SparseMatrix.h

Go to the documentation of this file.

/* $Id$Revision: */
/* vim:set shiftwidth=4 ts=8: */

/*************************************************************************
 * Copyright (c) 2011 AT&T Intellectual Property 
 * All rights reserved. This program and the accompanying materials
 * are made available under the terms of the Eclipse Public License v1.0
 * which accompanies this distribution, and is available at
 * http://www.eclipse.org/legal/epl-v10.html
 *
 * Contributors: See CVS logs. Details at http://www.graphviz.org/
 *************************************************************************/
#ifndef SPARSEMATRIX_H
#define  SPARSEMATRIX_H

#include <general.h>
#include <stdio.h>

#define SYMMETRY_EPSILON 0.0000001
enum {FORMAT_CSC, FORMAT_CSR, FORMAT_COORD};
enum {UNMASKED = -10, MASKED = 1};
enum {MATRIX_PATTERN_SYMMETRIC = 1<<0, MATRIX_SYMMETRIC = 1<<1, MATRIX_SKEW = 1<<2, MATRIX_HERMITIAN = 1<<3, MATRIX_UNDIRECTED = 1<<4};
enum {BIPARTITE_RECT = 0, BIPARTITE_PATTERN_UNSYM, BIPARTITE_UNSYM, BIPARTITE_ALWAYS};


struct SparseMatrix_struct {
  int m; /* row dimension */
  int n; /* column dimension */
  int nz;/* The actual length used is nz, for CSR/CSC matrix this is the same as ia[n] */
  int nzmax; /* the current length of ja and a (if exists) allocated.*/
  int type; /* whether it is real/complex matrix, or pattern only */
  int *ia; /* row pointer for CSR format, or row indices for coordinate format. 0-based */
  int *ja; /* column indices. 0-based */
  void *a; /* entry values. If NULL, pattern matrix */
  int format;/* whether it is CSR, CSC, COORD. By default it is in CSR format */
  int property; /* pattern_symmetric/symmetric/skew/hermitian*/
  int size;/* size of each entry. This allows for general matrix where each entry is, say, a matrix itself */
};

typedef struct SparseMatrix_struct* SparseMatrix;

enum {MATRIX_TYPE_REAL = 1<<0, MATRIX_TYPE_COMPLEX = 1<<1, MATRIX_TYPE_INTEGER = 1<<2, MATRIX_TYPE_PATTERN = 1<<3, MATRIX_TYPE_UNKNOWN = 1<<4};

/* SparseMatrix_general is more general and allow elements to be 
   any data structure, not just real/int/complex etc */
SparseMatrix SparseMatrix_new(int m, int n, int nz, int type, int format);
SparseMatrix SparseMatrix_general_new(int m, int n, int nz, int type, size_t sz, int format);

/* this version sum repeated entries */
SparseMatrix SparseMatrix_from_coordinate_format(SparseMatrix A);
/* what_to_sum is SUM_REPEATED_NONE, SUM_REPEATED_ALL, SUM_REPEATED_REAL_PART, SUM_REPEATED_IMAGINARY_PART, SUM_IMGINARY_KEEP_LAST_REAL*/
SparseMatrix SparseMatrix_from_coordinate_format_not_compacted(SparseMatrix A, int what_to_sum);

SparseMatrix SparseMatrix_from_coordinate_arrays(int nz, int m, int n, int *irn, int *jcn, void *val, int type, size_t sz);
SparseMatrix SparseMatrix_from_coordinate_arrays_not_compacted(int nz, int m, int n, int *irn, int *jcn, void *val, int type, size_t sz, int what_to_sum);


void SparseMatrix_print(char *, SparseMatrix A);/*print to stdout in Mathematica format*/

void SparseMatrix_export(FILE *f, SparseMatrix A);/* export into MM format except the header */

SparseMatrix SparseMatrix_import_binary(char *name);
SparseMatrix SparseMatrix_import_binary_fp(FILE *f);/* import into a preopenned file */

void SparseMatrix_export_binary(char *name, SparseMatrix A, int *flag);
void SparseMatrix_export_binary_fp(FILE *f, SparseMatrix A);/* export binary into a file preopened */

void SparseMatrix_delete(SparseMatrix A);

SparseMatrix SparseMatrix_add(SparseMatrix A, SparseMatrix B);
SparseMatrix SparseMatrix_multiply(SparseMatrix A, SparseMatrix B);
SparseMatrix SparseMatrix_multiply3(SparseMatrix A, SparseMatrix B, SparseMatrix C);

