDTH, RATE>
where
    F: UniformSamplingField + PrimeField64,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    /// Generic implementation for uniform bit sampling, parameterized by a strategy.
    #[inline]
    fn sample_uniform_bits_with_strategy<S>(
        &mut self,
        bits: usize,
    ) -> Result<usize, ResamplingError>
    where
        S: BitSamplingStrategy<F, P, WIDTH, RATE>,
    {
        if bits == 0 {
            return Ok(0);
        };
        assert!(bits < usize::BITS as usize, "bit count must be valid");
        assert!(
            (1u64 << bits) < F::ORDER_U64,
            "bit count exceeds field order"
        );
        let m = F::SAMPLING_BITS_M[bits];
        if bits <= F::MAX_SINGLE_SAMPLE_BITS {
            // Fast path: Only one sample is needed for sufficient uniformity.
            let rand_f = S::sample_value(self, m);
            Ok(rand_f?.as_canonical_u64() as usize & ((1 << bits) - 1))
        } else {
            // Slow path: Sample twice to construct the required number of bits.
            // This reduces the bias introduced by a single, larger sample.
            let half_bits1 = bits / 2;
            let half_bits2 = bits - half_bits1;
            // Sample the first chunk of bits.
            let rand1 = S::sample_value(self, F::SAMPLING_BITS_M[half_bits1]);
            let chunk1 = rand1?.as_canonical_u64() as usize & ((1 << half_bits1) - 1);
            // Sample the second chunk of bits.
            let rand2 = S::sample_value(self, F::SAMPLING_BITS_M[half_bits2]);
            let chunk2 = rand2?.as_canonical_u64() as usize & ((1 << half_bits2) - 1);

            // Combine the chunks.
            Ok(chunk1 | (chunk2 << half_bits1))
        }
    }
}

impl<F, P, const WIDTH: usize, const RATE: usize> CanSampleUniformBits<F>
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: UniformSamplingField + PrimeField64,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    fn sample_uniform_bits<const RESAMPLE: bool>(
        &mut self,
        bits: usize,
    ) -> Result<usize, ResamplingError> {
        if RESAMPLE {
            self.sample_uniform_bits_with_strategy::<ResampleOnRejection>(bits)
        } else {
            self.sample_uniform_bits_with_strategy::<ErrorOnRejection>(bits)
        }
    }
}

impl<F, P, const WIDTH: usize, const RATE: usize> CanFinalizeDigest
    for DuplexChallenger<F, P, WIDTH, RATE>
where
    F: Copy,
    P: CryptographicPermutation<[F; WIDTH]>,
{
    type Digest = [F; RATE];

    fn finalize(mut self) -> [F; RATE] {
        // Unconditionally duplex: absorb any pending input and permute.
        //
        // Note: sampling only pops from the output buffer without modifying
        // sponge state, so it does not necessarily affect the digest (e.g.
        // when the last observe already triggered auto-duplexing).
        self.duplexing();
        self.sponge_state[..RATE].try_into().unwrap()
    }
}

#[cfg(test)]
mod tests {
    use core::iter;

    use p3_baby_bear::BabyBear;
    use p3_field::PrimeCharacteristicRing;
    use p3_field::extension::BinomialExtensionField;
    use p3_goldilocks::Goldilocks;
    use p3_symmetric::Permutation;

    use super::*;
    use crate::grinding_challenger::GrindingChallenger;

    const WIDTH: usize = 24;
    const RATE: usize = 16;

    type G = Goldilocks;
    type EF2G = BinomialExtensionField<G, 2>;

    type BB = BabyBear;

    #[derive(Clone)]
    struct TestPermutation {}

    impl<F: Clone> Permutation<[F; WIDTH]> for TestPermutation {
        fn permute_mut(&self, input: &mut [F; WIDTH]) {
            input.reverse();
        }
    }

    impl<F: Clone> CryptographicPermutation<[F; WIDTH]> for TestPermutation {}

    #[test]
    fn test_duplex_challenger() {
        type Chal = DuplexChallenger<G, TestPermutation, WIDTH, RATE>;
        let permutation = TestPermutation {};
        let mut duplex_challenger = DuplexChallenger::new(permutation);

        // Observe 12 elements.
        (0..12).for_each(|element| duplex_challenger.observe(G::from_u8(element as u8)));

        let state_after_duplexing: Vec<_> = iter::repeat_n(G::ZERO, 12)
            .chain((0..12).map(G::from_u8).rev())
            .collect();

        let expected_samples: Vec<G> = state_after_duplexing[..16].iter().copied().rev().collect();
        let samples = <Chal as CanSample<G>>::sample_vec(&mut duplex_challenger, 16);
        assert_eq!(samples, expected_samples);
    }

