1/* $Id$Revision: */
2/* vim:set shiftwidth=4 ts=8: */
3
4/*************************************************************************
5 * Copyright (c) 2011 AT&T Intellectual Property
6 * All rights reserved. This program and the accompanying materials
7 * are made available under the terms of the Eclipse Public License v1.0
8 * which accompanies this distribution, and is available at
9 * http://www.eclipse.org/legal/epl-v10.html
10 *
11 * Contributors: See CVS logs. Details at http://www.graphviz.org/
12 *************************************************************************/
13#ifndef SPARSEMATRIX_H
14#define SPARSEMATRIX_H
15
16#include <general.h>
17#include <stdio.h>
18
19#define SYMMETRY_EPSILON 0.0000001
20enum {FORMAT_CSC, FORMAT_CSR, FORMAT_COORD};
21enum {UNMASKED = -10, MASKED = 1};
22enum {MATRIX_PATTERN_SYMMETRIC = 1<<0, MATRIX_SYMMETRIC = 1<<1, MATRIX_SKEW = 1<<2, MATRIX_HERMITIAN = 1<<3, MATRIX_UNDIRECTED = 1<<4};
23enum {BIPARTITE_RECT = 0, BIPARTITE_PATTERN_UNSYM, BIPARTITE_UNSYM, BIPARTITE_ALWAYS};
24
25
26struct SparseMatrix_struct {
27 int m; /* row dimension */
28 int n; /* column dimension */
29 int nz;/* The actual length used is nz, for CSR/CSC matrix this is the same as ia[n] */
30 int nzmax; /* the current length of ja and a (if exists) allocated.*/
31 int type; /* whether it is real/complex matrix, or pattern only */
32 int *ia; /* row pointer for CSR format, or row indices for coordinate format. 0-based */
33 int *ja; /* column indices. 0-based */
34 void *a; /* entry values. If NULL, pattern matrix */
35 int format;/* whether it is CSR, CSC, COORD. By default it is in CSR format */
36 int property; /* pattern_symmetric/symmetric/skew/hermitian*/
37 int size;/* size of each entry. This allows for general matrix where each entry is, say, a matrix itself */
38};
39
40typedef struct SparseMatrix_struct* SparseMatrix;
41
42enum {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};
43
44/* SparseMatrix_general is more general and allow elements to be
45 any data structure, not just real/int/complex etc */
46SparseMatrix SparseMatrix_new(int m, int n, int nz, int type, int format);
47SparseMatrix SparseMatrix_general_new(int m, int n, int nz, int type, size_t sz, int format);
48
49/* this version sum repeated entries */
50SparseMatrix SparseMatrix_from_coordinate_format(SparseMatrix A);
51/* what_to_sum is SUM_REPEATED_NONE, SUM_REPEATED_ALL, SUM_REPEATED_REAL_PART, SUM_REPEATED_IMAGINARY_PART, SUM_IMGINARY_KEEP_LAST_REAL*/
52SparseMatrix SparseMatrix_from_coordinate_format_not_compacted(SparseMatrix A, int what_to_sum);
53
54SparseMatrix SparseMatrix_from_coordinate_arrays(int nz, int m, int n, int *irn, int *jcn, void *val, int type, size_t sz);
55SparseMatrix 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);
56
57
58void SparseMatrix_print(char *, SparseMatrix A);/*print to stdout in Mathematica format*/
59
60void SparseMatrix_export(FILE *f, SparseMatrix A);/* export into MM format except the header */
61
62SparseMatrix SparseMatrix_import_binary(char *name);
63SparseMatrix SparseMatrix_import_binary_fp(FILE *f);/* import into a preopenned file */
64
65void SparseMatrix_export_binary(char *name, SparseMatrix A, int *flag);
66void SparseMatrix_export_binary_fp(FILE *f, SparseMatrix A);/* export binary into a file preopened */
67
68void SparseMatrix_delete(SparseMatrix A);
69
70SparseMatrix SparseMatrix_add(SparseMatrix A, SparseMatrix B);
71SparseMatrix SparseMatrix_multiply(SparseMatrix A, SparseMatrix B);
72SparseMatrix SparseMatrix_multiply3(SparseMatrix A, SparseMatrix B, SparseMatrix C);
73
74/* For complex matrix:
75 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
76 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
77 For other matrix, what_to_sum = SUM_REPEATED_REAL_PART is the same as what_to_sum = SUM_REPEATED_IMAGINARY_PART
78 or what_to_sum = SUM_REPEATED_ALL
79 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,
80 . and we use this to indicate that a user watched R1+R2 seconds, last watch is I2.
