rl 0.4.0

A reinforcement learning library
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

use rand::{seq::IteratorRandom, thread_rng};

use crate::env::{DiscreteActionSpace, Environment, Report};

/// The possible types of squares in the [`FrozenLake`] grid
#[derive(PartialEq)]
pub enum Square {
    Frozen = 0,
    Hole = 1,
    Start = 2,
    Goal = 3,
}

/// Actions for the [`FrozenLake`] environment, representing taking a step in a direction
#[derive(Clone, Copy, PartialEq, Eq, Hash, Debug)]
pub enum FLAction {
    Left = 0,
    Down = 1,
    Right = 2,
    Up = 3,
}

/// A very simple RL environment taken from Python [gymnasium](https://gymnasium.farama.org/)
///
/// Intended for use with a [QTableAgent](crate::algo::q_table::QTableAgent)
pub struct FrozenLake {
    map: [Square; 16],
    pos: usize,
    pub report: Report,
}

impl Default for FrozenLake {
    fn default() -> Self {
        Self::new()
    }
}

impl FrozenLake {
    pub fn new() -> Self {
        // TODO: Support custom maps
        let map = [
            Square::Start,
            Square::Frozen,
            Square::Frozen,
            Square::Frozen,
            Square::Frozen,
            Square::Hole,
            Square::Frozen,
            Square::Hole,
            Square::Frozen,
            Square::Frozen,
            Square::Frozen,
            Square::Hole,
            Square::Hole,
            Square::Frozen,
            Square::Frozen,
            Square::Goal,
        ];
        Self {
            map,
            pos: 0,
            report: Report::new(vec!["reward", "steps"]),
        }
    }
}

impl DiscreteActionSpace for FrozenLake {
    fn actions(&self) -> Vec<Self::Action> {
        let mut actions = Vec::with_capacity(4);

        if self.pos % 4 != 0 {
            actions.push(FLAction::Left)
        }
        if self.pos < 12 {
            actions.push(FLAction::Down)
        }
        if self.pos % 4 != 3 {
            actions.push(FLAction::Right)
        }
        if self.pos > 3 {
            actions.push(FLAction::Up)
        }

        actions
    }
}

impl Environment for FrozenLake {
    type State = usize;
    type Action = FLAction;

    fn is_active(&self) -> bool {
        match self.map[self.pos] {
            Square::Frozen | Square::Start => true,
            Square::Hole | Square::Goal => false,
        }
    }

    fn random_action(&self) -> Self::Action {
        self.actions()
            .into_iter()
            .choose(&mut thread_rng())
            .expect("There is always at least one available action in this environment")
    }

    fn step(&mut self, action: Self::Action) -> (Option<Self::State>, f32) {
        self.report.entry("steps").and_modify(|x| *x += 1.0);

        match action {
            FLAction::Left => self.pos -= 1,
            FLAction::Down => self.pos += 4,
            FLAction::Right => self.pos += 1,
            FLAction::Up => self.pos -= 4,
        };

        let (next_state, reward) = match self.map[self.pos] {
            Square::Hole => (None, -1.0),
            Square::Goal => (None, 1.0),
            _ => (Some(self.pos), -0.1),
        };

        self.report.entry("reward").and_modify(|x| *x += reward);

        (next_state, reward as f32)
    }

    fn reset(&mut self) -> Self::State {
        self.pos = 0;
        self.pos
    }
}