/** load a matrix market file into a matrix */
void load_matrix_market_matrix(const std::string & filename, int offset, int D){
MM_typecode matcode;
uint i,I,J;
double val;
uint rows, cols;
size_t nnz;
FILE * f = open_file(filename.c_str() ,"r");
int rc = mm_read_banner(f, &matcode);
if (rc != 0)
logstream(LOG_FATAL)<<"Failed to load matrix market banner in file: " << filename << std::endl;
if (mm_is_sparse(matcode)){
int rc = mm_read_mtx_crd_size(f, &rows, &cols, &nnz);
if (rc != 0)
logstream(LOG_FATAL)<<"Failed to load matrix market banner in file: " << filename << std::endl;
}
else { //dense matrix
rc = mm_read_mtx_array_size(f, &rows, &cols);
if (rc != 0)
logstream(LOG_FATAL)<<"Failed to load matrix market banner in file: " << filename << std::endl;
nnz = rows * cols;
}
if (D != (int)cols)
logstream(LOG_FATAL)<<"Wrong matrix size detected, command line argument should be --D=" << D << " instead of : " << cols << std::endl;
for (i=0; i<nnz; i++){
if (mm_is_sparse(matcode)){
rc = fscanf(f, "%u %u %lg\n", &I, &J, &val);
if (rc != 3)
logstream(LOG_FATAL)<<"Error reading input line " << i << std::endl;
I--; J--;
assert(I >= 0 && I < rows);
assert(J >= 0 && J < cols);
//set_val(a, I, J, val);
latent_factors_inmem[I+offset].set_val(J,val);
}
else {
rc = fscanf(f, "%lg", &val);
if (rc != 1)
logstream(LOG_FATAL)<<"Error reading nnz " << i << std::endl;
I = i / cols;
J = i % cols;
latent_factors_inmem[I+offset].set_val(J, val);
}
}
logstream(LOG_INFO) << "Factors from file: loaded matrix of size " << rows << " x " << cols << " from file: " << filename << " total of " << nnz << " entries. "<< i << std::endl;
fclose(f);
}
/* READ_MTX - read symmetric sparse matrix in MatrixMarket format
*/
void read_MTX_SSS(char *fname, int *n,
double **va, double **da, int **ja, int **ia) {
int m, nz, ret_code, i;
double *v_coo;
int *i_coo, *j_coo;
MM_typecode matcode;
FILE *f;
f = fopen(fname, "r");
assert(f != NULL);
ret_code = mm_read_banner(f, &matcode);
assert(ret_code == 0);
assert(mm_is_real(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) && mm_is_symmetric(matcode));
ret_code = mm_read_mtx_crd_size(f, &m, n, &nz);
assert(ret_code == 0);
assert(m == *n);
/* read COO format */
i_coo = (int *)malloc(nz * sizeof(int));
j_coo = (int *)malloc(nz * sizeof(int));
v_coo = (double *)malloc(nz * sizeof(double));
assert(i_coo && j_coo && v_coo);
for (i = 0; i < nz; i ++) {
fscanf(f, "%d %d %lg\n", &i_coo[i], &j_coo[i], &v_coo[i]);
i_coo[i]--; /* adjust from 1-based to 0-based */
j_coo[i]--;
}
fclose(f);
/* convert to SSS format */
convert_COO_SSS(*n, nz, i_coo, j_coo, v_coo, ia, ja, va, da);
free(i_coo); free(j_coo); free(v_coo);
}
/**
compute validation rmse
*/
void validation_rmse3(float (*prediction_func)(const vertex_data & user, const vertex_data & movie, const vertex_data & time, float rating, double & prediction)
,graphchi_context & gcontext,int tokens_per_row = 4) {
int ret_code;
MM_typecode matcode;
FILE *f;
size_t nz;
if ((f = fopen(validation.c_str(), "r")) == NULL) {
std::cout<<std::endl;
return; //missing validaiton data, nothing to compute
}
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner. File: " << validation << std::endl;
if (mm_is_complex(matcode) || !mm_is_sparse(matcode))
logstream(LOG_FATAL) << "Sorry, this application does not support complex values and requires a sparse matrix." << std::endl;
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &Me, &Ne, &nz)) !=0) {
logstream(LOG_FATAL) << "Failed reading matrix size: error=" << ret_code << std::endl;
}
if ((M > 0 && N > 0) && (Me != M || Ne != N))
logstream(LOG_FATAL)<<"Input size of validation matrix must be identical to training matrix, namely " << M << "x" << N << std::endl;
Le = nz;
last_validation_rmse = dvalidation_rmse;
dvalidation_rmse = 0;
int I, J;
double val, time = 1.0;
for (size_t i=0; i<nz; i++)
{
int rc;
rc = fscanf(f, "%d %d %lg %lg\n", &I, &J, &time, &val);
if (rc != tokens_per_row)
logstream(LOG_FATAL)<<"Error when reading input file on line: " << i << " . should have" << tokens_per_row << std::endl;
if (val < minval || val > maxval)
logstream(LOG_FATAL)<<"Value is out of range: " << val << " should be: " << minval << " to " << maxval << std::endl;
if ((uint)time > K)
logstream(LOG_FATAL)<<"Third column value time should be smaller than " << K << " while observed " << time << " in line : " << i << std::endl;
I--; /* adjust from 1-based to 0-based */
J--;
double prediction;
(*prediction_func)(latent_factors_inmem[I], latent_factors_inmem[J+M], latent_factors_inmem[M+N+(uint)time], val, prediction);
dvalidation_rmse += pow(prediction - val, 2);
}
fclose(f);
assert(Le > 0);
dvalidation_rmse = sqrt(dvalidation_rmse / (double)Le);
std::cout<<" Validation RMSE: " << std::setw(10) << dvalidation_rmse << std::endl;
if (halt_on_rmse_increase && dvalidation_rmse > last_validation_rmse && gcontext.iteration > 0){
logstream(LOG_WARNING)<<"Stopping engine because of validation RMSE increase" << std::endl;
gcontext.set_last_iteration(gcontext.iteration);
}
}
void convert_2_vec(const char * filename, shotgun_data * prob )
{
int ret_code;
MM_typecode matcode;
FILE *f;
int M, N, nz;
int i;
if ((f = fopen(filename, "r")) == NULL) {
printf("Could not file vector input file : %s.\n", filename);
exit(1);
}
if (mm_read_banner(f, &matcode) != 0)
{
printf("Could not process Matrix Market banner in input file %s.\n", filename);
exit(1);
}
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
exit(1);
}
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0){
printf("Could not process Matrix Market size in input file: %s.\n", filename);
exit(1);
}
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
prob->y.reserve(nz);
int I,J;
double val;
for (i=0; i<nz; i++)
{
fscanf(f, "%d %d %lg\n", &I, &J, &val);
I--; /* adjust from 1-based to 0-based */
J--;
assert(J==0);
prob->y.push_back(val);
}
if (f !=stdin) fclose(f);
}
int mm_read_mtx_crd(char *fname, int *M, int *N, int *nz, int **I, int **J,
double **val, MM_typecode *matcode)
{
int ret_code;
FILE *f;
if (strcmp(fname, "stdin") == 0) f=stdin;
else
if ((f = fopen(fname, "r")) == NULL)
return MM_COULD_NOT_READ_FILE;
if ((ret_code = mm_read_banner(f, matcode)) != 0)
return ret_code;
if (!(mm_is_valid(*matcode) && mm_is_sparse(*matcode) &&
mm_is_matrix(*matcode)))
return MM_UNSUPPORTED_TYPE;
if ((ret_code = mm_read_mtx_crd_size(f, M, N, nz)) != 0)
return ret_code;
//*I = (int *) malloc(*nz * sizeof(int));
//*J = (int *) malloc(*nz * sizeof(int));
//*val = NULL;
*I = new int[*nz];
*J = new int[*nz];
*val = 0;
if (mm_is_complex(*matcode))
{
//*val = (double *) malloc(*nz * 2 * sizeof(double));
*val = new double[2*(*nz)];
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_real(*matcode))
{
//*val = (double *) malloc(*nz * sizeof(double));
*val = new double[*nz];
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_pattern(*matcode))
{
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
if (f != stdin) fclose(f);
return 0;
}
char *mm_typecode_to_str(MM_typecode matcode)
{
char buffer[MM_MAX_LINE_LENGTH];
char *types[4];
char *mm_strdup(const char *);
int error =0;
/* check for MTX type */
if (mm_is_matrix(matcode))
types[0] = MM_MTX_STR;
else
error=1;
/* check for CRD or ARR matrix */
if (mm_is_sparse(matcode))
types[1] = MM_SPARSE_STR;
else
if (mm_is_dense(matcode))
types[1] = MM_DENSE_STR;
else
return NULL;