    #[test]
    #[should_panic]
    fn test_duplex_challenger_sample_bits_security() {
        type GoldilocksChal = DuplexChallenger<G, TestPermutation, WIDTH, RATE>;
        let permutation = TestPermutation {};
        let mut duplex_challenger = GoldilocksChal::new(permutation);

        for _ in 0..100 {
            assert!(duplex_challenger.sample_bits(129) < 4);
        }
    }

    #[test]
    #[should_panic]
    fn test_duplex_challenger_sample_bits_security_small_field() {
        type BabyBearChal = DuplexChallenger<BB, TestPermutation, WIDTH, RATE>;
        let permutation = TestPermutation {};
        let mut duplex_challenger = BabyBearChal::new(permutation);

        for _ in 0..100 {
            assert!(duplex_challenger.sample_bits(40) < 1 << 31);
        }
    }

    #[test]
    #[should_panic]
    fn test_duplex_challenger_grind_security() {
        type GoldilocksChal = DuplexChallenger<G, TestPermutation, WIDTH, RATE>;
        let permutation = TestPermutation {};
        let mut duplex_challenger = GoldilocksChal::new(permutation);

        // This should cause sample_bits (and hence grind and check_witness) to
        // panic. If bit sizes were not constrained correctly inside the
        // challenger, (1 << too_many_bits) would loop around, incorrectly
        // grinding and accepting a 1-bit PoW.
        let too_many_bits = usize::BITS as usize;

        let witness = duplex_challenger.grind(too_many_bits);
        assert!(duplex_challenger.check_witness(too_many_bits, witness));
    }

    #[test]
    fn test_observe_single_value() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        chal.observe(G::from_u8(42));
        assert_eq!(chal.input_buffer, vec![G::from_u8(42)]);
        assert!(chal.output_buffer.is_empty());
    }

    #[test]
    fn test_observe_array_of_values() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        chal.observe([G::from_u8(1), G::from_u8(2), G::from_u8(3)]);
        assert_eq!(
            chal.input_buffer,
            vec![G::from_u8(1), G::from_u8(2), G::from_u8(3)]
        );
        assert!(chal.output_buffer.is_empty());
    }

    #[test]
    fn test_observe_hash_array() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        let hash = Hash::<G, G, 4>::from([G::from_u8(10); 4]);
        chal.observe(hash);
        assert_eq!(chal.input_buffer, vec![G::from_u8(10); 4]);
    }

    #[test]
    fn test_observe_nested_vecs() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        chal.observe(vec![
            vec![G::from_u8(1), G::from_u8(2)],
            vec![G::from_u8(3)],
        ]);
        assert_eq!(
            chal.input_buffer,
            vec![G::from_u8(1), G::from_u8(2), G::from_u8(3)]
        );
    }

    #[test]
    fn test_sample_triggers_duplex() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        chal.observe(G::from_u8(5));
        assert!(chal.output_buffer.is_empty());
        let _sample: G = chal.sample();
        assert!(!chal.output_buffer.is_empty());
    }

    #[test]
    fn test_sample_multiple_extension_field() {
        use p3_field::extension::BinomialExtensionField;
        type EF = BinomialExtensionField<G, 2>;
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        chal.observe(G::from_u8(1));
        chal.observe(G::from_u8(2));
        let _: EF = chal.sample();
        let _: EF = chal.sample();
    }

    #[test]
    fn test_sample_bits_within_bounds() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        for i in 0..RATE {
            chal.observe(G::from_u8(i as u8));
        }

        // With RATE=16 and input = 0..15, the reversed sponge_state will be 15..0
        // The first RATE elements of that, i.e. output_buffer, are 15..0
        // sample_bits(3) will sample the last of those: G::from_u8(0)

        let bits = 3;
        let value = chal.sample_bits(bits);
        let expected = G::ZERO.as_canonical_u64() as usize & ((1 << bits) - 1);
        assert_eq!(value, expected);
    }

    #[test]
    fn test_sample_bits_trigger_duplex_when_empty() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        // Force empty buffers
        assert_eq!(chal.input_buffer.len(), 0);
        assert_eq!(chal.output_buffer.len(), 0);

        // sampling bits should not panic, should return 0
        let bits = 2;
        let sample = chal.sample_bits(bits);
        let expected = G::ZERO.as_canonical_u64() as usize & ((1 << bits) - 1);
        assert_eq!(sample, expected);
    }

