eevee 0.2.1

Generalized NeuroEvolution toolkit, based on NEAT
Documentation 
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
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150

use super::{FromGenome, Network};
use crate::{Connection, Genome};
use core::ops::Range;

/// A simple neural network, because man, what the fuck is going on. lol
/// Walks through connections oldest to newest, evaluating them on a flat state
#[derive(Debug)]
#[cfg_attr(
    all(feature = "serialize", not(feature = "serialize_json")),
    derive(serde::Serialize, serde::Deserialize),
    serde(bound(
        serialize = "C: serde::Serialize",
        deserialize = "C: serde::Deserialize<'de>"
    ))
)]
pub struct Simple<C: Connection> {
    pub(crate) connections: Vec<C>,
    pub(crate) bias: Vec<f64>,
    #[cfg_attr(
        all(feature = "serialize", not(feature = "serialize_json")),
        serde(skip_serializing)
    )]
    pub(crate) state: Vec<f64>,
    #[cfg_attr(
        all(feature = "serialize", not(feature = "serialize_json")),
        serde(skip_serializing)
    )]
    pub(crate) sensory: Range<usize>,
    #[cfg_attr(
        all(feature = "serialize", not(feature = "serialize_json")),
        serde(skip_serializing)
    )]
    pub(crate) action: Range<usize>,
}

impl<C: Connection> Network for Simple<C> {
    fn step<F: Fn(f64) -> f64>(&mut self, prec: usize, input: &[f64], σ: F) {
        debug_assert!(input.len() == self.sensory.len());
        self.state[self.sensory.start..self.sensory.end].copy_from_slice(input);
        if !self.connections.is_empty() {
            for _ in 0..prec {
                for c in self.connections.iter() {
                    self.state[c.to()] +=
                        σ((self.bias[c.from()] + self.state[c.from()]) * c.weight())
                }
            }
        }
    }

    fn flush(&mut self) {
        self.state = vec![0.; self.state.len()];
    }

    fn output(&self) -> &[f64] {
        &self.state[self.action.start..self.action.end]
    }
}

impl<C: Connection, G: Genome<C>> FromGenome<C, G> for Simple<C> {
    fn from_genome(genome: &G) -> Self {
        Simple {
            connections: genome.connections().to_owned(),
            bias: vec![0.; genome.node_count()],
            state: vec![0.; genome.node_count()],
            sensory: genome.sensory(),
            action: genome.action(),
        }
    }
}

#[cfg(test)]
mod test {
    use super::*;
    use crate::genome::{self, connection::BWConnection, WConnection};
    use eevee_macros::fn_matrix;

    fn_matrix! {
        C: WConnection | BWConnection,
        G: genome::Recurrent<C>,

        /// all connections copied from genome (including disabled)
        #[test]
        fn test_preserves_all_connections() {
            let (genome, _) = G::new(2, 2);
            let nn = Simple::from_genome(&genome);

            assert_eq!(nn.connections.len(), genome.connections().len());
        }

        /// enabled/disabled status preserved from genome
        #[test]
        fn test_preserves_enabled_status() {
            let (mut genome, _) = G::new(2, 2);
            if let Some(c) = genome.connections_mut().first_mut() {
                c.disable();
            }

            let nn = Simple::from_genome(&genome);
            assert!(!nn.connections[0].enabled());
            assert!(nn.connections[1].enabled());
        }

        /// state/bias vectors match node count
        #[test]
        fn test_state_size() {
            let (genome, _) = G::new(3, 2);
            let nn = Simple::from_genome(&genome);

            assert_eq!(nn.state.len(), genome.node_count());
            assert_eq!(nn.bias.len(), genome.node_count());
        }

        /// input range matches sensory neurons
        #[test]
        fn test_input_range() {
            let (genome, _) = G::new(3, 2);
            let nn = Simple::from_genome(&genome);

            assert_eq!(nn.sensory.start, genome.sensory().start);
            assert_eq!(nn.sensory.end, genome.sensory().end);
            assert_eq!(nn.sensory.len(), genome.sensory().len());
        }

        /// state initialized to zeros
        #[test]
        fn test_initial_state() {
            let (genome, _) = G::new(2, 2);
            let nn = Simple::from_genome(&genome);

            assert!(nn.state.iter().all(|&x| x == 0.0));
        }

        /// flush() clears state vector
        #[test]
        fn test_flush_clears_state() {
            let (genome, _) = G::new(2, 2);
            let mut nn = Simple::from_genome(&genome);
            let input = vec![1.0, 0.5];

            nn.step(2, &input, |x| x);
            let state_before = nn.state.clone();

            nn.flush();
            let state_after = nn.state.clone();

            assert!(state_before.iter().any(|&x| x != 0.0));
            assert!(state_after.iter().all(|&x| x == 0.0));
        }
    }
}