/* check for element data type */
if (mm_is_real(matcode))
types[2] = MM_REAL_STR;
else
if (mm_is_complex(matcode))
types[2] = MM_COMPLEX_STR;
else
if (mm_is_pattern(matcode))
types[2] = MM_PATTERN_STR;
else
if (mm_is_integer(matcode))
types[2] = MM_INT_STR;
else
return NULL;
/* check for symmetry type */
if (mm_is_general(matcode))
types[3] = MM_GENERAL_STR;
else
if (mm_is_symmetric(matcode))
types[3] = MM_SYMM_STR;
else
if (mm_is_hermitian(matcode))
types[3] = MM_HERM_STR;
else
if (mm_is_skew(matcode))
types[3] = MM_SKEW_STR;
else
return NULL;
sprintf(buffer,"%s %s %s %s", types[0], types[1], types[2], types[3]);
return mm_strdup(buffer);
}
void mm_typecode_to_str(MM_typecode matcode, char * buffer)
{
char type0[20];
char type1[20];
char type2[20];
char type3[20];
int error =0;
/* check for MTX type */
if (mm_is_matrix(matcode))
strcpy(type0, MM_MTX_STR);
else
error=1;
/* check for CRD or ARR matrix */
if (mm_is_sparse(matcode))
strcpy(type1, MM_SPARSE_STR);
else
if (mm_is_dense(matcode))
strcpy(type1, MM_DENSE_STR);
else
return;
/* check for element data type */
if (mm_is_real(matcode))
strcpy(type2, MM_REAL_STR);
else
if (mm_is_complex(matcode))
strcpy(type2, MM_COMPLEX_STR);
else
if (mm_is_pattern(matcode))
strcpy(type2, MM_PATTERN_STR);
else
if (mm_is_integer(matcode))
strcpy(type2, MM_INT_STR);
else
return;
/* check for symmetry type */
if (mm_is_general(matcode))
strcpy(type3, MM_GENERAL_STR);
else
if (mm_is_symmetric(matcode))
strcpy(type3, MM_SYMM_STR);
else
if (mm_is_hermitian(matcode))
strcpy(type3, MM_HERM_STR);
else
if (mm_is_skew(matcode))
strcpy(type3, MM_SKEW_STR);
else
return;
sprintf(buffer,"%s %s %s %s", type0, type1, type2, type3);
return;
}
/** Reads a matrix market file for a symmetric real valued sparse
matrix and returns the matrix in 1-indexed CSR form. */
void read_mm_sym_matrix(FILE* f, MM_typecode mcode,
int n, int nnzs,
double *values, int* col_ind, int *row_ptr
) {
if (!(mm_is_real(mcode) && mm_is_matrix(mcode) &&
mm_is_sparse(mcode) && mm_is_symmetric(mcode)) ) {
printf("First argument must be a symmetric, real-valued, sparse matrix\n");
printf("Market Market type: [%s]\n", mm_typecode_to_str(mcode));
exit(1);
}
// read Matrix Market matrix in COO format
int* I = (int *) malloc(nnzs * sizeof(int));
int *J = (int *) malloc(nnzs * sizeof(int));
double *val = (double *) malloc(nnzs * sizeof(double));
int i;
for (i=0; i<nnzs; i++) {
fscanf(f, "%d %d %lg\n", &I[i], &J[i], &val[i]);
I[i]--;
J[i]--;
}
MKL_INT job[] = {
1, // convert COO to CSR
1, // use 1 based indexing for CSR matrix (required by mkl_dcsrsymv)
0,
0, // Is this used?
nnzs,
0
};
MKL_INT info;
// convert COO matrix to CSR
mkl_dcsrcoo(job,
&n,
values,
col_ind,
row_ptr,
&nnzs,
val,
I,
J,
&info);
if (info != 0) {
printf("CSR to COO conversion failed: not enough memory!\n");
exit(1);
}
}
void test_predictions3(float (*prediction_func)(const vertex_data & user, const vertex_data & movie, const vertex_data & time, float rating, double & prediction)) {
int ret_code;
MM_typecode matcode;
FILE *f;
uint Me, Ne;
size_t nz;
if ((f = fopen(test.c_str(), "r")) == NULL) {
return; //missing validaiton data, nothing to compute
}
FILE * fout = fopen((test + ".predict").c_str(),"w");
if (fout == NULL)
logstream(LOG_FATAL)<<"Failed to open test prediction file for writing"<<std::endl;
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner. File: " << test << std::endl;
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) || !mm_is_sparse(matcode))
logstream(LOG_FATAL) << "Sorry, this application does not support complex values and requires a sparse matrix." << std::endl;
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &Me, &Ne, &nz)) !=0) {
logstream(LOG_FATAL) << "Failed reading matrix size: error=" << ret_code << std::endl;
}
if ((M > 0 && N > 0 ) && (Me != M || Ne != N))
logstream(LOG_FATAL)<<"Input size of test matrix must be identical to training matrix, namely " << M << "x" << N << std::endl;
mm_write_banner(fout, matcode);
mm_write_mtx_crd_size(fout ,M,N,nz);
for (uint i=0; i<nz; i++)
{
int I, J;
double val;
int time;
int rc = fscanf(f, "%d %d %d %lg\n", &I, &J, &time, &val);
if (rc != 4)
logstream(LOG_FATAL)<<"Error when reading input file: " << i << std::endl;
I--; /* adjust from 1-based to 0-based */
J--;
double prediction;
(*prediction_func)(latent_factors_inmem[I], latent_factors_inmem[J+M], latent_factors_inmem[time+M+N], 1, prediction);
fprintf(fout, "%d %d %12.8lg\n", I+1, J+1, prediction);
}
fclose(f);
fclose(fout);
logstream(LOG_INFO)<<"Finished writing " << nz << " predictions to file: " << test << ".predict" << std::endl;
}
int mm_read_mtx_crd(char *fname, int *M, int *N, int *nz, int **I, int **J,
double **val, MM_typecode *matcode)
{
int ret_code;
ZOLTAN_FILE* f;
if ((f = ZOLTAN_FILE_open(fname, "r", STANDARD)) == NULL)
return MM_COULD_NOT_READ_FILE;
if ((ret_code = mm_read_banner(f, matcode)) != 0)
return ret_code;
if (!(mm_is_valid(*matcode) && mm_is_sparse(*matcode) &&
mm_is_matrix(*matcode)))
return MM_UNSUPPORTED_TYPE;
if ((ret_code = mm_read_mtx_crd_size(f, M, N, nz)) != 0)
return ret_code;
*I = (int *) malloc(*nz * sizeof(int));
*J = (int *) malloc(*nz * sizeof(int));
*val = NULL;
if (mm_is_complex(*matcode))
{
*val = (double *) malloc(*nz * 2 * sizeof(double));
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_real(*matcode))
{
*val = (double *) malloc(*nz * sizeof(double));
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_pattern(*matcode))
{
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
ZOLTAN_FILE_close(f);
return 0;
}
void read_matrix_market_banner_and_size(FILE * f, MM_typecode & matcode, uint & Me, uint & Ne, size_t & nz, const std::string & filename){
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner. File: " << filename << std::endl;
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) || !mm_is_sparse(matcode))
logstream(LOG_FATAL) << "Sorry, this application does not support complex values and requires a sparse matrix." << std::endl;
/* find out size of sparse matrix .... */
if (mm_read_mtx_crd_size(f, &Me, &Ne, &nz) != 0) {
logstream(LOG_FATAL) << "Failed reading matrix size: error" << std::endl;
}
}
/*** fread MatrixMarket format file ***/
mm_real *
mm_real_fread (FILE *fp)
{
MM_typecode typecode;
mm_real *x;
if (mm_read_banner (fp, &typecode) != 0) error_and_exit ("mm_real_fread", "failed to read mm_real.", __FILE__, __LINE__);
if (!is_type_supported (typecode)) {
char msg[128];
sprintf (msg, "matrix type does not supported :[%s].", mm_typecode_to_str (typecode));
error_and_exit ("mm_real_fread", msg, __FILE__, __LINE__);
}
x = (mm_is_sparse (typecode)) ? mm_real_fread_sparse (fp, typecode) : mm_real_fread_dense (fp, typecode);
if (!x) error_and_exit ("mm_real_fread", "failed to read mm_real.", __FILE__, __LINE__);
if (mm_real_is_symmetric (x) && x->m != x->n) error_and_exit ("mm_real_fread", "symmetric matrix must be square.", __FILE__, __LINE__);
return x;
}
bool loadMmProperties(int *rowsCount,
int *columnsCount,
int *nonZerosCount,
bool *isStoredSparse,
int* matrixStorage,
int* matrixType,
FILE *file)
{
MM_typecode matcode;
// supports only valid matrices
if ((mm_read_banner(file, &matcode) != 0)
|| (!mm_is_matrix(matcode))
|| (!mm_is_valid(matcode)))
return false;
if ( mm_read_mtx_crd_size(file, rowsCount, columnsCount, nonZerosCount) != 0 )
return false;
// is it stored sparse?
