Blame view

fvn_sparse/UMFPACK/Source/umf_transpose.c 8.77 KB
422234dc3   daniau   git-svn-id: https...
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
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
  /* ========================================================================== */
  /* === UMF_transpose ======================================================== */
  /* ========================================================================== */
  
  /* -------------------------------------------------------------------------- */
  /* UMFPACK Copyright (c) Timothy A. Davis, CISE,                              */
  /* Univ. of Florida.  All Rights Reserved.  See ../Doc/License for License.   */
  /* web: http://www.cise.ufl.edu/research/sparse/umfpack                       */
  /* -------------------------------------------------------------------------- */
  
  /*  Not user-callable.  Computes a permuted transpose, R = (A (P,Q(1:nq)))' in
  	MATLAB notation, where R is in column-form.  A is n_row-by-n_col, the
  	row-form matrix R is n_row-by-nq, where nq <= n_col.  A may be singular.
  	The complex version can do transpose (') or array transpose (.').
  
  	Uses Gustavson's method (Two Fast Algorithms for Sparse Matrices:
  	Multiplication and Permuted Transposition, ACM Trans. on Math. Softw.,
  	vol 4, no 3, pp. 250-269).
  */
  
  #include "umf_internal.h"
  #include "umf_is_permutation.h"
  
  GLOBAL Int UMF_transpose
  (
      Int n_row,			/* A is n_row-by-n_col */
      Int n_col,
      const Int Ap [ ],		/* size n_col+1 */
      const Int Ai [ ],		/* size nz = Ap [n_col] */
      const double Ax [ ],	/* size nz if present */
  
      const Int P [ ],	/* P [k] = i means original row i is kth row in A(P,Q)*/
  			/* P is identity if not present */
  			/* size n_row, if present */
  
      const Int Q [ ],	/* Q [k] = j means original col j is kth col in A(P,Q)*/
  			/* Q is identity if not present */
  			/* size nq, if present */
      Int nq,		/* size of Q, ignored if Q is (Int *) NULL */
  
  			/* output matrix: Rp, Ri, Rx, and Rz: */
      Int Rp [ ],		/* size n_row+1 */
      Int Ri [ ],		/* size nz */
      double Rx [ ],	/* size nz, if present */
  
      Int W [ ],		/* size max (n_row,n_col) workspace */
  
      Int check		/* if true, then check inputs */
  #ifdef COMPLEX
      , const double Az [ ]	/* size nz */
      , double Rz [ ]		/* size nz */
      , Int do_conjugate		/* if true, then do conjugate transpose */
  				/* otherwise, do array transpose */
  #endif
  )
  {
  
      /* ---------------------------------------------------------------------- */
      /* local variables */
      /* ---------------------------------------------------------------------- */
  
      Int i, j, k, p, bp, newj, do_values ;
  #ifdef COMPLEX
      Int split ;
  #endif
  
      /* ---------------------------------------------------------------------- */
      /* check inputs */
      /* ---------------------------------------------------------------------- */
  
  #ifndef NDEBUG
      Int nz ;
      ASSERT (n_col >= 0) ;
      nz = (Ap != (Int *) NULL) ? Ap [n_col] : 0 ;
      DEBUG2 (("UMF_transpose: "ID"-by-"ID" nz "ID"
  ", n_row, n_col, nz)) ;
  #endif
  
      if (check)
      {
  	/* UMFPACK_symbolic skips this check */
  	/* UMFPACK_transpose always does this check */
  	if (!Ai || !Ap || !Ri || !Rp || !W)
  	{
  	    return (UMFPACK_ERROR_argument_missing) ;
  	}
  	if (n_row <= 0 || n_col <= 0)		/* n_row,n_col must be > 0 */
  	{
  	    return (UMFPACK_ERROR_n_nonpositive) ;
  	}
  	if (!UMF_is_permutation (P, W, n_row, n_row) ||
  	    !UMF_is_permutation (Q, W, nq, nq))
  	{
  	    return (UMFPACK_ERROR_invalid_permutation) ;
  	}
  	if (AMD_valid (n_row, n_col, Ap, Ai) != AMD_OK)
  	{
  	    return (UMFPACK_ERROR_invalid_matrix) ;
  	}
      }
  
