light_curve_feature/features/
eta.rs1use crate::evaluator::*;
2use itertools::Itertools;
3
4macro_const! {
5 const DOC: &str = r"
6Von Neummann $\eta$
7
8$$
9\eta \equiv \frac1{(N - 1)\\,\sigma_m^2} \sum_{i=0}^{N-2}(m_{i+1} - m_i)^2,
10$$
11where $N$ is the number of observations,
12$\sigma_m = \sqrt{\sum_i (m_i - \langle m \rangle)^2 / (N-1)}$ is the magnitude standard deviation.
13
14- Depends on: **magnitude**
15- Minimum number of observations: **2**
16- Number of features: **1**
17
18Kim et al. 2014, [DOI:10.1051/0004-6361/201323252](https://doi.org/10.1051/0004-6361/201323252)
19";
20}
21
22#[doc = DOC!()]
23#[derive(Clone, Default, Debug, Serialize, Deserialize, JsonSchema)]
24pub struct Eta {}
25
26impl Eta {
27 pub fn new() -> Self {
28 Self {}
29 }
30
31 pub const fn doc() -> &'static str {
32 DOC
33 }
34}
35
36lazy_info!(
37 ETA_INFO,
38 Eta,
39 size: 1,
40 min_ts_length: 2,
41 t_required: false,
42 m_required: true,
43 w_required: false,
44 sorting_required: true,
45);
46
47impl FeatureNamesDescriptionsTrait for Eta {
48 fn get_names(&self) -> Vec<&str> {
49 vec!["eta"]
50 }
51
52 fn get_descriptions(&self) -> Vec<&str> {
53 vec!["Von Neummann eta-coefficient for magnitude sample"]
54 }
55}
56
57impl<T> FeatureEvaluator<T> for Eta
58where
59 T: Float,
60{
61 fn eval(&self, ts: &mut TimeSeries<T>) -> Result<Vec<T>, EvaluatorError> {
62 self.check_ts_length(ts)?;
63 let m_std2 = get_nonzero_m_std2(ts)?;
64 let value =
65 ts.m.as_slice()
66 .iter()
67 .tuple_windows()
68 .map(|(&a, &b)| (b - a).powi(2))
69 .sum::<T>()
70 / (ts.lenf() - T::one())
71 / m_std2;
72 Ok(vec![value])
73 }
74}
75
76#[cfg(test)]
77#[allow(clippy::unreadable_literal)]
78#[allow(clippy::excessive_precision)]
79mod tests {
80 use super::*;
81 use crate::tests::*;
82
83 check_feature!(Eta);
84
85 feature_test!(
86 eta,
87 [Eta::new()],
88 [1.11338],
89 [1.0_f32, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 109.0],
90 );
91}