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

use super::{Feedback, Summary};
use crate::logging::{LogError, LogValue, Loggable, StatsLogger};
use crate::utils::stats::OnlineMeanVariance;
use derive_more::{Add, AddAssign, Sub, SubAssign};
use num_traits::{Bounded, ToPrimitive};
use serde::{Deserialize, Serialize};
use std::fmt;

/// A reward-signal environment feedback
#[derive(
    Debug,
    Default,
    Copy,
    Clone,
    PartialEq,
    PartialOrd,
    Add,
    AddAssign,
    Sub,
    SubAssign,
    Serialize,
    Deserialize,
)]
pub struct Reward(pub f64);

impl fmt::Display for Reward {
    #[inline]
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        fmt::Display::fmt(&self.0, f)
    }
}

impl From<f64> for Reward {
    #[inline]
    fn from(r: f64) -> Self {
        Self(r)
    }
}

impl From<Reward> for f64 {
    #[inline]
    fn from(Reward(r): Reward) -> Self {
        r
    }
}

impl Reward {
    /// Get the reward value as a float
    #[inline]
    #[must_use]
    pub const fn unwrap(self) -> f64 {
        self.0
    }
}

impl Feedback for Reward {
    type EpisodeFeedback = Self;
    type StepSummary = RewardSummary;
    type EpisodeSummary = RewardSummary;
    #[inline]
    fn add_to_episode_feedback(&self, episode_feedback: &mut Self::EpisodeFeedback) {
        episode_feedback.0 += self.0
    }
}

impl Bounded for Reward {
    #[inline]
    fn min_value() -> Self {
        Self(Bounded::min_value())
    }
    #[inline]
    fn max_value() -> Self {
        Self(Bounded::max_value())
    }
}

impl ToPrimitive for Reward {
    #[inline]
    fn to_i64(&self) -> Option<i64> {
        self.0.to_i64()
    }
    #[inline]
    fn to_u64(&self) -> Option<u64> {
        self.0.to_u64()
    }
    #[inline]
    fn to_f32(&self) -> Option<f32> {
        self.0.to_f32()
    }
    #[inline]
    fn to_f64(&self) -> Option<f64> {
        self.0.to_f64()
    }
}

impl Loggable for Reward {
    fn log<L: StatsLogger + ?Sized>(
        &self,
        name: &'static str,
        logger: &mut L,
    ) -> Result<(), LogError> {
        logger
            .with_scope(name)
            .log("reward".into(), LogValue::Scalar(self.0))
    }
}

/// A summmary of reward values.
#[derive(Debug, Default, Copy, Clone, PartialEq, Add, AddAssign, Serialize, Deserialize)]
pub struct RewardSummary(pub OnlineMeanVariance<f64>);

impl fmt::Display for RewardSummary {
    #[inline]
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        fmt::Display::fmt(&self.0, f)
    }
}

impl Summary for RewardSummary {
    type Item = Reward;
    #[inline]
    fn push(&mut self, reward: &Reward) {
        self.0.push(reward.0)
    }
    #[inline]
    fn size(&self) -> u64 {
        self.0.count()
    }
    #[inline]
    fn merge(&mut self, other: Self) {
        self.0 += other.0
    }
}

impl Loggable for RewardSummary {
    fn log<L: StatsLogger + ?Sized>(
        &self,
        name: &'static str,
        logger: &mut L,
    ) -> Result<(), LogError> {
        self.0.log("reward", &mut logger.with_scope(name))
    }
}