if (mm_is_sparse(matcode))
*isStoredSparse = true;
else
*isStoredSparse = false;
if (mm_is_integer(matcode))
*matrixStorage = MATRIX_STORAGE_INTEGER;
else if (mm_is_real(matcode))
*matrixStorage = MATRIX_STORAGE_REAL;
else if (mm_is_complex(matcode))
*matrixStorage = MATRIX_STORAGE_COMPLEX;
else if (mm_is_pattern(matcode))
*matrixStorage = MATRIX_STORAGE_PATTERN;
if (mm_is_general(matcode))
*matrixType = MATRIX_TYPE_GENERAL;
else if (mm_is_symmetric(matcode))
*matrixType = MATRIX_TYPE_SYMMETRIC;
else if (mm_is_skew(matcode))
*matrixType = MATRIX_TYPE_SKEW;
else if (mm_is_hermitian(matcode))
*matrixType = MATRIX_TYPE_HERMITIAN;
return true;
}
int DenseMatrix_mm_read_strassen(DenseMatrix *m, char *file_name, int *nr_rows, int *nr_cols) {
int res;
MM_typecode matcode;
FILE *file;
file = fopen(file_name, "r");
CHECK_NULL_RETURN(file);
res = mm_read_banner(file, &matcode);
CHECK_ZERO_ERROR_RETURN(res, "Failed to read matrix code");
CHECK_ERROR_RETURN(!mm_is_matrix(matcode), "File is not a matrix", 1);
if (mm_is_sparse(matcode)) {
int nr_sparse_elements;
res = mm_read_mtx_crd_size(file, nr_rows, nr_cols, &nr_sparse_elements);
CHECK_ZERO_ERROR_RETURN(res, "Failed to read sparse mm dimensions");
int dim = pow2dim(*nr_rows, *nr_cols);
// Initialzie matrix to zero to fill in with sparse elements
res = DenseMatrix_init_zero(m, dim, dim);
CHECK_ZERO_ERROR_RETURN(res, "Failed to allocate memory for mm matrix");
res = DenseMatrix_parse_mm_sparse(m, file, nr_sparse_elements);
} else if (mm_is_dense(matcode)) {
res = mm_read_mtx_array_size(file, nr_rows, nr_cols);
CHECK_ZERO_ERROR_RETURN(res, "Failed to read dense mm dimensions");
int dim = pow2dim(*nr_rows, *nr_cols);
res = DenseMatrix_init_zero(m, dim, dim);
CHECK_ZERO_ERROR_RETURN(res, "Failed to allocate memory for mm matrix");
res = DenseMatrix_parse_mm_dense(m, file);
} else {
ERROR("mm matrix code is not supported. Only supports dense and sparse matrices");
}
CHECK_ZERO_ERROR_RETURN(res, "Failed to parse mm file");
return 0;
}
int get_mm_info(const char *file, int *m, int *n, int *nz)
{
FILE *fp;
MM_typecode matcode;
if ((fp = fopen(file, "r")) == NULL)
{
fprintf(stderr,"ERROR: Could not open file: %s\n",file);
exit(1);
}
if (mm_read_banner(fp, &matcode) != 0)
{
fprintf(stderr,"ERROR: Could not process Matrix Market banner.\n");
exit(1);
}
if (!(mm_is_real(matcode) || mm_is_integer(matcode)))
{
fprintf(stderr,"ERROR: Market Market type: [%s] not supported\n",
mm_typecode_to_str(matcode));
exit(1);
}
if (mm_is_sparse(matcode))
{
if (mm_read_mtx_crd_size(fp, m, n, nz) !=0)
{ /* find out size of sparse matrix */
exit(1);
}
}
else
{
if (mm_read_mtx_array_size(fp, m, n) !=0)
{ /* find out size of dense matrix */
exit(1);
}
*nz = -1;
}
fclose(fp);
return 0;
}
/*---------------------------------------------*
* READ COO Matrix Market *
*---------------------------------------------*/
int read_coo_MM(coo_t *coo, options_t *opts) {
char *matfile = opts->fmatname;
MM_typecode matcode;
FILE *p = fopen(matfile,"r");
if (p == NULL) {
printf("Unable to open file %s\n", matfile);
exit(1);
}
/*----------- READ MM banner */
if (mm_read_banner(p, &matcode) != 0) {
printf("Could not process Matrix Market banner.\n");
exit(1);
}
if (!mm_is_valid(matcode)) {
printf("Invalid Matrix Market file.\n");
exit(1);
}
if (!(mm_is_real(matcode) && mm_is_coordinate(matcode)
&& mm_is_sparse(matcode))) {
printf("Only sparse real-valued coordinate \
matrices are supported\n");
exit(1);
}
//------------------------------------------------------------------------
int test_mm_read(std::string filename)
{
int ret_code;
MM_typecode matcode;
FILE *f;
int M, N, nz;
int i, *I, *J;
double *val;
int err=0;
if ((f = fopen(filename.c_str(), "r")) == NULL)
return -1;
if (mm_read_banner(f, &matcode) != 0)
{
printf("Could not process Matrix Market banner.\n");
return -2;
}
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
return -3;
}
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0)
return -4;
/* reseve memory for matrices */
I = (int *) malloc(nz * sizeof(int));
J = (int *) malloc(nz * sizeof(int));
val = (double *) malloc(nz * sizeof(double));
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
for (i=0; i<nz; i++)
{
fscanf(f, "%d %d %lg\n", &I[i], &J[i], &val[i]);
I[i]--; /* adjust from 1-based to 0-based */
J[i]--;
}
if (f !=stdin) fclose(f);
/************************/
/* now write out matrix */
/************************/
fprintf(stdout, "Read full file contents: \n================================\n");
err += mm_write_banner(stdout, matcode);
err += mm_write_mtx_crd_size(stdout, M, N, nz);
for (i=0; i<nz; i++)
{
fprintf(stdout, "%d %d %lg\n", I[i]+1, J[i]+1, val[i]);
}
return err;
}
int mm_read_unsymmetric_sparse(const char *fname, int *M_, int *N_, int *nz_,
double **val_, int **I_, int **J_)
{
FILE *f;
MM_typecode matcode;
int M, N, nz;
int i;
double *val;
int *I, *J;
if ((f = fopen(fname, "r")) == NULL)
return -1;
if (mm_read_banner(f, &matcode) != 0)
{
printf("mm_read_unsymetric: Could not process Matrix Market banner ");
printf(" in file [%s]\n", fname);
return -1;
}
if ( !(mm_is_real(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode)))
{
fprintf(stderr, "Sorry, this application does not support ");
fprintf(stderr, "Market Market type: [%s]\n",
mm_typecode_to_str(matcode));
return -1;
}
/* find out size of sparse matrix: M, N, nz .... */
if (mm_read_mtx_crd_size(f, &M, &N, &nz) !=0)
{
fprintf(stderr, "read_unsymmetric_sparse(): could not parse matrix size.\n");
return -1;
}
*M_ = M;
*N_ = N;
*nz_ = nz;
/* reseve memory for matrices */
I = (int *) malloc(nz * sizeof(int));
J = (int *) malloc(nz * sizeof(int));
val = (double *) malloc(nz * sizeof(double));
*val_ = val;
*I_ = I;
*J_ = J;
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
for (i=0; i<nz; i++)
{
fscanf(f, "%d %d %lg\n", &I[i], &J[i], &val[i]);
I[i]--; /* adjust from 1-based to 0-based */
J[i]--;
}
fclose(f);
return 0;
}
int mm_read_mtx_crd(char *fname, int *M, int *N, int *nz, int **I, int **J,
double **val, MM_typecode *matcode)
/******************************************************************************/
/*
Purpose:
MM_READ_MTX_CRD reads the values in an MM coordinate file.
