Implements the novel CLOSURE technique for efficiently reconstructing all possible distributions of raw data from summary statistics. It is not about the Rust feature called closure.

The crate is mostly meant to serve as a backend for the R package unsum.

You will likely only need dfs_parallel().

Most of the code was written by Claude 3.5, translating Python code by Nathanael Larigaldie.

Enter summary data reported in a paper and call dfs_parallel().

let mean = 5.0;
let sd = 2.78;
let n = 30;
let scale_min = 1;
let scale_max = 7;
let rounding_error_mean = 0.005;
let rounding_error_sd = 0.005;

dfs_parallel(
    mean,
    sd,
    n,
    scale_min,
    scale_max,
    rounding_error_mean,
    rounding_error_sd,
)