rlst 0.10.0

A Rust native linear algebra library.
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

//! Working with a distributed CSR matrix.

fn main() {
    use mpi::traits::Root;
    use mpi::{self, traits::Communicator};
    use rlst::dense::array::DynArray;
    use rlst::{
        self, Shape,
        io::matrix_market::{read_array_mm, read_coordinate_mm},
    };
    use rlst::{AijIteratorByValue, AsMatrixApply, FromAijDistributed, assert_array_relative_eq};
    use std::path::PathBuf;
    use std::rc::Rc;

    fn relative_file(filename: &str) -> String {
        let file = PathBuf::from(file!());
        let dir = file.parent().unwrap();
        format!("{}/{filename}", dir.display())
    }

    let universe = mpi::initialize().unwrap();
    let world = universe.world();
    let rank = world.rank();

    let (dist_mat, dist_x) = {
        if rank == 0 {
            println!("Loading CSR matrix example on process {rank}");

            // Read the sparse matrix in matrix market format.
            let sparse_mat = read_coordinate_mm::<f64>(&relative_file("mat_507_313.mm")).unwrap();

            // Communicate the shape of the matrix to all processes.

            let mut shape = sparse_mat.shape();

            world.this_process().broadcast_into(shape.as_mut_slice());

            // We create the domain layout and range layout.

            let domain_layout = Rc::new(
                rlst::distributed_tools::IndexLayout::from_equidistributed_chunks(
                    shape[1], 1, &world,
                ),
            );
            let range_layout = Rc::new(
                rlst::distributed_tools::IndexLayout::from_equidistributed_chunks(
                    shape[0], 1, &world,
                ),
            );

            // We now distribute the matrix across processes.
            let dist_mat = rlst::sparse::distributed_csr_mat::DistributedCsrMatrix::from_aij_iter(
                domain_layout.clone(),
                range_layout.clone(),
                sparse_mat.iter_aij_value(),
            );

            //  We now read the vector x and distribute it across.

            let x = read_array_mm::<f64>(&relative_file("x_313.mm")).unwrap();
            let dist_x = x.slice(1, 0).scatter_from_one_root(domain_layout.clone());

            (dist_mat, dist_x)
        } else {
            // Other processes just need to receive the shape and the distributed matrix.

            let mut shape = [0, 0];

            world
                .process_at_rank(0)
                .broadcast_into(shape.as_mut_slice());

            let domain_layout = Rc::new(
                rlst::distributed_tools::IndexLayout::from_equidistributed_chunks(
                    shape[1], 1, &world,
                ),
            );
            let range_layout = Rc::new(
                rlst::distributed_tools::IndexLayout::from_equidistributed_chunks(
                    shape[0], 1, &world,
                ),
            );

            let dist_mat =
                rlst::sparse::distributed_csr_mat::DistributedCsrMatrix::<'_, f64, _>::from_aij(
                    domain_layout.clone(),
                    range_layout.clone(),
                    Default::default(),
                    Default::default(),
                    Default::default(),
                );

            let dist_x = DynArray::<f64, 1>::scatter_from_one(0, domain_layout.clone());

            (dist_mat, dist_x)
        }
    };

    // We now apply the  distributed matrix vector multiplication.

    let mut dist_y = rlst::dist_vec!(f64, dist_mat.range_layout().clone());
    dist_mat.apply(1.0, &dist_x, 0.0, &mut dist_y);

    // We now gather the result on process 0 and compare to the expected result.

    if rank == 0 {
        let result = dist_y.gather_to_one_root();

        // Read the expected result vector in matrix market format and slice to one dimension.
        let y_expected = read_array_mm::<f64>(&relative_file("/y_507.mm"))
            .unwrap()
            .slice(1, 0);

        assert_array_relative_eq!(result, y_expected, 1E-10);

        println!("Test passed.")
    } else {
        dist_y.gather_to_one(0);
    }
}