abd-clam 0.21.0

Clustered Learning of Approximate Manifolds
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

//! Linear search for the k nearest neighbors of a query.

use core::cmp::Ordering;

use distances::Number;

use crate::Dataset;

/// Linear search for the nearest neighbors of a query.
///
/// # Arguments
///
/// * `data` - The dataset to search.
/// * `query` - The query to search around.
/// * `k` - The number of neighbors to search for.
/// * `indices` - The indices to search.
///
/// # Returns
///
/// A vector of 2-tuples, where the first element is the index of the instance
/// and the second element is the distance from the query to the instance.
pub fn search<T, U, D>(data: &D, query: T, k: usize, indices: &[usize]) -> Vec<(usize, U)>
where
    T: Send + Sync + Copy,
    U: Number,
    D: Dataset<T, U>,
{
    let distances = data.query_to_many(query, indices);
    let mut hits = indices.iter().copied().zip(distances.into_iter()).collect::<Vec<_>>();
    hits.sort_by(|(_, a), (_, b)| a.partial_cmp(b).unwrap_or(Ordering::Less));
    hits[..k].to_vec()
}

#[cfg(test)]
mod tests {

    use distances::vectors::euclidean;
    use symagen::random_data;

    use crate::{Cakes, PartitionCriteria, VecDataset};

    #[test]
    fn linear() {
        let (cardinality, dimensionality) = (1_000, 10);
        let (min_val, max_val) = (-1.0, 1.0);
        let seed = 42;

        let data = random_data::random_f32(cardinality, dimensionality, min_val, max_val, seed);
        let data = data.iter().map(Vec::as_slice).collect::<Vec<_>>();
        let data = VecDataset::new("knn-test".to_string(), data, euclidean::<_, f32>, false);

        let query = random_data::random_f32(1, dimensionality, min_val, max_val, seed * 2);
        let query = query[0].as_slice();

        let criteria = PartitionCriteria::default();
        let model = Cakes::new(data, Some(seed), criteria);
        let tree = model.tree();

        for k in [100, 10, 1] {
            let linear_nn = super::search(tree.data(), query, k, tree.indices());

            assert_eq!(
                linear_nn.len(),
                k,
                "Linear search returned {} neighbors instead of {}",
                linear_nn.len(),
                k
            );
        }
    }
}