ferray-window

NumPy-equivalent window functions (plus a few functional-programming utilities) for the ferray workspace.

Part of the ferray workspace — a Rust-native, drop-in replacement for NumPy.

Overview

This crate provides window functions for signal processing and spectral analysis. All window functions take a length m: usize and return an Array1<f64> (the f32 siblings return Array1<f32>), matching NumPy's output to high precision.

The five NumPy window functions:

bartlett(m) — triangular window
blackman(m) — Blackman window
hamming(m) — Hamming window
hanning(m) — Hann window
kaiser(m, beta) — Kaiser window parameterised by beta

Each also has an f32 sibling: bartlett_f32, blackman_f32, hamming_f32, hanning_f32, kaiser_f32.

SciPy-style extras (beyond NumPy's five): boxcar, triang, cosine, exponential, gaussian, general_cosine, general_hamming, nuttall, parzen, bohman, flattop, lanczos, tukey, taylor, chebwin, and dpss (Slepian taper).

Functional-programming utilities are also re-exported: vectorize, piecewise, apply_along_axis, apply_over_axes, and sum_axis_keepdims.

NumPy correspondence

ferray	NumPy
`bartlett(m)`	`np.bartlett(M)`
`blackman(m)`	`np.blackman(M)`
`hamming(m)`	`np.hamming(M)`
`hanning(m)`	`np.hanning(M)`
`kaiser(m, beta)`	`np.kaiser(M, beta)`

Feature flags

None. The crate builds with default features only and adds no extra runtime dependencies.

Example

use ferray_window::{hanning, kaiser};

// Hann window of length 64.
let w = hanning(64)?;

// Kaiser window of length 64 with beta = 14.0.
let k = kaiser(64, 14.0)?;

assert_eq!(w.len(), 64);
assert_eq!(k.len(), 64);
# Ok::<(), ferray_core::FerrayError>(())

MSRV & edition