  #ifndef NDEBUG
      DEBUG2 (("UMF_transpose, input matrix:
  ")) ;
      UMF_dump_col_matrix (Ax,
  #ifdef COMPLEX
  	Az,
  #endif
  	Ai, Ap, n_row, n_col, nz) ;
  #endif
  
      /* ---------------------------------------------------------------------- */
      /* count the entries in each row of A */
      /* ---------------------------------------------------------------------- */
  
      /* use W as workspace for RowCount */
  
      for (i = 0 ; i < n_row ; i++)
      {
  	W [i] = 0 ;
  	Rp [i] = 0 ;
      }
  
      if (Q != (Int *) NULL)
      {
  	for (newj = 0 ; newj < nq ; newj++)
  	{
  	    j = Q [newj] ;
  	    ASSERT (j >= 0 && j < n_col) ;
  	    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  	    {
  		i = Ai [p] ;
  		ASSERT (i >= 0 && i < n_row) ;
  		W [i]++ ;
  	    }
  	}
      }
      else
      {
  	for (j = 0 ; j < n_col ; j++)
  	{
  	    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  	    {
  		i = Ai [p] ;
  		ASSERT (i >= 0 && i < n_row) ;
  		W [i]++ ;
  	    }
  	}
      }
  
      /* ---------------------------------------------------------------------- */
      /* compute the row pointers for R = A (P,Q) */
      /* ---------------------------------------------------------------------- */
  
      if (P != (Int *) NULL)
      {
  	Rp [0] = 0 ;
  	for (k = 0 ; k < n_row ; k++)
  	{
  	    i = P [k] ;
  	    ASSERT (i >= 0 && i < n_row) ;
  	    Rp [k+1] = Rp [k] + W [i] ;
  	}
  	for (k = 0 ; k < n_row ; k++)
  	{
  	    i = P [k] ;
  	    ASSERT (i >= 0 && i < n_row) ;
  	    W [i] = Rp [k] ;
  	}
      }
      else
      {
  	Rp [0] = 0 ;
  	for (i = 0 ; i < n_row ; i++)
  	{
  	    Rp [i+1] = Rp [i] + W [i] ;
  	}
  	for (i = 0 ; i < n_row ; i++)
  	{
  	    W [i] = Rp [i] ;
  	}
      }
      ASSERT (Rp [n_row] <= Ap [n_col]) ;
  
      /* at this point, W holds the permuted row pointers */
  
      /* ---------------------------------------------------------------------- */
      /* construct the row form of B */
      /* ---------------------------------------------------------------------- */
  
      do_values = Ax && Rx ;
  
  #ifdef COMPLEX
      split = SPLIT (Az) && SPLIT (Rz) ;
  