    #[test]
    fn test_output_buffer_pops_correctly() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // Observe RATE elements, causing a duplexing
        for i in 0..RATE {
            chal.observe(G::from_u8(i as u8));
        }

        // we expect the output buffer to be reversed
        let expected = [
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(15),
            G::from_u8(14),
            G::from_u8(13),
            G::from_u8(12),
            G::from_u8(11),
            G::from_u8(10),
            G::from_u8(9),
            G::from_u8(8),
        ]
        .to_vec();

        assert_eq!(chal.output_buffer, expected);

        let first: G = chal.sample();
        let second: G = chal.sample();

        // sampling pops from end of output buffer
        assert_eq!(first, G::from_u8(8));
        assert_eq!(second, G::from_u8(9));
    }

    #[test]
    fn test_duplexing_only_when_needed() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        chal.output_buffer = vec![G::from_u8(10), G::from_u8(20)];

        // Sample should not call duplexing; just pop from the buffer
        let sample: G = chal.sample();
        assert_eq!(sample, G::from_u8(20));
        assert_eq!(chal.output_buffer, vec![G::from_u8(10)]);
    }

    #[test]
    fn test_flush_when_input_full() {
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // Observe RATE elements, causing a duplexing
        for i in 0..RATE {
            chal.observe(G::from_u8(i as u8));
        }

        // We expect the output buffer to be reversed
        let expected_output = [
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(0),
            G::from_u8(15),
            G::from_u8(14),
            G::from_u8(13),
            G::from_u8(12),
            G::from_u8(11),
            G::from_u8(10),
            G::from_u8(9),
            G::from_u8(8),
        ]
        .to_vec();

        // Input buffer should be drained after duplexing
        assert!(chal.input_buffer.is_empty());

        // Output buffer should match expected state from duplexing
        assert_eq!(chal.output_buffer, expected_output);
    }

    #[test]
    fn test_observe_base_as_algebra_element_consistency_with_direct_observe() {
        // Create two identical challengers to verify behavior equivalence
        let mut chal1 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        let mut chal2 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        let base_val = G::from_u8(99);

        // Method 1: Use the convenience method for base-to-extension observation
        chal1.observe_base_as_algebra_element::<EF2G>(base_val);

        // Method 2: Manually convert to extension field then observe
        let ext_val = EF2G::from(base_val);
        chal2.observe_algebra_element(ext_val);

        // Both methods must produce identical internal state
        assert_eq!(chal1.input_buffer, chal2.input_buffer);
        assert_eq!(chal1.output_buffer, chal2.output_buffer);
        assert_eq!(chal1.sponge_state, chal2.sponge_state);
    }

    #[test]
    fn test_observe_base_as_algebra_element_stream_consistency() {
        // Create two identical challengers for stream observation test
        let mut chal1 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        let mut chal2 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // Define a base value vector
        let base_values: Vec<_> = (0u8..25).map(G::from_u8).collect();

        // Method 1: Observe stream using convenience method
        for &val in &base_values {
            chal1.observe_base_as_algebra_element::<EF2G>(val);
        }

        // Method 2: Manually convert each element before observing
        for &val in &base_values {
            let ext_val = EF2G::from(val);
            chal2.observe_algebra_element(ext_val);
        }

        // Verify identical state through sequential observations and duplexing.
        assert_eq!(chal1.input_buffer, chal2.input_buffer);
        assert_eq!(chal1.output_buffer, chal2.output_buffer);
        assert_eq!(chal1.sponge_state, chal2.sponge_state);

        // Verify sampling produces identical challenges
        let sample1: EF2G = chal1.sample_algebra_element();
        let sample2: EF2G = chal2.sample_algebra_element();
        assert_eq!(sample1, sample2);

        // Verify state consistency is maintained after sampling
        assert_eq!(chal1.input_buffer, chal2.input_buffer);
        assert_eq!(chal1.output_buffer, chal2.output_buffer);
        assert_eq!(chal1.sponge_state, chal2.sponge_state);
    }

    #[test]
    fn test_observe_algebra_elements_equivalence() {
        // Test that the two following paths give the same results:
        // - `observe_algebra_slice`
        // - `observe_algebra_element` in a loop
        let mut chal1 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        let mut chal2 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // Create a slice of extension field elements
        let ext_values: Vec<EF2G> = (0u8..10).map(|i| EF2G::from(G::from_u8(i))).collect();

