solvr 0.2.0-beta.2

Advanced computing library for real-world problem solving - optimization, differential equations, interpolation, statistics, and more
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

//! CPU implementation of HDBSCAN clustering.

use crate::cluster::impl_generic::hdbscan_impl;
use crate::cluster::traits::hdbscan::{HdbscanAlgorithms, HdbscanOptions, HdbscanResult};
use numr::error::Result;
use numr::runtime::cpu::{CpuClient, CpuRuntime};
use numr::tensor::Tensor;

impl HdbscanAlgorithms<CpuRuntime> for CpuClient {
    fn hdbscan(
        &self,
        data: &Tensor<CpuRuntime>,
        options: &HdbscanOptions,
    ) -> Result<HdbscanResult<CpuRuntime>> {
        hdbscan_impl(self, data, options)
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use numr::runtime::cpu::CpuDevice;

    fn setup() -> (CpuClient, CpuDevice) {
        let device = CpuDevice::new();
        let client = CpuClient::new(device.clone());
        (client, device)
    }

    #[test]
    fn test_hdbscan_basic() {
        let (client, device) = setup();

        // Two well-separated clusters with enough points for min_cluster_size
        #[rustfmt::skip]
        let data = Tensor::<CpuRuntime>::from_slice(
            &[
                0.0, 0.0,
                0.1, 0.1,
                0.2, 0.0,
                0.0, 0.2,
                0.1, 0.0,
                10.0, 10.0,
                10.1, 10.1,
                10.2, 10.0,
                10.0, 10.2,
                10.1, 10.0,
            ],
            &[10, 2],
            &device,
        );

        let options = HdbscanOptions {
            min_cluster_size: 3,
            min_samples: Some(2),
            ..Default::default()
        };

        let result = client.hdbscan(&data, &options).unwrap();
        assert_eq!(result.labels.shape(), &[10]);
        assert_eq!(result.probabilities.shape(), &[10]);
    }

    #[test]
    fn test_hdbscan_cluster_separation() {
        let (client, device) = setup();

        // Two clearly separated clusters — first 5 near origin, last 5 near (100,100)
        #[rustfmt::skip]
        let data = Tensor::<CpuRuntime>::from_slice(
            &[
                0.0, 0.0, 0.1, 0.1, 0.2, 0.0, 0.0, 0.2, 0.1, 0.0,
                100.0, 100.0, 100.1, 100.1, 100.2, 100.0, 100.0, 100.2, 100.1, 100.0,
            ],
            &[10, 2],
            &device,
        );

        let options = HdbscanOptions {
            min_cluster_size: 3,
            min_samples: Some(2),
            ..Default::default()
        };

        let result = client.hdbscan(&data, &options).unwrap();
        let labels: Vec<f64> = result.labels.to_vec();
        // Points in same group should have same label
        let first_group: Vec<f64> = labels[0..5].to_vec();
        let second_group: Vec<f64> = labels[5..10].to_vec();
        // All in first group should match
        assert!(first_group.iter().all(|&l| l == first_group[0]));
        // All in second group should match
        assert!(second_group.iter().all(|&l| l == second_group[0]));
        // Two groups should differ (unless both noise)
        if first_group[0] >= 0.0 && second_group[0] >= 0.0 {
            assert_ne!(first_group[0], second_group[0]);
        }
    }

    #[test]
    fn test_hdbscan_noise_detection() {
        let (client, device) = setup();

        // Dense cluster + isolated outlier
        #[rustfmt::skip]
        let data = Tensor::<CpuRuntime>::from_slice(
            &[
                0.0, 0.0, 0.1, 0.1, 0.2, 0.0, 0.0, 0.2, 0.1, 0.0,
                50.0, 50.0,  // outlier
            ],
            &[6, 2],
            &device,
        );

        let options = HdbscanOptions {
            min_cluster_size: 3,
            min_samples: Some(2),
            ..Default::default()
        };

        let result = client.hdbscan(&data, &options).unwrap();
        let labels: Vec<f64> = result.labels.to_vec();
        // Outlier should be noise (-1) or at least different from the cluster
        let cluster_label = labels[0];
        if cluster_label >= 0.0 {
            // The outlier should be noise
            assert!(labels[5] < 0.0 || labels[5] != cluster_label);
        }
    }
}