      if (do_conjugate && do_values)
      {
  	if (Q != (Int *) NULL)
  	{
  	    if (split)
  	    {
  		/* R = A (P,Q)' */
  		for (newj = 0 ; newj < nq ; newj++)
  		{
  		    j = Q [newj] ;
  		    ASSERT (j >= 0 && j < n_col) ;
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			bp = W [Ai [p]]++ ;
  			Ri [bp] = newj ;
  			Rx [bp] = Ax [p] ;
  			Rz [bp] = -Az [p] ;
  		    }
  		}
  	    }
  	    else
  	    {
  		/* R = A (P,Q)' (merged complex values) */
  		for (newj = 0 ; newj < nq ; newj++)
  		{
  		    j = Q [newj] ;
  		    ASSERT (j >= 0 && j < n_col) ;
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			bp = W [Ai [p]]++ ;
  			Ri [bp] = newj ;
  			Rx [2*bp] = Ax [2*p] ;
  			Rx [2*bp+1] = -Ax [2*p+1] ;
  		    }
  		}
  	    }
  	}
  	else
  	{
  	    if (split)
  	    {
  		/* R = A (P,:)' */
  		for (j = 0 ; j < n_col ; j++)
  		{
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			bp = W [Ai [p]]++ ;
  			Ri [bp] = j ;
  			Rx [bp] = Ax [p] ;
  			Rz [bp] = -Az [p] ;
  		    }
  		}
  	    }
  	    else
  	    {
  		/* R = A (P,:)' (merged complex values) */
  		for (j = 0 ; j < n_col ; j++)
  		{
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			bp = W [Ai [p]]++ ;
  			Ri [bp] = j ;
  			Rx [2*bp] = Ax [2*p] ;
  			Rx [2*bp+1] = -Ax [2*p+1] ;
  		    }
  		}
  	    }
  	}
      }
      else
  #endif
      {
  	if (Q != (Int *) NULL)
  	{
  	    if (do_values)
  	    {
  #ifdef COMPLEX
  		if (split)
  #endif
  		{
  		    /* R = A (P,Q).' */
  		    for (newj = 0 ; newj < nq ; newj++)
  		    {
  			j = Q [newj] ;
  			ASSERT (j >= 0 && j < n_col) ;
  			for (p = Ap [j] ; p < Ap [j+1] ; p++)
  			{
  			    bp = W [Ai [p]]++ ;
  			    Ri [bp] = newj ;
  			    Rx [bp] = Ax [p] ;
  #ifdef COMPLEX
  			    Rz [bp] = Az [p] ;
  #endif
  			}
  		    }
  		}
  #ifdef COMPLEX
  		else
  		{
  		    /* R = A (P,Q).' (merged complex values) */
  		    for (newj = 0 ; newj < nq ; newj++)
  		    {
  			j = Q [newj] ;
  			ASSERT (j >= 0 && j < n_col) ;
  			for (p = Ap [j] ; p < Ap [j+1] ; p++)
  			{
  			    bp = W [Ai [p]]++ ;
  			    Ri [bp] = newj ;
  			    Rx [2*bp] = Ax [2*p] ;
  			    Rx [2*bp+1] = Ax [2*p+1] ;
  			}
  		    }
  		}
  #endif
  	    }
  	    else
  	    {
  		/* R = pattern of A (P,Q).' */
  		for (newj = 0 ; newj < nq ; newj++)
  		{
  		    j = Q [newj] ;
  		    ASSERT (j >= 0 && j < n_col) ;
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			Ri [W [Ai [p]]++] = newj ;
  		    }
  		}
  	    }
  	}
  	else
  	{
  	    if (do_values)
  	    {
  #ifdef COMPLEX
  		if (split)
  #endif
  		{
  		    /* R = A (P,:).' */
  		    for (j = 0 ; j < n_col ; j++)
  		    {
  			for (p = Ap [j] ; p < Ap [j+1] ; p++)
  			{
  			    bp = W [Ai [p]]++ ;
  			    Ri [bp] = j ;
  			    Rx [bp] = Ax [p] ;
  #ifdef COMPLEX
  			    Rz [bp] = Az [p] ;
  #endif
  			}
  		    }
  		}
  #ifdef COMPLEX
  		else
  		{
  		    /* R = A (P,:).' (merged complex values) */
  		    for (j = 0 ; j < n_col ; j++)
  		    {
  			for (p = Ap [j] ; p < Ap [j+1] ; p++)
  			{
  			    bp = W [Ai [p]]++ ;
  			    Ri [bp] = j ;
  			    Rx [2*bp] = Ax [2*p] ;
  			    Rx [2*bp+1] = Ax [2*p+1] ;
  			}
  		    }
  		}
  #endif
  	    }
  	    else
  	    {
  		/* R = pattern of A (P,:).' */
  		for (j = 0 ; j < n_col ; j++)
  		{
  		    for (p = Ap [j] ; p < Ap [j+1] ; p++)
  		    {
  			Ri [W [Ai [p]]++] = j ;
  		    }
  		}
  	    }
  	}
      }
  
  #ifndef NDEBUG
      for (k = 0 ; k < n_row ; k++)
      {
  	if (P != (Int *) NULL)
  	{
  	    i = P [k] ;
  	}
  	else
  	{
  	    i = k ;
  	}
  	DEBUG3 ((ID":  W[i] "ID" Rp[k+1] "ID"
  ", i, W [i], Rp [k+1])) ;
  	ASSERT (W [i] == Rp [k+1]) ;
      }
      DEBUG2 (("UMF_transpose, output matrix:
  ")) ;
      UMF_dump_col_matrix (Rx,
  #ifdef COMPLEX
  	Rz,
  #endif
  	Ri, Rp, n_col, n_row, Rp [n_row]) ;
      ASSERT (AMD_valid (n_col, n_row, Rp, Ri) == AMD_OK) ;
  #endif
  
      return (UMFPACK_OK) ;
  }