        // Method 1: Use observe_algebra_slice with slice
        chal1.observe_algebra_slice(&ext_values);

        // Method 2: Call observe_algebra_element individually
        for ext_val in &ext_values {
            chal2.observe_algebra_element(*ext_val);
        }

        // Verify identical internal state
        assert_eq!(chal1.input_buffer, chal2.input_buffer);
        assert_eq!(chal1.output_buffer, chal2.output_buffer);
        assert_eq!(chal1.sponge_state, chal2.sponge_state);

        // Verify sampling produces identical challenges
        let sample1: EF2G = chal1.sample_algebra_element();
        let sample2: EF2G = chal2.sample_algebra_element();
        assert_eq!(sample1, sample2);
    }

    #[test]
    fn test_observe_algebra_elements_empty_slice() {
        // Test that observing an empty slice does not change state
        let mut chal1 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});
        let mut chal2 =
            DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // Observe some values first to have non-trivial state
        chal1.observe(G::from_u8(42));
        chal2.observe(G::from_u8(42));

        // Observe empty slice
        let empty: Vec<EF2G> = vec![];
        chal1.observe_algebra_slice(&empty);

        // Verify state unchanged
        assert_eq!(chal1.input_buffer, chal2.input_buffer);
        assert_eq!(chal1.output_buffer, chal2.output_buffer);
        assert_eq!(chal1.sponge_state, chal2.sponge_state);
    }

    #[test]
    fn test_observe_algebra_elements_triggers_duplexing() {
        // Test that observing enough elements triggers duplexing
        let mut chal = DuplexChallenger::<G, TestPermutation, WIDTH, RATE>::new(TestPermutation {});

        // EF2G has dimension 2, so we need RATE/2 elements to fill the buffer
        //
        // With RATE=16, we need 8 EF2G elements to trigger duplexing
        let ext_values: Vec<EF2G> = (0u8..8).map(|i| EF2G::from(G::from_u8(i))).collect();

        assert!(chal.input_buffer.is_empty());
        assert!(chal.output_buffer.is_empty());

        chal.observe_algebra_slice(&ext_values);

        // After observing 8 EF2G elements (16 base field elements), duplexing should occur
        assert!(chal.input_buffer.is_empty());
        assert!(!chal.output_buffer.is_empty());
    }

    #[test]
    fn test_finalize() {
        let new_chal = || DuplexChallenger::<G, _, WIDTH, RATE>::new(TestPermutation {});

        // Deterministic: same observations produce same digest.
        let mut c1 = new_chal();
        let mut c2 = new_chal();
        for i in 0..5u8 {
            c1.observe(G::from_u8(i));
            c2.observe(G::from_u8(i));
        }
        assert_eq!(c1.finalize(), c2.finalize());

        // Different observations produce different digests.
        let mut c1 = new_chal();
        let mut c2 = new_chal();
        for i in 0..10u8 {
            c1.observe(G::from_u8(i));
            c2.observe(G::from_u8(i + 1));
        }
        assert_ne!(c1.finalize(), c2.finalize());
    }

    /// Document how sampling interacts with finalize.
    ///
    /// Sampling does not modify the sponge state — it only pops from the
    /// output buffer. This means the digest only changes when a sample
    /// triggers a new duplexing (because the output buffer was empty or
    /// there was pending input). Within one "batch" of RATE outputs, all
    /// sample counts produce the same digest.
    #[test]
    fn test_finalize_sample_interaction() {
        let digest = |n_samples: usize| {
            let mut c = DuplexChallenger::<G, _, WIDTH, RATE>::new(TestPermutation {});
            for i in 0..5u8 {
                c.observe(G::from_u8(i));
            }
            for _ in 0..n_samples {
                let _: G = c.sample();
            }
            c.finalize()
        };

        // The first sample triggers duplexing (absorbs pending input),
        // so finalize's duplexing is now an extra permutation on an
        // already-duplexed state — different from the 0-sample case.
        assert_ne!(digest(0), digest(1));

        // Samples 1 through RATE all come from the same output batch.
        // They don't trigger another duplexing, so the sponge state
        // (and thus the digest) is identical.
        assert_eq!(digest(1), digest(2));
        assert_eq!(digest(1), digest(RATE));

        // The (RATE+1)-th sample exhausts the output buffer and triggers
        // a fresh duplexing, changing the sponge state again.
        assert_ne!(digest(RATE), digest(RATE + 1));

        // Within the second batch, the digest is again stable.
        assert_eq!(digest(RATE + 1), digest(RATE + 2));
    }
}