flow_dimensional_reduction/lib.rs
1//! Dimensionality reduction primitives for flow cytometry.
2//!
3//! Currently provides [`Pca`], a faer-based principal component analysis using
4//! the covariance method: it decomposes the `d × d` covariance matrix rather
5//! than the `n × d` data matrix. For flow cytometry workloads (n ≈ 10⁶–10⁷,
6//! d ≈ 10–50) this is dramatically cheaper.
7//!
8//! Data is `f32` on the boundary; means and covariance are accumulated in `f64`
9//! and downcast only when the final basis is stored.
10
11pub mod pca;
12
13pub use pca::{Pca, PcaComponent, PcaError, PcaResult};