Discussion:
This function allocates the storage for the arrays.
Modified:
31 October 2008
Parameters:
*/
/*
mm_read_mtx_crd() fills M, N, nz, array of values, and return
type code, e.g. 'MCRS'
if matrix is complex, values[] is of size 2*nz,
(nz pairs of real/imaginary values)
*/
{
int ret_code;
FILE *f;
if (strcmp(fname, "stdin") == 0) f=stdin;
else
if ((f = fopen(fname, "r")) == NULL)
return MM_COULD_NOT_READ_FILE;
if ((ret_code = mm_read_banner(f, matcode)) != 0)
return ret_code;
if (!(mm_is_valid(*matcode) && mm_is_sparse(*matcode) &&
mm_is_matrix(*matcode)))
return MM_UNSUPPORTED_TYPE;
if ((ret_code = mm_read_mtx_crd_size(f, M, N, nz)) != 0)
return ret_code;
*I = (int *) malloc(*nz * sizeof(int));
*J = (int *) malloc(*nz * sizeof(int));
*val = NULL;
if (mm_is_complex(*matcode))
{
*val = (double *) malloc(*nz * 2 * sizeof(double));
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_real(*matcode))
{
*val = (double *) malloc(*nz * sizeof(double));
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
else if (mm_is_pattern(*matcode))
{
ret_code = mm_read_mtx_crd_data(f, *M, *N, *nz, *I, *J, *val,
*matcode);
if (ret_code != 0) return ret_code;
}
if (f != stdin) fclose(f);
return 0;
}
vec load_matrix_market_vector(const std::string & filename, bool optional_field, bool allow_zeros)
{
int ret_code;
MM_typecode matcode;
uint M, N;
size_t i,nz;
logstream(LOG_INFO) <<"Going to read matrix market vector from input file: " << filename << std::endl;
FILE * f = open_file(filename.c_str(), "r", optional_field);
//if optional file not found return
if (f== NULL && optional_field){
return zeros(1);
}
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner." << std::endl;
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
logstream(LOG_FATAL) << "sorry, this application does not support " << std::endl <<
"Market Market type: " << mm_typecode_to_str(matcode) << std::endl;
/* find out size of sparse matrix .... */
if (mm_is_sparse(matcode)){
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0)
logstream(LOG_FATAL) << "failed to read matrix market cardinality size " << std::endl;
}
else {
if ((ret_code = mm_read_mtx_array_size(f, &M, &N))!= 0)
logstream(LOG_FATAL) << "failed to read matrix market vector size " << std::endl;
if (N > M){ //if this is a row vector, transpose
int tmp = N;
N = M;
M = tmp;
}
nz = M*N;
}
vec ret = zeros(M);
uint row,col;
double val;
for (i=0; i<nz; i++)
{
if (mm_is_sparse(matcode)){
int rc = fscanf(f, "%u %u %lg\n", &row, &col, &val);
if (rc != 3){
logstream(LOG_FATAL) << "Failed reading input file: " << filename << "Problm at data row " << i << " (not including header and comment lines)" << std::endl;
}
row--; /* adjust from 1-based to 0-based */
col--;
}
else {
int rc = fscanf(f, "%lg\n", &val);
if (rc != 1){
logstream(LOG_FATAL) << "Failed reading input file: " << filename << "Problm at data row " << i << " (not including header and comment lines)" << std::endl;
}
row = i;
col = 0;
}
//some users have gibrish in text file - better check both I and J are >=0 as well
assert(row >=0 && row< M);
assert(col == 0);
if (val == 0 && !allow_zeros)
logstream(LOG_FATAL)<<"Zero entries are not allowed in a sparse matrix market vector. Use --zero=true to avoid this error"<<std::endl;
//set observation value
ret[row] = val;
}
fclose(f);
logstream(LOG_INFO)<<"Succesfully read a vector of size: " << M << " [ " << nz << "]" << std::endl;
return ret;
}
cs *read_matrix(const char *filename, MM_typecode &matcode)
{
LogInfo("Reading Matrix from " << std::string(filename) << "\n");
FILE *file = fopen(filename, "r");
if (!file)
{
LogError("Error: Cannot read file " << std::string(filename) << "\n");
return NULL;
}
LogInfo("Reading Matrix Market banner...");
if (mm_read_banner(file, &matcode) != 0)
{
LogError("Error: Could not process Matrix Market banner\n");
fclose(file);
return NULL;
}
if (!mm_is_matrix(matcode) || !mm_is_sparse(matcode)
|| mm_is_complex(matcode))
{
LogError(
"Error: Unsupported matrix format - Must be real and sparse\n");
fclose(file);
return NULL;
}
Int M, N, nz;
if ((mm_read_mtx_crd_size(file, &M, &N, &nz)) != 0)
{
LogError("Error: Could not parse matrix dimension and size.\n");
fclose(file);
return NULL;
}
if (M != N)
{
LogError("Error: Matrix must be square.\n");
fclose(file);
return NULL;
}
LogInfo("Reading matrix data...\n");
Int *I = (Int *)SuiteSparse_malloc(static_cast<size_t>(nz), sizeof(Int));
Int *J = (Int *)SuiteSparse_malloc(static_cast<size_t>(nz), sizeof(Int));
double *val
= (double *)SuiteSparse_malloc(static_cast<size_t>(nz), sizeof(double));
if (!I || !J || !val)
{
LogError("Error: Ran out of memory in Mongoose::read_matrix\n");
SuiteSparse_free(I);
SuiteSparse_free(J);
SuiteSparse_free(val);
fclose(file);
return NULL;
}
mm_read_mtx_crd_data(file, M, N, nz, I, J, val, matcode);
fclose(file); // Close the file
for (Int k = 0; k < nz; k++)
{
--I[k];
--J[k];
if (mm_is_pattern(matcode))
val[k] = 1;
}
cs *A = (cs *)SuiteSparse_malloc(1, sizeof(cs));
if (!A)
{
LogError("Error: Ran out of memory in Mongoose::read_matrix\n");
SuiteSparse_free(I);
SuiteSparse_free(J);
SuiteSparse_free(val);
return NULL;
}
A->nzmax = nz;
A->m = M;
A->n = N;
A->p = J;
A->i = I;
A->x = val;
A->nz = nz;
LogInfo("Compressing matrix from triplet to CSC format...\n");
cs *compressed_A = cs_compress(A);
cs_spfree(A);
if (!compressed_A)
{
LogError("Error: Ran out of memory in Mongoose::read_matrix\n");
return NULL;
}
return compressed_A;
}
static void read_mtx_and_return_csr(
int argc, char **argv, int *mm, int *nn, int **ia, int **ja, double **aa)
{
int ret_code;
MM_typecode matcode;
FILE *f;
int m, n, nz;
int i, *I, *J;
double *val;
if (argc < 2)
{
fprintf(stderr, "Usage: %s [martix-market-filename]\n", argv[0]);
exit(1);
}
else
{
printf("\n===================================\n");
printf("mtx file = %s\n", argv[1]);
if ((f = fopen(argv[1], "r")) == NULL)
{
printf("Could not open %s\n", argv[1]);
exit(1);
}
}
if (mm_read_banner(f, &matcode) != 0)
{
printf("Could not process Matrix Market banner.\n");
exit(1);
}
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if ( mm_is_pattern(matcode)
|| mm_is_dense(matcode)
|| mm_is_array(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
exit(1);
}
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
exit(1);
}
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &m, &n, &nz)) !=0)
exit(1);
/* reseve memory for matrices */
I = (int *) malloc(nz * sizeof(int));
J = (int *) malloc(nz * sizeof(int));
val = (double *) malloc(nz * sizeof(double));
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
for (i=0; i<nz; i++)
{
fscanf(f, "%d %d %lg\n", &I[i], &J[i], &val[i]);
I[i]--; /* adjust from 1-based to 0-based */
J[i]--;
}
if (f !=stdin) fclose(f);
if (m != n) exit(1);
*mm = m;
*nn = n;
*ia = (int*) malloc (sizeof(int) * (n+1));
*ja = (int*) malloc (sizeof(int) * nz);
*aa = (double*) malloc (sizeof(double) * nz);
coo2csr(n, nz, val, I, J, *aa, *ja, *ia);
free (I);
free (J);
free (val);
}
magma_int_t
magma_d_csr_mtx(
magma_d_matrix *A,
const char *filename,
magma_queue_t queue )
{
char buffer[ 1024 ];
magma_int_t info = 0;
int csr_compressor = 0; // checks for zeros in original file
magma_d_matrix B={Magma_CSR};
magma_index_t *coo_col = NULL;
magma_index_t *coo_row = NULL;
double *coo_val = NULL;
double *new_val = NULL;
magma_index_t* new_row = NULL;
magma_index_t* new_col = NULL;
magma_int_t symmetric = 0;
std::vector< std::pair< magma_index_t, double > > rowval;
FILE *fid = NULL;
MM_typecode matcode;
fid = fopen(filename, "r");
if (fid == NULL) {
printf("%% Unable to open file %s\n", filename);
info = MAGMA_ERR_NOT_FOUND;
goto cleanup;
}
printf("%% Reading sparse matrix from file (%s):", filename);
fflush(stdout);
if (mm_read_banner(fid, &matcode) != 0) {
printf("\n%% Could not process Matrix Market banner: %s.\n", matcode);
info = MAGMA_ERR_NOT_SUPPORTED;
goto cleanup;
}
if (!mm_is_valid(matcode)) {
printf("\n%% Invalid Matrix Market file.\n");
info = MAGMA_ERR_NOT_SUPPORTED;
goto cleanup;
}
if ( ! ( ( mm_is_real(matcode) ||
mm_is_integer(matcode) ||
mm_is_pattern(matcode) ||
mm_is_real(matcode) ) &&
mm_is_coordinate(matcode) &&
mm_is_sparse(matcode) ) )
{
mm_snprintf_typecode( buffer, sizeof(buffer), matcode );
printf("\n%% Sorry, MAGMA-sparse does not support Market Market type: [%s]\n", buffer );
printf("%% Only real-valued or pattern coordinate matrices are supported.\n");
info = MAGMA_ERR_NOT_SUPPORTED;
goto cleanup;
}
magma_index_t num_rows, num_cols, num_nonzeros;
if (mm_read_mtx_crd_size(fid, &num_rows, &num_cols, &num_nonzeros) != 0) {
info = MAGMA_ERR_UNKNOWN;
goto cleanup;
}
A->storage_type = Magma_CSR;
A->memory_location = Magma_CPU;
A->num_rows = num_rows;
A->num_cols = num_cols;
A->nnz = num_nonzeros;
A->fill_mode = MagmaFull;
CHECK( magma_index_malloc_cpu( &coo_col, A->nnz ) );
CHECK( magma_index_malloc_cpu( &coo_row, A->nnz ) );
CHECK( magma_dmalloc_cpu( &coo_val, A->nnz ) );
if (mm_is_real(matcode) || mm_is_integer(matcode)) {
for(magma_int_t i = 0; i < A->nnz; ++i) {
magma_index_t ROW, COL;
double VAL; // always read in a double and convert later if necessary
fscanf(fid, " %d %d %lf \n", &ROW, &COL, &VAL);
if ( VAL == 0 )
csr_compressor = 1;
coo_row[i] = ROW - 1;
coo_col[i] = COL - 1;
coo_val[i] = MAGMA_D_MAKE( VAL, 0.);
}
} else if (mm_is_pattern(matcode) ) {
for(magma_int_t i = 0; i < A->nnz; ++i) {
magma_index_t ROW, COL;
fscanf(fid, " %d %d \n", &ROW, &COL );
coo_row[i] = ROW - 1;
coo_col[i] = COL - 1;
coo_val[i] = MAGMA_D_MAKE( 1.0, 0.);
}
} else if (mm_is_real(matcode) ){
for(magma_int_t i = 0; i < A->nnz; ++i) {
magma_index_t ROW, COL;
double VAL, VALC; // always read in a double and convert later if necessary
fscanf(fid, " %d %d %lf %lf\n", &ROW, &COL, &VAL, &VALC);
coo_row[i] = ROW - 1;
coo_col[i] = COL - 1;
coo_val[i] = MAGMA_D_MAKE( VAL, VALC);
}
// printf(" ...successfully read real matrix... ");
} else {
printf("\n%% Unrecognized data type\n");
info = MAGMA_ERR_NOT_SUPPORTED;
goto cleanup;
}
fclose(fid);
fid = NULL;
printf(" done. Converting to CSR:");
fflush(stdout);
A->sym = Magma_GENERAL;
if( mm_is_symmetric(matcode) ) {
symmetric = 1;
}
if ( mm_is_symmetric(matcode) || mm_is_symmetric(matcode) ) {
// duplicate off diagonal entries
printf("\n%% Detected symmetric case.");
A->sym = Magma_SYMMETRIC;
magma_index_t off_diagonals = 0;
for(magma_int_t i = 0; i < A->nnz; ++i) {
if (coo_row[i] != coo_col[i])
++off_diagonals;
}
magma_index_t true_nonzeros = 2*off_diagonals + (A->nnz - off_diagonals);
//printf("%% total number of nonzeros: %d\n%%", int(A->nnz));
CHECK( magma_index_malloc_cpu( &new_row, true_nonzeros ));
CHECK( magma_index_malloc_cpu( &new_col, true_nonzeros ));
CHECK( magma_dmalloc_cpu( &new_val, true_nonzeros ));
magma_index_t ptr = 0;
for(magma_int_t i = 0; i < A->nnz; ++i) {
if (coo_row[i] != coo_col[i]) {
new_row[ptr] = coo_row[i];
new_col[ptr] = coo_col[i];
new_val[ptr] = coo_val[i];
ptr++;
new_col[ptr] = coo_row[i];
new_row[ptr] = coo_col[i];
new_val[ptr] = (symmetric == 0) ? coo_val[i] : conj(coo_val[i]);
ptr++;
} else {
new_row[ptr] = coo_row[i];
new_col[ptr] = coo_col[i];
new_val[ptr] = coo_val[i];
ptr++;
}
}
magma_free_cpu(coo_row);
magma_free_cpu(coo_col);
magma_free_cpu(coo_val);
coo_row = new_row;
coo_col = new_col;
coo_val = new_val;
A->nnz = true_nonzeros;
//printf("total number of nonzeros: %d\n", A->nnz);
} // end symmetric case
CHECK( magma_dmalloc_cpu( &A->val, A->nnz ));
CHECK( magma_index_malloc_cpu( &A->col, A->nnz ));
CHECK( magma_index_malloc_cpu( &A->row, A->num_rows+1 ));
// original code from Nathan Bell and Michael Garland
for (magma_index_t i = 0; i < num_rows; i++)
(A->row)[i] = 0;
for (magma_index_t i = 0; i < A->nnz; i++)
(A->row)[coo_row[i]]++;
// cumulative sum the nnz per row to get row[]
magma_int_t cumsum;
cumsum = 0;
for(magma_int_t i = 0; i < num_rows; i++) {
magma_index_t temp = (A->row)[i];
(A->row)[i] = cumsum;
cumsum += temp;
}
(A->row)[num_rows] = A->nnz;
// write Aj,Ax into Bj,Bx
for(magma_int_t i = 0; i < A->nnz; i++) {
magma_index_t row_ = coo_row[i];
magma_index_t dest = (A->row)[row_];
(A->col)[dest] = coo_col[i];
(A->val)[dest] = coo_val[i];
(A->row)[row_]++;
}
magma_free_cpu(coo_row);
magma_free_cpu(coo_col);
magma_free_cpu(coo_val);
coo_row = NULL;
coo_col = NULL;
coo_val = NULL;
int last;
last = 0;
for(int i = 0; i <= num_rows; i++) {
int temp = (A->row)[i];
(A->row)[i] = last;
last = temp;
}
(A->row)[A->num_rows] = A->nnz;
// sort column indices within each row
// copy into vector of pairs (column index, value), sort by column index, then copy back
for (magma_index_t k=0; k < A->num_rows; ++k) {
int kk = (A->row)[k];
int len = (A->row)[k+1] - (A->row)[k];
rowval.resize( len );
for( int i=0; i < len; ++i ) {
rowval[i] = std::make_pair( (A->col)[kk+i], (A->val)[kk+i] );
}
std::sort( rowval.begin(), rowval.end(), compare_first );
for( int i=0; i < len; ++i ) {
(A->col)[kk+i] = rowval[i].first;
(A->val)[kk+i] = rowval[i].second;
}
}
if ( csr_compressor > 0) { // run the CSR compressor to remove zeros
//printf("removing zeros: ");
CHECK( magma_dmtransfer( *A, &B, Magma_CPU, Magma_CPU, queue ));
CHECK( magma_d_csr_compressor(
&(A->val), &(A->row), &(A->col),
&B.val, &B.row, &B.col, &B.num_rows, queue ));
B.nnz = B.row[num_rows];
//printf(" remaining nonzeros:%d ", B.nnz);
magma_free_cpu( A->val );
magma_free_cpu( A->row );
magma_free_cpu( A->col );
CHECK( magma_dmtransfer( B, A, Magma_CPU, Magma_CPU, queue ));
//printf("done.\n");
}
A->true_nnz = A->nnz;
printf(" done.\n");
cleanup:
if ( fid != NULL ) {
fclose( fid );
fid = NULL;
}
magma_dmfree( &B, queue );
magma_free_cpu(coo_row);
magma_free_cpu(coo_col);
magma_free_cpu(coo_val);
return info;
}
bool LoadMatrixMarketFile(const std::string& file_path,
SparseMatrix<T>& A,
unsigned int& height,
unsigned int& width,
unsigned int& nnz)
{
std::ifstream infile(file_path);
if (!infile)
return false;
char mm_typecode[4];
// read the matrix market banner (header)
if (0 != mm_read_banner(infile, mm_typecode))
return false;
if (!mm_is_valid(mm_typecode))
return false;
// this reader supports these matrix types:
//
// sparse, real/integer/pattern, general/symm/skew
//
if (!mm_is_sparse(mm_typecode))
{
std::cerr << "Only sparse MatrixMarket files are supported." << std::endl;
return false;
}
if (!mm_is_real(mm_typecode) && !mm_is_integer(mm_typecode) && !mm_is_pattern(mm_typecode))
{
std::cerr << "Only real, integer, and pattern MatrixMarket formats are supported." << std::endl;
return false;
}
if (!mm_is_general(mm_typecode) && !mm_is_symmetric(mm_typecode) && !mm_is_skew(mm_typecode))
{
std::cerr << "Only general, symmetric, and skew-symmetric MatrixMarket formats are supported."
<< std::endl;
return false;
}
// read the number of rows, cols, nonzeros
if (0 != mm_read_mtx_crd_size(infile, height, width, nnz))
{
std::cerr << "could not read matrix coordinate information" << std::endl;
height = width = nnz = 0;
return false;
}
// read the data according to the type
bool is_real = mm_is_real(mm_typecode);
bool is_int = mm_is_integer(mm_typecode);
bool is_symmetric = mm_is_symmetric(mm_typecode);
bool is_skew = mm_is_skew(mm_typecode);
std::string line;
unsigned int reserve_size = nnz;
if (is_symmetric || is_skew)
reserve_size *= 2;
A.Clear();
A.Reserve(height, width, reserve_size);
// load num random entries of A
A.BeginLoad();
unsigned int row, col, count;
if (is_real)
{
double val;
for (count=0; count != nnz; ++count)
{
infile >> row; assert(row >= 1);
infile >> col; assert(col >= 1);
infile >> val;
// convert to 0-based indexing
row -= 1;
col -= 1;
A.Load(row, col, val);
if (row != col)
{
if (is_symmetric)
A.Load(col, row, val);
else if (is_skew)
A.Load(col, row, -val);
}
}
}
else if (is_int)
csr_matrix_t *csr_mm_load(char *filename)
{
int ret_code;
MM_typecode matcode;
FILE *f;
int entry_count;
int i, M, N, nz;
matrix_entry_t *entries;
matrix_entry_t *entry;
int *row_start;
int *col_index;
double *val;
csr_matrix_t *A;
if ((f = fopen(filename, "r")) == NULL)
exit(1);
if (mm_read_banner(f, &matcode) != 0) {
printf("Could not process Matrix Market banner.\n");
exit(1);
}
if (!mm_is_sparse(matcode) || !mm_is_symmetric(matcode) ||
!mm_is_real(matcode)) {
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
exit(1);
}
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) != 0)
exit(1);
/* reserve memory for matrices */
row_start = (int *) malloc((M + 1) * sizeof(int));
col_index = (int *) malloc((2 * nz - M) * sizeof(int));
val = (double *) malloc((2 * nz - M) * sizeof(double));
entries =
(matrix_entry_t *) malloc((2 * nz - M) * sizeof(matrix_entry_t));
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
entry = entries;
entry_count = 0;
for (i = 0; i < nz; i++) {
int row, col;
double val;
fscanf(f, "%d %d %lg\n", &row, &col, &val);
--row; /* adjust to 0-based */
--col;
assert(row >= 0 && col >= 0);
assert(entry_count++ < 2 * nz - M);
entry->i = row;
entry->j = col;
entry->val = val;
++entry;
if (row != col) { /* Fill out the other half... */
assert(entry_count++ < 2 * nz - M);
entry->i = col;
entry->j = row;
entry->val = val;
++entry;
}
}
if (f != stdin)
fclose(f);
/**********************************/
/* now make CSR version of matrix */
/**********************************/
nz = 2 * nz - M;
qsort(entries, nz, sizeof(matrix_entry_t), entry_comparison);
entry = entries;
row_start[0] = 0;
for (i = 0; i < nz; ++i) {
row_start[entry->i + 1] = i + 1;
col_index[i] = entry->j;
val[i] = entry->val;
++entry;
}
free(entries);
A = (csr_matrix_t *) malloc(sizeof(csr_matrix_t));
A->m = M;
A->n = N;
A->nz = nz;
A->row_start = row_start;
A->col_idx = col_index;
A->val = val;
return A;
}
char *mm_typecode_to_str ( MM_typecode matcode )
/******************************************************************************/
/*
Purpose:
MM_TYPECODE_TO_STR converts the internal typecode to an MM header string.