/* For complex matrix:
   if what_to_sum = SUM_REPEATED_REAL_PART, we find entries {i,j,x + i y} and sum the x's if {i,j,Round(y)} are the same
   if what_to_sum = SUM_REPEATED_IMAGINARY_PART, we find entries {i,j,x + i y} and sum the y's if {i,j,Round(x)} are the same
   For other matrix, what_to_sum = SUM_REPEATED_REAL_PART is the same as what_to_sum = SUM_REPEATED_IMAGINARY_PART
   or what_to_sum = SUM_REPEATED_ALL
   if what_to_sum = SUM_IMGINARY_KEEP_LAST_REAL, we merge {i,j,R1,I1} and {i,j,R2,I2} into {i,j,R1+R2,I2}. Useful if I1 and I2 are time stamps, 
   .   and we use this to indicate that a user watched R1+R2 seconds, last watch is I2.
*/
enum {SUM_REPEATED_NONE = 0, SUM_REPEATED_ALL, SUM_REPEATED_REAL_PART, SUM_REPEATED_IMAGINARY_PART, SUM_IMGINARY_KEEP_LAST_REAL};
SparseMatrix SparseMatrix_sum_repeat_entries(SparseMatrix A, int what_to_sum);
SparseMatrix SparseMatrix_coordinate_form_add_entries(SparseMatrix A, int nentries, int *irn, int *jcn, void *val);
int SparseMatrix_is_symmetric(SparseMatrix A, int test_pattern_symmetry_only);
SparseMatrix SparseMatrix_transpose(SparseMatrix A);
SparseMatrix SparseMatrix_symmetrize(SparseMatrix A, int pattern_symmetric_only);
SparseMatrix SparseMatrix_symmetrize_nodiag(SparseMatrix A, int pattern_symmetric_only);
void SparseMatrix_multiply_vector(SparseMatrix A, real *v, real **res, int transposed);/* if v = NULL, v is assumed to be {1,1,...,1}*/
SparseMatrix SparseMatrix_remove_diagonal(SparseMatrix A);
SparseMatrix SparseMatrix_remove_upper(SparseMatrix A);/* remove diag and upper diag */
SparseMatrix SparseMatrix_divide_row_by_degree(SparseMatrix A);
SparseMatrix SparseMatrix_get_real_adjacency_matrix_symmetrized(SparseMatrix A);  /* symmetric, all entries to 1, diaginal removed */
SparseMatrix SparseMatrix_normalize_to_rowsum1(SparseMatrix A);/* for real only! */
void SparseMatrix_multiply_dense(SparseMatrix A, int ATranspose, real *v, int vTransposed, real **res, int res_transpose, int dim);
SparseMatrix SparseMatrix_apply_fun(SparseMatrix A, double (*fun)(double x));/* for real only! */
SparseMatrix SparseMatrix_apply_fun_general(SparseMatrix A, void (*fun)(int i, int j, int n, double *x));/* for real and complex (n=2) */
SparseMatrix SparseMatrix_copy(SparseMatrix A);
int SparseMatrix_has_diagonal(SparseMatrix A);
SparseMatrix SparseMatrix_normalize_by_row(SparseMatrix A);/* divide by max of each row */
SparseMatrix SparseMatrix_crop(SparseMatrix A, real epsilon);/*remove any entry <= epsilon*/
SparseMatrix SparseMatrix_scaled_by_vector(SparseMatrix A, real *v, int apply_to_row);
SparseMatrix SparseMatrix_multiply_by_scaler(SparseMatrix A, real s);
SparseMatrix SparseMatrix_make_undirected(SparseMatrix A);/* make it strictly low diag only, and set flag to undirected */
int SparseMatrix_connectedQ(SparseMatrix A);
real SparseMatrix_pseudo_diameter_only(SparseMatrix A);
real SparseMatrix_pseudo_diameter_weighted(SparseMatrix A0, int root, int aggressive, int *end1, int *end2, int *connectedQ); /* assume real distances, unsymmetric matrix ill be symmetrized */
real SparseMatrix_pseudo_diameter_unweighted(SparseMatrix A0, int root, int aggressive, int *end1, int *end2, int *connectedQ); /* assume unit edge length, unsymmetric matrix ill be symmetrized */
void SparseMatrix_level_sets(SparseMatrix A, int root, int *nlevel, int **levelset_ptr, int **levelset, int **mask, int reintialize_mask);
void SparseMatrix_level_sets_khops(int khops, SparseMatrix A, int root, int *nlevel, int **levelset_ptr, int **levelset, int **mask, int reintialize_mask);
void SparseMatrix_weakly_connected_components(SparseMatrix A0, int *ncomp, int **comps, int **comps_ptr);
void SparseMatrix_decompose_to_supervariables(SparseMatrix A, int *ncluster, int **cluster, int **clusterp);
SparseMatrix SparseMatrix_get_submatrix(SparseMatrix A, int nrow, int ncol, int *rindices, int *cindices);
SparseMatrix SparseMatrix_exclude_submatrix(SparseMatrix A, int nrow, int ncol, int *rindices, int *cindices);

