Sync Rust normal_dist_inv_cdf with Python equivalent

jackoconnordev · jackoconnordev · commit 3ef5264f614b · 2025-07-25T10:46:01.000+01:00
See python/cpython#95265. To quote: > Restores alignment with random.gauss(mu, sigma) and random.normalvariate(mu, sigma) both. of which are equivalent to sampling from NormalDist(mu, sigma).inv_cdf(random()). The two functions in the random module happy accept sigma=0 and give a well-defined result. > This also lets the function gently handle a sigma getting smaller, eventually becoming zero. As sigma decrease, NormalDist(mu, sigma).inv_cdf(p) forms a tighter and tighter internal around mu and becoming exactly mu in the limit. For example, NormalDist(100, 1E-300).inv_cdf(0.3) cleanly evaluates to 100.0but withsigma=1e-500`` the function previously would raised an unexpected error.
diff --git a/stdlib/src/statistics.rs b/stdlib/src/statistics.rs
@@ -7,7 +7,7 @@ mod _statistics {
     // See https://github.com/python/cpython/blob/6846d6712a0894f8e1a91716c11dd79f42864216/Modules/_statisticsmodule.c#L28-L120
     #[allow(clippy::excessive_precision)]
     fn normal_dist_inv_cdf(p: f64, mu: f64, sigma: f64) -> Option<f64> {
-        if p <= 0.0 || p >= 1.0 || sigma <= 0.0 {
+        if p <= 0.0 || p >= 1.0 {
             return None;
         }
 

Original file line number	Diff line number	Diff line change
`@@ -7,7 +7,7 @@ mod _statistics {`
`7`	`7`	`// See https://github.com/python/cpython/blob/6846d6712a0894f8e1a91716c11dd79f42864216/Modules/_statisticsmodule.c#L28-L120`
`8`	`8`	`#[allow(clippy::excessive_precision)]`
`9`	`9`	`fn normal_dist_inv_cdf(p: f64, mu: f64, sigma: f64) -> Option<f64> {`
`10`		`- if p <= 0.0 \|\| p >= 1.0 \|\| sigma <= 0.0 {`
	`10`	`+ if p <= 0.0 \|\| p >= 1.0 {`
`11`	`11`	`return None;`
`12`	`12`	`}`
`13`	`13`