Modified:
31 October 2008
*/
{
char buffer[MM_MAX_LINE_LENGTH];
char *types[4];
char *mm_strdup(const char *);
//int error =0;
/*
check for MTX type
*/
if (mm_is_matrix(matcode))
types[0] = MM_MTX_STR;
// else
// error=1;
/*
check for CRD or ARR matrix
*/
if (mm_is_sparse(matcode))
types[1] = MM_SPARSE_STR;
else
if (mm_is_dense(matcode))
types[1] = MM_DENSE_STR;
else
return NULL;
/*
check for element data type
*/
if (mm_is_real(matcode))
types[2] = MM_REAL_STR;
else
if (mm_is_complex(matcode))
types[2] = MM_COMPLEX_STR;
else
if (mm_is_pattern(matcode))
types[2] = MM_PATTERN_STR;
else
if (mm_is_integer(matcode))
types[2] = MM_INT_STR;
else
return NULL;
/*
check for symmetry type
*/
if (mm_is_general(matcode))
types[3] = MM_GENERAL_STR;
else
if (mm_is_symmetric(matcode))
types[3] = MM_SYMM_STR;
else
if (mm_is_hermitian(matcode))
types[3] = MM_HERM_STR;
else
if (mm_is_skew(matcode))
types[3] = MM_SKEW_STR;
else
return NULL;
sprintf(buffer,"%s %s %s %s", types[0], types[1], types[2], types[3]);
return mm_strdup(buffer);
}
bool MatrixMarket::read_matrix(const char *file_name) {
int ret_code;
MM_typecode matcode;
FILE *f;
int i;//, *I(NULL); //, *J;
// double *val;
if ((f = fopen(file_name, "r")) == NULL)
return false;
if (mm_read_banner(f, &matcode) != 0)
{
printf("Could not process Matrix Market banner.\n");
return false;
}
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
printf("Sorry, this application does not support ");
printf("Market Market type: [%s]\n", mm_typecode_to_str(matcode));
return false;
}
/* find out size of sparse matrix .... */
int _num_rows, _num_cols, _num_nnzs;
if ((ret_code = mm_read_mtx_crd_size(f, &_num_rows, &_num_cols, &_num_nnzs)) !=0)
return false;
bool is_vector = (_num_rows == _num_nnzs) && (_num_cols == 1);
// convert to 64 bit
m_num_rows = _num_rows;
m_num_cols = _num_cols;
m_num_nnzs = _num_nnzs;
/* reseve memory for matrices */
// I = (int *) malloc(nz * sizeof(int));
// J = (int *) malloc(nz * sizeof(int));
// val = (double *) malloc(nz * sizeof(double));
m_rows.resize(m_num_nnzs);
m_cols.resize(m_num_nnzs);
m_coefs.resize(m_num_nnzs);
/* NOTE: when reading in doubles, ANSI C requires the use of the "l" */
/* specifier as in "%lg", "%lf", "%le", otherwise errors will occur */
/* (ANSI C X3.159-1989, Sec. 4.9.6.2, p. 136 lines 13-15) */
if (is_vector) {
for (i=0; i<m_num_nnzs; i++) {
fscanf(f, "%lg\n", &(m_coefs[i]));
m_rows[i] = i;
m_cols[i] = 0;
}
if (f !=stdin) fclose(f);
return true;
}
std::vector<std::map<int32_t,double> > elements;
elements.resize(m_num_rows);
for (i=0; i<m_num_nnzs; i++)
{
fscanf(f, "%d %d %lg\n", &(m_rows[i]), &(m_cols[i]), &(m_coefs[i]));
m_rows[i]--; /* adjust from 1-based to 0-based */
m_cols[i]--;
elements[m_rows[i]][m_cols[i]] += m_coefs[i];
}
// now sort by rows
std::vector<int> m_rows2;
std::vector<int32_t> m_cols2;
std::vector<double> m_coefs2;
m_rows2.resize(m_num_nnzs);
m_cols2.resize(m_num_nnzs);
m_coefs2.resize(m_num_nnzs);
int index=0;
for (int row=0; row<m_num_rows; row++) {
assert((int)elements[row].size() > 0);
// Make sure diagonal element is first
if (elements[row].find(row) != elements[row].end()) {
m_cols2[index] = row;
m_rows2[index] = row;
m_coefs2[index] = elements[row][row];
assert (m_coefs2[index] != 0.0);
index++;
}
else
assert(0);
for (auto iter = elements[row].begin(); iter != elements[row].end(); iter++) {
int col = iter->first;
if (col != row) {
double C = iter->second;
assert(col < m_num_cols);
m_rows2[index] = row;
m_cols2[index] = col;
m_coefs2[index] = C;
index++;
}
}
}
if (f !=stdin) fclose(f);
m_rows.swap(m_rows2);
m_cols.swap(m_cols2);
m_coefs.swap(m_coefs2);
/* convert coo to csr row ptrs */
coo_to_csr_rows(m_rows, m_num_rows, m_row_ptrs);
return true;
// /************************/
// /* now write out matrix */
// /************************/
// mm_write_banner(stdout, matcode);
// mm_write_mtx_crd_size(stdout, m_num_rows, m_num_cols, m_num_nnzs);
// for (i=0; i<nz; i++)
// fprintf(stdout, "%d %d %20.19g\n", I[i]+1, J[i]+1, val[i]);
// free(I);
}
int convert_matrixmarket(std::string base_filename, SharderPreprocessor<als_edge_type> * preprocessor = NULL) {
// Note, code based on: http://math.nist.gov/MatrixMarket/mmio/c/example_read.c
int ret_code;
MM_typecode matcode;
FILE *f;
size_t nz;
std::string suffix = "";
if (preprocessor != NULL) {
suffix = preprocessor->getSuffix();
}
/**
* Create sharder object
*/
int nshards;
if ((nshards = find_shards<als_edge_type>(base_filename+ suffix, get_option_string("nshards", "auto")))) {
logstream(LOG_INFO) << "File " << base_filename << " was already preprocessed, won't do it again. " << std::endl;
FILE * inf = fopen((base_filename + ".gm").c_str(), "r");
int rc = fscanf(inf,"%d\n%d\n%ld\n%lg\n%d\n",&M, &N, &L, &globalMean, &K);
if (rc != 5)
logstream(LOG_FATAL)<<"Failed to read global mean from file" << base_filename+ suffix << ".gm" << std::endl;
fclose(inf);
logstream(LOG_INFO) << "Opened matrix size: " <<M << " x " << N << " Global mean is: " << globalMean << " time bins: " << K << " Now creating shards." << std::endl;
return nshards;
}
sharder<als_edge_type> sharderobj(base_filename + suffix);
sharderobj.start_preprocessing();
if ((f = fopen(base_filename.c_str(), "r")) == NULL) {
logstream(LOG_FATAL) << "Could not open file: " << base_filename << ", error: " << strerror(errno) << std::endl;
}
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner. File: " << base_filename << std::endl;
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) || !mm_is_sparse(matcode))
logstream(LOG_FATAL) << "Sorry, this application does not support complex values and requires a sparse matrix." << std::endl;
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0) {
logstream(LOG_FATAL) << "Failed reading matrix size: error=" << ret_code << std::endl;
}
L=nz;
logstream(LOG_INFO) << "Starting to read matrix-market input. Matrix dimensions: "
<< M << " x " << N << ", non-zeros: " << nz << std::endl;
uint I, J;
double val;
if (!sharderobj.preprocessed_file_exists()) {
for (size_t i=0; i<nz; i++)
{
int rc = fscanf(f, "%d %d %lg\n", &I, &J, &val);
if (rc != 3)
logstream(LOG_FATAL)<<"Error when reading input file: " << i << std::endl;
I--; /* adjust from 1-based to 0-based */
J--;
if (I >= M)
logstream(LOG_FATAL)<<"Row index larger than the matrix row size " << I << " > " << M << " in line: " << i << std::endl;
if (J >= N)
logstream(LOG_FATAL)<<"Col index larger than the matrix col size " << J << " > " << N << " in line; " << i << std::endl;
globalMean += val;
sharderobj.preprocessing_add_edge(I, M + J, als_edge_type((float)val));
}
uint toadd = 0;
if (implicitratingtype == IMPLICIT_RATING_RANDOM)
toadd = add_implicit_edges(implicitratingtype, sharderobj);
globalMean += implicitratingvalue * toadd;
L += toadd;
sharderobj.end_preprocessing();
globalMean /= L;