SparseMatrix SparseMatrix_get_augmented(SparseMatrix A);

/* bipartite_options:
   BIPARTITE_RECT -- turn rectangular matrix into square), 
   BIPARTITE_PATTERN_UNSYM -- pattern unsummetric as bipartite
   BIPARTITE_UNSYM -- unsymmetric as square
   BIPARTITE_ALWAYS -- always as square
*/
SparseMatrix SparseMatrix_to_square_matrix(SparseMatrix A, int bipartite_options);

SparseMatrix SparseMatrix_largest_component(SparseMatrix A);

/* columns with <= threhold entries are deleted */
SparseMatrix SparseMatrix_delete_empty_columns(SparseMatrix A, int **new2old, int *nnew, int inplace);
SparseMatrix SparseMatrix_delete_sparse_columns(SparseMatrix A, int threshold, int **new2old, int *nnew, int inplace);

SparseMatrix SparseMatrix_sort(SparseMatrix A);

SparseMatrix SparseMatrix_set_entries_to_real_one(SparseMatrix A);

SparseMatrix SparseMatrix_complement(SparseMatrix A, int undirected);

int SparseMatrix_k_centers(SparseMatrix D, int weighted, int K, int root, 
                          int **centers, int centering, real **dist);

int SparseMatrix_k_centers_user(SparseMatrix D, int weighted, int K, 
                                int *centers_user, int centering, real **dist);

SparseMatrix SparseMatrix_distance_matrix_k_centers(int K, SparseMatrix D, int weighted);

int SparseMatrix_distance_matrix(SparseMatrix A, int weighted,  real **dist_matrix);
SparseMatrix SparseMatrix_distance_matrix_khops(int khops, SparseMatrix A, int weighted);
SparseMatrix SparseMatrix_distance_matrix_k_centers(int K, SparseMatrix D, int weighted);

void SparseMatrix_kcoreness(SparseMatrix A, int **coreness);/* assign coreness to each node */
void SparseMatrix_kcore_decomposition(SparseMatrix A, int *coreness_max0, int **coreness_ptr0, int **coreness_list0);/* return the decomposition */

void SparseMatrix_khairness(SparseMatrix A, int **hairness);/* assign hairness to each node */
void SparseMatrix_khair_decomposition(SparseMatrix A, int *hairness_max0, int **hairness_ptr0, int **hairness_list0);/* return the decomposition */

SparseMatrix SparseMatrix_from_dense(int m, int n, real *x);

void SparseMatrix_page_rank(SparseMatrix A, real teleport_probablity, int weighted, real epsilon, real **page_rank);


#define SparseMatrix_set_undirected(A) set_flag((A)->property, MATRIX_UNDIRECTED)
#define SparseMatrix_set_symmetric(A) set_flag((A)->property, MATRIX_SYMMETRIC)
#define SparseMatrix_set_pattern_symmetric(A) set_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
#define SparseMatrix_set_skew(A) set_flag((A)->property, MATRIX_SKEW)
#define SparseMatrix_set_hemitian(A) set_flag((A)->property, MATRIX_HERMITIAN)


#define SparseMatrix_clear_undirected(A) clear_flag((A)->property, MATRIX_UNDIRECTED)
#define SparseMatrix_clear_symmetric(A) clear_flag((A)->property, MATRIX_SYMMETRIC)
#define SparseMatrix_clear_pattern_symmetric(A) clear_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
#define SparseMatrix_clear_skew(A) clear_flag((A)->property, MATRIX_SKEW)
#define SparseMatrix_clear_hemitian(A) clear_flag((A)->property, MATRIX_HERMITIAN)


#define SparseMatrix_known_undirected(A) test_flag((A)->property, MATRIX_UNDIRECTED)
#define SparseMatrix_known_symmetric(A) test_flag((A)->property, MATRIX_SYMMETRIC)
#define SparseMatrix_known_strucural_symmetric(A) test_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
#define SparseMatrix_known_skew(A) test_flag((A)->property, MATRIX_SKEW)
#define SparseMatrix_known_hemitian(A) test_flag((A)->property, MATRIX_HERMITIAN)




#endif