81*/
82enum {SUM_REPEATED_NONE = 0, SUM_REPEATED_ALL, SUM_REPEATED_REAL_PART, SUM_REPEATED_IMAGINARY_PART, SUM_IMGINARY_KEEP_LAST_REAL};
83SparseMatrix SparseMatrix_sum_repeat_entries(SparseMatrix A, int what_to_sum);
84SparseMatrix SparseMatrix_coordinate_form_add_entries(SparseMatrix A, int nentries, int *irn, int *jcn, void *val);
85int SparseMatrix_is_symmetric(SparseMatrix A, int test_pattern_symmetry_only);
86SparseMatrix SparseMatrix_transpose(SparseMatrix A);
87SparseMatrix SparseMatrix_symmetrize(SparseMatrix A, int pattern_symmetric_only);
88SparseMatrix SparseMatrix_symmetrize_nodiag(SparseMatrix A, int pattern_symmetric_only);
89void SparseMatrix_multiply_vector(SparseMatrix A, real *v, real **res, int transposed);/* if v = NULL, v is assumed to be {1,1,...,1}*/
90SparseMatrix SparseMatrix_remove_diagonal(SparseMatrix A);
91SparseMatrix SparseMatrix_remove_upper(SparseMatrix A);/* remove diag and upper diag */
92SparseMatrix SparseMatrix_divide_row_by_degree(SparseMatrix A);
93SparseMatrix SparseMatrix_get_real_adjacency_matrix_symmetrized(SparseMatrix A); /* symmetric, all entries to 1, diaginal removed */
94SparseMatrix SparseMatrix_normalize_to_rowsum1(SparseMatrix A);/* for real only! */
95void SparseMatrix_multiply_dense(SparseMatrix A, int ATranspose, real *v, int vTransposed, real **res, int res_transpose, int dim);
96SparseMatrix SparseMatrix_apply_fun(SparseMatrix A, double (*fun)(double x));/* for real only! */
97SparseMatrix SparseMatrix_apply_fun_general(SparseMatrix A, void (*fun)(int i, int j, int n, double *x));/* for real and complex (n=2) */
98SparseMatrix SparseMatrix_copy(SparseMatrix A);
99int SparseMatrix_has_diagonal(SparseMatrix A);
100SparseMatrix SparseMatrix_normalize_by_row(SparseMatrix A);/* divide by max of each row */
101SparseMatrix SparseMatrix_crop(SparseMatrix A, real epsilon);/*remove any entry <= epsilon*/
102SparseMatrix SparseMatrix_scaled_by_vector(SparseMatrix A, real *v, int apply_to_row);
103SparseMatrix SparseMatrix_multiply_by_scaler(SparseMatrix A, real s);
104SparseMatrix SparseMatrix_make_undirected(SparseMatrix A);/* make it strictly low diag only, and set flag to undirected */
105int SparseMatrix_connectedQ(SparseMatrix A);
106real SparseMatrix_pseudo_diameter_only(SparseMatrix A);
107real SparseMatrix_pseudo_diameter_weighted(SparseMatrix A0, int root, int aggressive, int *end1, int *end2, int *connectedQ); /* assume real distances, unsymmetric matrix ill be symmetrized */
108real 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 */
109void SparseMatrix_level_sets(SparseMatrix A, int root, int *nlevel, int **levelset_ptr, int **levelset, int **mask, int reintialize_mask);
110void SparseMatrix_level_sets_khops(int khops, SparseMatrix A, int root, int *nlevel, int **levelset_ptr, int **levelset, int **mask, int reintialize_mask);
111void SparseMatrix_weakly_connected_components(SparseMatrix A0, int *ncomp, int **comps, int **comps_ptr);
112void SparseMatrix_decompose_to_supervariables(SparseMatrix A, int *ncluster, int **cluster, int **clusterp);
113SparseMatrix SparseMatrix_get_submatrix(SparseMatrix A, int nrow, int ncol, int *rindices, int *cindices);
114SparseMatrix SparseMatrix_exclude_submatrix(SparseMatrix A, int nrow, int ncol, int *rindices, int *cindices);
115
116SparseMatrix SparseMatrix_get_augmented(SparseMatrix A);
117
118/* bipartite_options:
119 BIPARTITE_RECT -- turn rectangular matrix into square),
