Skip to main content

ComputeEngine

chronix_analytics::compute

Trait ComputeEngine 

Source 
pub trait ComputeEngine: Send + Sync {
    // Required methods
    fn batch_dot_product(
        &self,
        a: &[f64],
        b: &[f64],
    ) -> Result<f64, ComputeError>;
    fn batch_matrix_solve(
        &self,
        a_flat: &[f64],
        b: &[f64],
        n: usize,
    ) -> Result<Vec<f64>, ComputeError>;
    fn batch_exponential_smooth(
        &self,
        values: &[f64],
        alpha: f64,
    ) -> Result<Vec<f64>, ComputeError>;
    fn batch_difference(
        &self,
        values: &[f64],
        d: usize,
    ) -> Result<Vec<f64>, ComputeError>;
    fn batch_autocorrelation(
        &self,
        values: &[f64],
        max_lag: usize,
    ) -> Result<Vec<f64>, ComputeError>;
    fn batch_least_squares(
        &self,
        a_flat: &[f64],
        b: &[f64],
        m: usize,
        n: usize,
    ) -> Result<Vec<f64>, ComputeError>;
}
Expand description

Compute engine trait — defines batch numerical operations used by forecast, anomaly detection, and multivariate engines.

Required Methods§

Source

fn batch_dot_product(&self, a: &[f64], b: &[f64]) -> Result<f64, ComputeError>

Dot product of two vectors.

Source

fn batch_matrix_solve( &self, a_flat: &[f64], b: &[f64], n: usize, ) -> Result<Vec<f64>, ComputeError>

Solve a linear system Ax = b using LU decomposition.

a_flat is the row-major flattened NxN matrix A. b is the right-hand-side vector of length N. Returns x such that Ax = b.

Source

fn batch_exponential_smooth( &self, values: &[f64], alpha: f64, ) -> Result<Vec<f64>, ComputeError>

Exponential smoothing on a time series.

Returns smoothed values with the given alpha parameter.

Source

fn batch_difference( &self, values: &[f64], d: usize, ) -> Result<Vec<f64>, ComputeError>

Compute d-th order differences of a series.

Source

fn batch_autocorrelation( &self, values: &[f64], max_lag: usize, ) -> Result<Vec<f64>, ComputeError>

Compute autocorrelation at the given lags.

Source

fn batch_least_squares( &self, a_flat: &[f64], b: &[f64], m: usize, n: usize, ) -> Result<Vec<f64>, ComputeError>

Solve least-squares: find x minimizing ||Ax - b||².

a_flat is row-major M×N matrix, b is length-M vector. Returns x of length N via Householder QR decomposition (numerically stable — preserves κ(A) instead of squaring it like normal equations).

Dyn Compatibility§

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors§