distributed_sddmm/serial_conjgrad.cpp at master · PASSIONLab/distributed_sddmm · GitHub

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
#include <iostream>
#include <memory>
#include <Eigen/Dense>
#include <cassert>
#include "mpi.h"
#include "io_utils.h"
#include "sparse_kernels.h"
#include "als_conjugate_gradients.h"
#include "CombBLAS/CombBLAS.h"

using namespace std;
using namespace Eigen;

class SingleNodeALS : public ALS_CG {
public:
    // This constructor tests with a random matrix.
    spmat_local_t S;
    VectorXd ground_truth;
    int R;

    StandardKernel spOps;

    void initialize_dense_matrix(DenseMatrix &X) {
        X.setRandom();
        X /= X.cols();
    }

    SingleNodeALS(int logM, int nnz_per_row, int R) {
        // Generate latent factor Matrices
        this->R = R;
        int n = 1 << logM;

        // Generate two random sets of latent factors
        DenseMatrix Agt(n, R);
        DenseMatrix Bgt(n, R);

        initialize_dense_matrix(Agt);
        initialize_dense_matrix(Bgt);

        // Generate a random sparse matrix using CombBLAS
        shared_ptr<CommGrid> grid;
        grid.reset(new CommGrid(MPI_COMM_WORLD, 1, 1));

        VectorXd SValues; // We discard and generate ourselves

        generateRandomMatrix(logM,
            nnz_per_row,
            grid,
            S,
            SValues
        );

        // Compute a ground truth using an SDDMM, setting all sparse values to 1
        VectorXd ones = VectorXd::Constant(S.local_nnz, 1.0);
        new (&ground_truth) VectorXd(S.local_nnz);

        spOps.sddmm_local(S,
                    ones,
                    Agt,
                    Bgt,
                    ground_truth,
                    0,
                    S.local_nnz);
    }

    void computeRHS(MatMode matrix_to_optimize,
                            DenseMatrix &rhs) {
        if(matrix_to_optimize == Amat) {
            spOps.spmm_local(S, ground_truth, rhs, B, Amat, 0, S.local_nnz);
        }
        else {
            spOps.spmm_local(S, ground_truth, A, rhs, Bmat, 0, S.local_nnz);
        }
    }

    void initializeEmbeddings() {
        new (&A) DenseMatrix(S.nrows, R);
        new (&B) DenseMatrix(S.ncols, R);
        initialize_dense_matrix(A);
        initialize_dense_matrix(B);
    }

    ~SingleNodeALS() {
        // Empty destructor
    }

    double computeResidual() {
        VectorXd ones = VectorXd::Constant(S.local_nnz, 1.0);
        VectorXd sddmm_result = VectorXd::Zero(S.local_nnz);

        spOps.sddmm_local(S,
                    ones,
                    A,
                    B,
                    sddmm_result,
                    0,
                    S.local_nnz);

        return (sddmm_result - ground_truth).norm();
    }

    void computeQueries(
                        DenseMatrix &A,
                        DenseMatrix &B,
                        MatMode matrix_to_optimize,
                        DenseMatrix &result) {

        double lambda = 1e-7;

        VectorXd ones = VectorXd::Constant(S.local_nnz, 1.0);
        result.setZero();
        VectorXd sddmm_result = VectorXd::Zero(S.local_nnz);

        spOps.sddmm_local(S,
                    ones,
                    A,
                    B,
                    sddmm_result,
                    0,
                    S.local_nnz);

        if(matrix_to_optimize == Amat) {
            spOps.spmm_local(S, sddmm_result, result, B, Amat, 0, S.local_nnz);
            result += lambda * A;
        }
        else if(matrix_to_optimize == Bmat) {
            spOps.spmm_local(S, sddmm_result, A, result, Bmat, 0, S.local_nnz);
            result += lambda * B;
        }
    }
};

int main(int argc, char** argv) {
    MPI_Init(&argc, &argv);

    SingleNodeALS test(12, 8, 16);
    test.run_cg(20);

    MPI_Finalize();
}