120 BIPARTITE_PATTERN_UNSYM -- pattern unsummetric as bipartite
121 BIPARTITE_UNSYM -- unsymmetric as square
122 BIPARTITE_ALWAYS -- always as square
123*/
124SparseMatrix SparseMatrix_to_square_matrix(SparseMatrix A, int bipartite_options);
125
126SparseMatrix SparseMatrix_largest_component(SparseMatrix A);
127
128/* columns with <= threhold entries are deleted */
129SparseMatrix SparseMatrix_delete_empty_columns(SparseMatrix A, int **new2old, int *nnew, int inplace);
130SparseMatrix SparseMatrix_delete_sparse_columns(SparseMatrix A, int threshold, int **new2old, int *nnew, int inplace);
131
132SparseMatrix SparseMatrix_sort(SparseMatrix A);
133
134SparseMatrix SparseMatrix_set_entries_to_real_one(SparseMatrix A);
135
136SparseMatrix SparseMatrix_complement(SparseMatrix A, int undirected);
137
138int SparseMatrix_k_centers(SparseMatrix D, int weighted, int K, int root,
139 int **centers, int centering, real **dist);
140
141int SparseMatrix_k_centers_user(SparseMatrix D, int weighted, int K,
142 int *centers_user, int centering, real **dist);
143
144SparseMatrix SparseMatrix_distance_matrix_k_centers(int K, SparseMatrix D, int weighted);
145
146int SparseMatrix_distance_matrix(SparseMatrix A, int weighted, real **dist_matrix);
147SparseMatrix SparseMatrix_distance_matrix_khops(int khops, SparseMatrix A, int weighted);
148SparseMatrix SparseMatrix_distance_matrix_k_centers(int K, SparseMatrix D, int weighted);
149
150void SparseMatrix_kcoreness(SparseMatrix A, int **coreness);/* assign coreness to each node */
151void SparseMatrix_kcore_decomposition(SparseMatrix A, int *coreness_max0, int **coreness_ptr0, int **coreness_list0);/* return the decomposition */
152
153void SparseMatrix_khairness(SparseMatrix A, int **hairness);/* assign hairness to each node */
154void SparseMatrix_khair_decomposition(SparseMatrix A, int *hairness_max0, int **hairness_ptr0, int **hairness_list0);/* return the decomposition */
155
156SparseMatrix SparseMatrix_from_dense(int m, int n, real *x);
157
158void SparseMatrix_page_rank(SparseMatrix A, real teleport_probablity, int weighted, real epsilon, real **page_rank);
159
160
161#define SparseMatrix_set_undirected(A) set_flag((A)->property, MATRIX_UNDIRECTED)
162#define SparseMatrix_set_symmetric(A) set_flag((A)->property, MATRIX_SYMMETRIC)
163#define SparseMatrix_set_pattern_symmetric(A) set_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
164#define SparseMatrix_set_skew(A) set_flag((A)->property, MATRIX_SKEW)
165#define SparseMatrix_set_hemitian(A) set_flag((A)->property, MATRIX_HERMITIAN)
166
167
168#define SparseMatrix_clear_undirected(A) clear_flag((A)->property, MATRIX_UNDIRECTED)
169#define SparseMatrix_clear_symmetric(A) clear_flag((A)->property, MATRIX_SYMMETRIC)
170#define SparseMatrix_clear_pattern_symmetric(A) clear_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
171#define SparseMatrix_clear_skew(A) clear_flag((A)->property, MATRIX_SKEW)
172#define SparseMatrix_clear_hemitian(A) clear_flag((A)->property, MATRIX_HERMITIAN)
173
174
175#define SparseMatrix_known_undirected(A) test_flag((A)->property, MATRIX_UNDIRECTED)
176#define SparseMatrix_known_symmetric(A) test_flag((A)->property, MATRIX_SYMMETRIC)
177#define SparseMatrix_known_strucural_symmetric(A) test_flag((A)->property, MATRIX_PATTERN_SYMMETRIC)
178#define SparseMatrix_known_skew(A) test_flag((A)->property, MATRIX_SKEW)
179#define SparseMatrix_known_hemitian(A) test_flag((A)->property, MATRIX_HERMITIAN)
180
181
182
183
184#endif
185