gymnasia 3.0.5

OpenAI Gymnasium environments in pure Rust
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
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241

use std::fmt::Debug;

use rand::Rng;
use serde::Serialize;

use super::{sample_space::SampleSpace, space::Space};

/// Types that can serve as element-wise bounds in a [`BoxSpace`].
///
/// Implement this on observation structs, scalar types, or array types
/// to use them with `BoxSpace<B>`. We ship implementations for `f64`
/// and [`Tensor`].
///
/// # Example
///
/// ```
/// use gymnasia::spaces::Bounded;
/// use rand::Rng;
///
/// #[derive(Clone, Debug)]
/// struct MyObs { x: f64, y: f64 }
///
/// impl Bounded for MyObs {
///     fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool {
///         value.x >= low.x && value.x <= high.x
///             && value.y >= low.y && value.y <= high.y
///     }
///     fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self {
///         MyObs {
///             x: rng.gen_range(low.x..=high.x),
///             y: rng.gen_range(low.y..=high.y),
///         }
///     }
///     fn clamp(value: Self, low: &Self, high: &Self) -> Self {
///         MyObs {
///             x: value.x.clamp(low.x, high.x),
///             y: value.y.clamp(low.y, high.y),
///         }
///     }
/// }
/// ```
pub trait Bounded: Sized + Clone + Debug {
    /// Check whether `value` lies within `[low, high]` element-wise.
    fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool;

    /// Sample uniformly within `[low, high]` element-wise.
    fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self;

    /// Clamp `value` element-wise to `[low, high]`.
    fn clamp(value: Self, low: &Self, high: &Self) -> Self;
}

/// A bounded continuous space defined by low/high bounds.
///
/// Generic over the bounds representation `B`. Use typed observation
/// structs for classical control, [`Tensor`] for high-dimensional spaces,
/// or `f64` for scalar spaces.
///
/// # Examples
///
/// ```
/// use gymnasia::spaces::{BoxSpace, Space};
///
/// let space = BoxSpace::new(-1.0_f64, 1.0);
/// assert!(space.contains(&0.5));
/// assert!(!space.contains(&2.0));
/// ```
#[derive(Debug, Clone, Serialize)]
#[serde(bound = "B: Serialize")]
pub struct BoxSpace<B: Bounded> {
    /// The lower bound (inclusive).
    pub low: B,
    /// The upper bound (inclusive).
    pub high: B,
}

impl<B: Bounded> BoxSpace<B> {
    /// Create a new box space with the given bounds.
    pub fn new(low: B, high: B) -> Self {
        Self { low, high }
    }
}

impl<B: Bounded> AsRef<BoxSpace<B>> for BoxSpace<B> {
    fn as_ref(&self) -> &BoxSpace<B> {
        self
    }
}

impl<B: Bounded> Space for BoxSpace<B> {
    type Element = B;

    fn contains(&self, value: &B) -> bool {
        B::in_bounds(value, &self.low, &self.high)
    }
}

impl<B: Bounded> SampleSpace for BoxSpace<B> {
    type Mask = ();

    fn sample<R: Rng>(&self, rng: &mut R, _mask: Option<&Self::Mask>) -> B {
        B::sample_uniform(rng, &self.low, &self.high)
    }
}

// --- Shipped Bounded implementations ---

impl Bounded for Vec<f64> {
    fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool {
        value.len() == low.len()
            && value
                .iter()
                .zip(low.iter().zip(high.iter()))
                .all(|(v, (lo, hi))| v >= lo && v <= hi)
    }

    fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self {
        assert_eq!(low.len(), high.len(), "Vec<f64> bounds length mismatch");
        low.iter()
            .zip(high.iter())
            .map(|(lo, hi)| rng.gen_range(*lo..=*hi))
            .collect()
    }

    fn clamp(value: Self, low: &Self, high: &Self) -> Self {
        value
            .into_iter()
            .zip(low.iter().zip(high.iter()))
            .map(|(v, (lo, hi))| v.clamp(*lo, *hi))
            .collect()
    }
}

impl Bounded for f64 {
    fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool {
        value >= low && value <= high
    }

    fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self {
        rng.gen_range(*low..=*high)
    }

    fn clamp(value: Self, low: &Self, high: &Self) -> Self {
        value.clamp(*low, *high)
    }
}

impl Bounded for f32 {
    fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool {
        value >= low && value <= high
    }

    fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self {
        rng.gen_range(*low..=*high)
    }

    fn clamp(value: Self, low: &Self, high: &Self) -> Self {
        value.clamp(*low, *high)
    }
}

/// A flat array with shape metadata, for high-dimensional spaces.
///
/// Use this as the bound type in `BoxSpace<Tensor>` for image-based
/// or large state-vector observations.
#[derive(Clone, Debug, Serialize)]
pub struct Tensor {
    /// The flat data.
    pub data: Vec<f64>,
    /// The logical shape (e.g. `[210, 160, 3]` for an RGB image).
    pub shape: Vec<usize>,
}

impl Tensor {
    /// Create a new tensor with the given data and shape.
    ///
    /// # Panics
    ///
    /// Panics if `data.len()` does not equal the product of `shape`.
    pub fn new(data: Vec<f64>, shape: Vec<usize>) -> Self {
        let expected: usize = shape.iter().product();
        assert_eq!(
            data.len(),
            expected,
            "data length {} does not match shape {:?} (expected {})",
            data.len(),
            shape,
            expected
        );
        Self { data, shape }
    }

    /// Create a tensor filled with a constant value.
    pub fn filled(value: f64, shape: Vec<usize>) -> Self {
        let n: usize = shape.iter().product();
        Self {
            data: vec![value; n],
            shape,
        }
    }
}

impl Bounded for Tensor {
    fn in_bounds(value: &Self, low: &Self, high: &Self) -> bool {
        assert_eq!(value.shape, low.shape, "Tensor shape mismatch");
        assert_eq!(value.shape, high.shape, "Tensor shape mismatch");
        value
            .data
            .iter()
            .zip(low.data.iter().zip(high.data.iter()))
            .all(|(v, (lo, hi))| v >= lo && v <= hi)
    }

    fn sample_uniform<R: Rng>(rng: &mut R, low: &Self, high: &Self) -> Self {
        assert_eq!(low.shape, high.shape, "Tensor shape mismatch");
        let data: Vec<f64> = low
            .data
            .iter()
            .zip(high.data.iter())
            .map(|(lo, hi)| rng.gen_range(*lo..=*hi))
            .collect();
        Tensor {
            data,
            shape: low.shape.clone(),
        }
    }

    fn clamp(value: Self, low: &Self, high: &Self) -> Self {
        assert_eq!(value.shape, low.shape, "Tensor shape mismatch");
        let data: Vec<f64> = value
            .data
            .iter()
            .zip(low.data.iter().zip(high.data.iter()))
            .map(|(v, (lo, hi))| v.clamp(*lo, *hi))
            .collect();
        Tensor {
            data,
            shape: value.shape,
        }
    }
}