Struct StoredRotation 

pub struct StoredRotation { /* private fields */ }

Stored deterministic orthogonal rotation.

Implementations

impl StoredRotation

pub fn new(dim: usize, seed: u64) -> Result<Self>

Generate a Haar-like orthogonal matrix via QR decomposition.

pub fn dim(&self) -> usize

Dimension of this rotation.

pub fn seed(&self) -> u64

Seed used for deterministic generation.

pub fn rotation_schema(&self) -> &'static str

Stable rotation schema marker.

pub fn algorithm_version(&self) -> &'static str

Rotation algorithm identity.

pub fn digest(&self) -> Result<String>

Deterministic digest over the rotation identity and matrix values.

pub fn apply(&self, input: &[f64]) -> Result<Vec<f64>>

Apply y = Pi x.

pub fn apply_inverse(&self, input: &[f64]) -> Result<Vec<f64>>

Apply inverse x = Pi^T y.

pub fn matrix_f32(&self) -> Vec<f32>

Return the rotation matrix converted to f32. Useful for callers that need to reuse the converted matrix across many apply_inverse_f32_with_matrix calls without reconverting each time.

pub fn apply_inverse_f32(&self, input: &[f32]) -> Result<Vec<f32>>

Apply inverse x = Pi^T y in f32. Faster than the f64 version for batch decode of many small vectors, where the f32→f64 roundtrip is the bottleneck. Result is mathematically equivalent for the as f32 final-cast case; intermediate precision loss is below the codebook quantization noise floor.

pub fn apply_inverse_f32_with_matrix( &self, input: &[f32], matrix_f32: &[f32], ) -> Result<Vec<f32>>

Apply inverse x = Pi^T y using a pre-converted f32 matrix. This avoids the f64→f32 conversion on every call, which is the bottleneck in decode_batch_fast when decoding thousands of codes that share the same rotation.

pub fn apply_inverse_batch_f32( &self, inputs: &[&[f32]], ) -> Result<Vec<Vec<f32>>>

Apply inverse x = Pi^T y to a batch of inputs in one call. The matrix is converted to f32 once and reused across the batch.

Trait Implementations

impl Clone for StoredRotation

fn clone(&self) -> StoredRotation

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StoredRotation

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for StoredRotation

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for StoredRotation

fn eq(&self, other: &StoredRotation) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for StoredRotation

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for StoredRotation