logstream(LOG_INFO) << "Global mean is: " << globalMean << " Now creating shards." << std::endl;
if (preprocessor != NULL) {
preprocessor->reprocess(sharderobj.preprocessed_name(), base_filename);
}
FILE * outf = fopen((base_filename + ".gm").c_str(), "w");
fprintf(outf, "%d\n%d\n%ld\n%lg\n%d\n", M, N, L, globalMean, K);
fclose(outf);
} else {
logstream(LOG_INFO) << "Matrix already preprocessed, just run sharder." << std::endl;
}
fclose(f);
logstream(LOG_INFO) << "Now creating shards." << std::endl;
// Shard with a specified number of shards, or determine automatically if not defined
nshards = sharderobj.execute_sharding(get_option_string("nshards", "auto"));
logstream(LOG_INFO) << "Successfully finished sharding for " << base_filename + suffix << std::endl;
logstream(LOG_INFO) << "Created " << nshards << " shards." << std::endl;
return nshards;
}
void HostMatrixCOO<ValueType>::ReadFileMTX(const std::string filename) {
// Follow example_read.c (from Matrix Market web site)
int ret_code;
MM_typecode matcode;
FILE *f;
int M, N;
int fnz, nnz; // file nnz, real nnz
bool sym = false;
LOG_INFO("ReadFileMTX: filename="<< filename << "; reading...");
if ((f = fopen(filename.c_str(), "r")) == NULL) {
LOG_INFO("ReadFileMTX cannot open file " << filename);
FATAL_ERROR(__FILE__, __LINE__);
}
if (mm_read_banner(f, &matcode) != 0)
{
LOG_INFO("ReadFileMTX could not process Matrix Market banner.");
FATAL_ERROR(__FILE__, __LINE__);
}
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) && mm_is_matrix(matcode) &&
mm_is_sparse(matcode) )
{
LOG_INFO("ReadFileMTX does not support Market Market type:" << mm_typecode_to_str(matcode));
FATAL_ERROR(__FILE__, __LINE__);
}
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &fnz)) !=0) {
LOG_INFO("ReadFileMTX matrix size error");
FATAL_ERROR(__FILE__, __LINE__);
}
nnz = fnz ;
/* reseve memory for matrices */
if(mm_is_symmetric(matcode)) {
if (N != M) {
LOG_INFO("ReadFileMTX non-squared symmetric matrices are not supported");
LOG_INFO("What is symmetric and non-squared matrix? e-mail me");
FATAL_ERROR(__FILE__, __LINE__);
}
nnz = 2*(nnz - N) + N;
sym = true ;
}
this->AllocateCOO(nnz,M,N);
int ii=0;
int col, row;
double val;
int ret;
for (int i=0; i<fnz; ++i) {
ret = fscanf(f, "%d %d %lg\n", &row, &col, &val);
if (!ret) FATAL_ERROR(__FILE__, __LINE__);
row--; /* adjust from 1-based to 0-based */
col--;
assert (ret == 3);
// LOG_INFO(row << " " << col << " " << val);
this->mat_.row[ii] = row;
this->mat_.col[ii] = col;
this->mat_.val[ii] = val;
if (sym && (row!=col)) {
++ii;
this->mat_.row[ii] = col;
this->mat_.col[ii] = row;
this->mat_.val[ii] = val;
}
++ii;
}
LOG_INFO("ReadFileMTX: filename="<< filename << "; done");
fclose(f);
}
int convert_matrixmarket4(std::string base_filename, bool add_time_edges = false, bool square = false) {
// Note, code based on: http://math.nist.gov/MatrixMarket/mmio/c/example_read.c
int ret_code;
MM_typecode matcode;
FILE *f;
size_t nz;
/**
* Create sharder object
*/
int nshards;
if ((nshards = find_shards<als_edge_type>(base_filename, get_option_string("nshards", "auto")))) {
logstream(LOG_INFO) << "File " << base_filename << " was already preprocessed, won't do it again. " << std::endl;
FILE * inf = fopen((base_filename + ".gm").c_str(), "r");
int rc = fscanf(inf,"%d\n%d\n%ld\n%lg\n%d\n",&M, &N, &L, &globalMean, &K);
if (rc != 5)
logstream(LOG_FATAL)<<"Failed to read global mean from file" << base_filename << ".gm" << std::endl;
fclose(inf);
if (K <= 0)
logstream(LOG_FATAL)<<"Incorrect number of time bins K in .gm file " << base_filename << ".gm" << std::endl;
logstream(LOG_INFO) << "Read matrix of size " << M << " x " << N << " Global mean is: " << globalMean << " time bins: " << K << " Now creating shards." << std::endl;
return nshards;
}
sharder<als_edge_type> sharderobj(base_filename);
sharderobj.start_preprocessing();
if ((f = fopen(base_filename.c_str(), "r")) == NULL) {
logstream(LOG_FATAL) << "Could not open file: " << base_filename << ", error: " << strerror(errno) << std::endl;
}
if (mm_read_banner(f, &matcode) != 0)
logstream(LOG_FATAL) << "Could not process Matrix Market banner. File: " << base_filename << std::endl;
/* This is how one can screen matrix types if their application */
/* only supports a subset of the Matrix Market data types. */
if (mm_is_complex(matcode) || !mm_is_sparse(matcode))
logstream(LOG_FATAL) << "Sorry, this application does not support complex values and requires a sparse matrix." << std::endl;
/* find out size of sparse matrix .... */
if ((ret_code = mm_read_mtx_crd_size(f, &M, &N, &nz)) !=0) {
logstream(LOG_FATAL) << "Failed reading matrix size: error=" << ret_code << std::endl;
}
logstream(LOG_INFO) << "Starting to read matrix-market input. Matrix dimensions: "
<< M << " x " << N << ", non-zeros: " << nz << std::endl;
uint I, J;
double val, time;
if (!sharderobj.preprocessed_file_exists()) {
for (size_t i=0; i<nz; i++)
{
int rc = fscanf(f, "%d %d %lg %lg\n", &I, &J, &time, &val);
if (rc != 4)
logstream(LOG_FATAL)<<"Error when reading input file: " << i << std::endl;
if (time < 0)
logstream(LOG_FATAL)<<"Time (third columns) should be >= 0 " << std::endl;
I--; /* adjust from 1-based to 0-based */
J--;
if (I >= M)
logstream(LOG_FATAL)<<"Row index larger than the matrix row size " << I << " > " << M << " in line: " << i << std::endl;
if (J >= N)
logstream(LOG_FATAL)<<"Col index larger than the matrix col size " << J << " > " << N << " in line; " << i << std::endl;
K = std::max((int)time, (int)K);
//avoid self edges
if (square && I == J)
continue;
globalMean += val;
L++;
sharderobj.preprocessing_add_edge(I, (square? J : (M + J)), als_edge_type(val, time+M+N));
//in case of a tensor, add besides of the user-> movie edge also
//time -> user and time-> movie edges
if (add_time_edges){
sharderobj.preprocessing_add_edge((uint)time + M + N, I, als_edge_type(val, M+J));
sharderobj.preprocessing_add_edge((uint)time + M + N, M+J , als_edge_type(val, I));
}
}
uint toadd = 0;
if (implicitratingtype == IMPLICIT_RATING_RANDOM)
toadd = add_implicit_edges4(implicitratingtype, sharderobj);
globalMean += implicitratingvalue * toadd;
L += toadd;
sharderobj.end_preprocessing();
globalMean /= L;
logstream(LOG_INFO) << "Global mean is: " << globalMean << " time bins: " << K << " . Now creating shards." << std::endl;
FILE * outf = fopen((base_filename + ".gm").c_str(), "w");
fprintf(outf, "%d\n%d\n%ld\n%lg\n%d\n", M, N, L, globalMean, K);
fclose(outf);
} else {
logstream(LOG_INFO) << "Matrix already preprocessed, just run sharder." << std::endl;
}
fclose(f);
logstream(LOG_INFO) << "Now creating shards." << std::endl;
// Shard with a specified number of shards, or determine automatically if not defined
nshards = sharderobj.execute_sharding(get_option_string("nshards", "auto"));
return nshards;
}
