# rustymath

> Maths in plain Rust, from fast f16, f32 and f64 functions to exact rationals.

## Modules

- `rustymath.linalg`: rustymath's linear algebra, under the names of NumPy's similar functions. Each function takes a stack of matrices in the last two axes of its arguments, and broadcasts the axes before them as NumPy's generalized ufuncs do. float32, float64, complex64 and complex128 matrices compute at their own width. float16 matrices compute in float32, with each result rounded to float16. Integers and Booleans compute as float64, except in `matmul` and `semiring`, where integers compute in int64.

<!-- Generated by nbdev-export; edit the index notebook, not this file -->
