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

// Traits
use rand_distr::Distribution;
use crate::traits::{State, StateIterator, Transition};
use core::fmt::Debug;
use rand::Rng;

// Structs
use crate::errors::InvalidState;

// Functions
use core::mem;

/// Markov Chain in discrete time, with arbitrary space.
///
/// # Remarks
/// 
/// If your transition function `transition` could reuse of structs that implement
/// the `Distribution<T>` trait in order to sample the next state, then, 
/// for the best performance possible, create your own struct that implements
/// the `Transition<T, T>` trait.
#[derive(Debug, Clone)]
pub struct MarkovChain<T, F, R> {
    state: T,
    transition: F,
    rng: R,
}

impl<T, F, R> MarkovChain<T, F, R>
where
    R: Rng,
    F: Transition<T, T>,
{
    #[inline]
    pub fn new(state: T, transition: F, rng: R) -> Self {
        MarkovChain {
            state,
            transition,
            rng,
        }
    }
}

impl<T, F, R> State for MarkovChain<T, F, R>
where
    T: Debug + Clone,
{
    type Item = T;

    #[inline]
    fn state(&self) -> Option<&Self::Item> {
        Some(&self.state)
    }

    #[inline]
    fn state_mut(&mut self) -> Option<&mut Self::Item> {
        Some(&mut self.state)
    }

    #[inline]
    fn set_state(
        &mut self,
        mut new_state: Self::Item,
    ) -> Result<Option<Self::Item>, InvalidState<Self::Item>> {
        mem::swap(&mut self.state, &mut new_state);
        Ok(Some(new_state))
    }
}

impl<T, F, R> Iterator for MarkovChain<T, F, R>
where
    T: Debug + Clone,
    F: Transition<T, T>,
    R: Rng,
{
    type Item = T;

    #[inline]
    fn next(&mut self) -> Option<Self::Item> {
        self.state = self.transition.sample_from(&self.state, &mut self.rng);
        self.state().cloned()
    }
}

impl<T, F, R> StateIterator for MarkovChain<T, F, R>
where
    T: Debug + Clone,
    F: Transition<T, T>,
    R: Rng,
{
    #[inline]
    fn state_as_item(&self) -> Option<<Self as std::iter::Iterator>::Item> {
        self.state().cloned()
    }
}

impl<T, F, R> Distribution<T> for MarkovChain<T, F, R>
where
    T: Debug + Clone,
    F: Transition<T, T>,
    R: Rng,
{
    /// Sample a possible next state. 
    #[inline]
    fn sample<R2>(&self, rng: &mut R2) -> T
    where
        R2: Rng + ?Sized,
    { 
        self.transition.sample_from(&self.state, rng)
    }
}


#[cfg(test)]
mod tests {
    use super::*;
    use crate::distributions::Raw;
    use pretty_assertions::assert_eq;

    // use approx::abs_diff_eq;

    #[test]
    fn sampling_stability() {
        let rng = crate::tests::rng(1);
        let expected = 1;
        let transition = |_: &u64| Raw::new(vec![(1.0, expected)]);
        let mc = MarkovChain::new(0, transition, rng);
        for x in mc.take(100) {
            assert_eq!(x, expected);
        }

        let rng = crate::tests::rng(2);
        let transition = |_: &u64| Raw::new(vec![(0.5, 1), (0.5, 2)]);
        let mc = MarkovChain::new(0, transition, rng);
        for x in mc.take(100) {
            assert!(x == 1 || x == 2);
        }
    }

    #[test]
    fn value_stability() {
        let rng = crate::tests::rng(3);
        let expected = vec![1, 2, 1, 1];
        let transition = |_: &u64| Raw::new(vec![(0.5, 1), (0.5, 2)]);
        let mc = MarkovChain::new(0, transition, rng);
        let sample: Vec<u64> = mc.take(4).collect();

        assert_eq!(sample, expected);
    }

    #[test]
    fn construction() {
        let rng = crate::tests::rng(4);
        let expected = 0.39515292318166956;
        let transition = |_: &f64| rand_distr::StandardNormal;
        let mut mc = MarkovChain::new(0., transition, rng);
        let sample: f64 = mc.next().unwrap();

        assert_eq!(sample, expected);
    }
}