1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58

//! implementations about neural networks's mutation (weight and bias)

use rand::prelude::*;
use rand_distr::{Distribution, Normal};

use crate::{
    brain::{
        neuron::{BlockNN, BrainNN, GenericNN},
        nn::BaseNN,
    },
    consts::mutate_consts::*,
};

/// mutate Neuron Networks
pub fn mutate_nn(nnvec: &mut Vec<GenericNN>) {
    for nn in nnvec.iter_mut() {
        let mut rng: ThreadRng = thread_rng();

        if !rng.gen_bool(MUTATE_NN_PORB as f64) {
            continue;
        }

        match nn {
            GenericNN::BRAINNN(nn) => mutate_brain_nn(nn),
            GenericNN::BLOCKNN(nn) => mutate_block_nn(nn),
        }
    }
}

fn mutate_block_nn(nn: &mut BlockNN) {
    mutate_base_nn(&mut nn.inward_nn.nn);
    mutate_base_nn(&mut nn.outward_nn.nn);
}

fn mutate_brain_nn(nn: &mut BrainNN) {
    mutate_base_nn(&mut nn.nn);
}


/// add an random value to the existed weight and bias
fn mutate_base_nn(nn: &mut BaseNN) {
    let normal = Normal::new(0.0, MUTATE_NN_STD).unwrap();

    // Use the thread_rng to get a thread-local random number generator
    let mut rng: ThreadRng = thread_rng();

    for layer in &mut nn.layers {
        // Mutate weights
        for weight in layer.weights.iter_mut() {
            *weight += normal.sample(&mut rng) as f32;
        }

        // Mutate biases
        for bias in layer.bias.iter_mut() {
            *bias += normal.sample(&mut rng) as f32;
        }
    }
}