Running combination of loops in parallel

Question

The following code calculates a matrix of distances between each pair of elements in the corpus. However, for large corpora the process becomes quite lengthy. How can I parallelize it with rayon?

use std::iter::Iterator;
use ndarray::Array2;

type DistanceFn = dyn Fn(&f64, &f64) -> f64 + Sync;

fn calculate(corpus: &[f64], distance_fn: Box<DistanceFn>) -> Array2<f64>  {
    let mut distances: Array2<f64> = Array2::zeros((corpus.len(), corpus.len()));
    for (i, m1) in corpus.iter().enumerate() {
        for (j, m2) in corpus.iter().enumerate() {
            distances[[i, j]] = distance_fn(m1, m2);
        }
    }
    distances
}

fn diff_distance(m1:&f64, m2:&f64) -> f64 {
    m1-m2
}

fn main() {
    let corpus = &vec![5.0,6.0,7.0];
    let matrix = calculate(corpus, Box::new(diff_distance));
    print!("{}", matrix)
}