native_neural_network 0.3.1

use crate::sphere5d::NeuronPoint;

pub fn centroid(points: &[NeuronPoint]) -> [f32; 5] {
    let mut c = [0.0f32; 5];
    if points.is_empty() {
        return c;
    }
    for p in points {
        for (i, v) in p.position.iter().enumerate() {
            c[i] += *v;
        }
    }
    let scale = 1.0 / points.len() as f32;
    for v in &mut c {
        *v *= scale;
    }
    c
}

pub fn mean_radius(points: &[NeuronPoint], center: [f32; 5]) -> f32 {
    if points.is_empty() {
        return 0.0;
    }
    let mut acc = 0.0f32;
    for p in points {
        let mut d2 = 0.0f32;
        for (i, pos_v) in p.position.iter().enumerate() {
            let d = *pos_v - center[i];
            d2 += d * d;
        }
        acc += crate::math::sqrtf(d2);
    }
    acc / points.len() as f32
}

pub fn centroid_f64(points: &[[f64; 5]]) -> [f64; 5] {
    let mut c = [0.0f64; 5];
    if points.is_empty() {
        return c;
    }
    for p in points {
        for (i, v) in p.iter().enumerate() {
            c[i] += *v;
        }
    }
    let scale = 1.0 / points.len() as f64;
    for v in &mut c {
        *v *= scale;
    }
    c
}

pub fn mean_radius_f64(points: &[[f64; 5]], center: [f64; 5]) -> f64 {
    if points.is_empty() {
        return 0.0;
    }
    let mut acc = 0.0f64;
    for p in points {
        let mut d2 = 0.0f64;
        for (i, pos_v) in p.iter().enumerate() {
            let d = *pos_v - center[i];
            d2 += d * d;
        }
        acc += crate::math::sqrtd(d2);
    }
    acc / points.len() as f64
}