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

use ndarray::prelude::*;

use crate::{AvgLLH, Error, Latent, Parametrizable};

use super::utils::generate_random_expections;

#[derive(Clone, Debug)]

pub struct Finite {
    // pub dimension: i32,
    pub prior: Option<f64>,
    pub pmf: Array1<f64>,
    sufficient_statistics: Array2<f64>,
}

impl Finite {
    pub fn new(prior: Option<f64>) -> Finite {
        // let prior = prior.unwrap_or(1.0);
        Finite {
            // dimension,
            prior,
            pmf: Array1::<f64>::zeros(0),
            sufficient_statistics: Array2::<f64>::zeros((0, 0)),
        }
    }
}

impl Parametrizable for Finite {
    type SufficientStatistics = Array1<f64>;

    type Likelihood = Array2<f64>;

    type DataIn<'a> = ArrayView2<'a, f64>;

    type DataOut = Array2<f64>;

    fn expect(&self, _data: &Self::DataIn<'_>) -> Result<(Self::Likelihood, AvgLLH), Error> {
        Ok((
            self.pmf.slice(s![NewAxis, ..]).mapv(|x| x.ln()), //.to_owned()
            AvgLLH(f64::NAN),
        ))
    }

    fn compute(
        &self,
        _data: &Self::DataIn<'_>,
        responsibilities: &Self::Likelihood,
    ) -> Result<Self::SufficientStatistics, Error> {
        let sum_responsibilities = responsibilities.sum_axis(Axis(0)); // (K)

        Ok(sum_responsibilities)
    }

    fn maximize(
        &mut self,
        sufficient_statistics: &Self::SufficientStatistics,
    ) -> Result<(), Error> {
        if let Some(_x) = self.prior {
        } else {
            self.pmf = sufficient_statistics / sufficient_statistics.sum();
        }

        Ok(())
    }

    fn predict(&self, _data: &Self::DataIn<'_>) -> Result<Self::DataOut, Error> {
        Err(Error::ForbiddenCode)
    }

    fn update(
        &mut self,
        sufficient_statistics: &Self::SufficientStatistics,
        weight: f64,
    ) -> Result<(), Error> {
        self.sufficient_statistics =
            &self.sufficient_statistics * (1.0 - weight) + sufficient_statistics * weight;
        Ok(())
    }

    fn merge(
        sufficient_statistics: &[&Self::SufficientStatistics],
        weights: &[f64],
    ) -> Result<Self::SufficientStatistics, Error> {
        Ok(sufficient_statistics
            .iter()
            .zip(weights.iter())
            .map(|(s, w)| *s * *w)
            .reduce(|s1, s2| &s1 + &s2)
            .unwrap())
    }

    fn expect_rand(&self, data: &Self::DataIn<'_>, k: usize) -> Result<Self::Likelihood, Error> {
        generate_random_expections(data, k)
    }
}
use tracing::info;

/// FIXME compute the likelihood.. is it just the sum of all?
impl Latent<Finite> for Finite {
    // fn join(
    //     likelihood_a: &<Finite as Parametrizable>::Likelihood,
    //     likelihood_b: &<Finite as Parametrizable>::Likelihood,
    // ) -> Result<(<Finite as Parametrizable>::Likelihood, f64), Error> {
    // }

    fn expect(
        &self,
        data: &<Finite as Parametrizable>::DataIn<'_>,
        likelihood_b: &<Finite as Parametrizable>::Likelihood,
    ) -> Result<(<Finite as Parametrizable>::Likelihood, AvgLLH), Error> {
        let likelihood_a = Parametrizable::expect(self, data)?.0;
        info!(%likelihood_a);
        let log_weighted = likelihood_a + likelihood_b; // n x k
        info!(%likelihood_b);
        let weighted = log_weighted.mapv(|x| x.exp()); // sum?
        let s = weighted.shape();
        info!("{}x{}", s[0], s[1]);
        let log_weighted_norm = weighted.sum_axis(Axis(1)).mapv(|x| x.ln());
        let s = log_weighted_norm.shape();
        info!("{}", s[0]);

        let log_responsibilities = log_weighted - &log_weighted_norm.slice(s![.., NewAxis]);

        Ok((
            log_responsibilities.mapv(|x| x.exp()),
            AvgLLH(log_weighted_norm.mean().unwrap()),
        ))
    }
}

// #[cfg(test)]
// mod tests {
//     use super::*;

//     #[test]
//     fn it_works() {
//         // let result = add(2, 2);
//         // assert_eq!(result, 4);
//     }
// }