MODFLOW 6  version 6.6.0.dev0
USGS Modular Hydrologic Model
sparsekit.f90
Go to the documentation of this file.
1 subroutine amux ( n, x, y, a, ja, ia )
2 
3  !*****************************************************************************80
4  !
5  !! AMUX multiplies a CSR matrix A times a vector.
6  !
7  ! Discussion:
8  !
9  ! This routine multiplies a matrix by a vector using the dot product form.
10  ! Matrix A is stored in compressed sparse row storage.
11  !
12  ! Modified:
13  !
14  ! 07 January 2004
15  !
16  ! Author:
17  !
18  ! Youcef Saad
19  !
20  ! Parameters:
21  !
22  ! Input, integer ( kind = 4 ) N, the row dimension of the matrix.
23  !
24  ! Input, real X(*), and array of length equal to the column dimension
25  ! of A.
26  !
27  ! Input, real A(*), integer ( kind = 4 ) JA(*), IA(NROW+1), the matrix in CSR
28  ! Compressed Sparse Row format.
29  !
30  ! Output, real Y(N), the product A * X.
31  !
32  implicit none
33 
34  integer ( kind = 4 ) n
35 
36  real ( kind = 8 ) a(*)
37  integer ( kind = 4 ) i
38  integer ( kind = 4 ) ia(*)
39  integer ( kind = 4 ) ja(*)
40  integer ( kind = 4 ) k
41  real ( kind = 8 ) t
42  real ( kind = 8 ) x(*)
43  real ( kind = 8 ) y(n)
44 
45  do i = 1, n
46  !
47  ! Compute the inner product of row I with vector X.
48  !
49  t = 0.0d+00
50  do k = ia(i), ia(i+1)-1
51  t = t + a(k) * x(ja(k))
52  end do
53 
54  y(i) = t
55 
56  end do
57 
58  return
59  end
60 
61 subroutine dvperm ( n, x, perm )
62 
63  !*****************************************************************************80
64  !
65  !! DVPERM performs an in-place permutation of a real vector.
66  !
67  ! Discussion:
68  !
69  ! This routine permutes a real vector X using a permutation PERM.
70  !
71  ! On return, the vector X satisfies,
72  !
73  ! x(perm(j)) :== x(j), j = 1,2,.., n
74  !
75  ! Modified:
76  !
77  ! 07 January 2004
78  !
79  ! Author:
80  !
81  ! Youcef Saad
82  !
83  ! Parameters:
84  !
85  ! Input, integer ( kind = 4 ) N, the length of X.
86  !
87  ! Input/output, real X(N), the vector to be permuted.
88  !
89  ! Input, integer ( kind = 4 ) PERM(N), the permutation.
90  !
91  implicit none
92 
93  integer ( kind = 4 ) n
94 
95  integer ( kind = 4 ) ii
96  integer ( kind = 4 ) init
97  integer ( kind = 4 ) k
98  integer ( kind = 4 ) next
99  integer ( kind = 4 ) perm(n)
100  real ( kind = 8 ) tmp
101  real ( kind = 8 ) tmp1
102  real ( kind = 8 ) x(n)
103 
104  init = 1
105  tmp = x(init)
106  ii = perm(init)
107  perm(init)= -perm(init)
108  k = 0
109  !
110  ! The main loop.
111  !
112  6 continue
113 
114  k = k + 1
115  !
116  ! Save the chased element.
117  !
118  tmp1 = x(ii)
119  x(ii) = tmp
120  next = perm(ii)
121 
122  if ( next < 0 ) then
123  go to 65
124  end if
125  !
126  ! Test for end.
127  !
128  if ( n < k ) then
129  perm(1:n) = -perm(1:n)
130  return
131  end if
132 
133  tmp = tmp1
134  perm(ii) = -perm(ii)
135  ii = next
136  !
137  ! End of the loop.
138  !
139  go to 6
140  !
141  ! Reinitialize cycle.
142  !
143  65 continue
144 
145  init = init + 1
146 
147  if ( n < init ) then
148  perm(1:n) = -perm(1:n)
149  return
150  end if
151 
152  if ( perm(init) < 0 ) then
153  go to 65
154  end if
155 
156  tmp = x(init)
157  ii = perm(init)
158  perm(init) = -perm(init)
159  go to 6
160 
161 end subroutine dvperm
162 
163 subroutine cperm ( nrow, a, ja, ia, ao, jao, iao, perm, job )
164 
165  !*****************************************************************************80
166  !
167  !! CPERM permutes the columns of a matrix.
168  !
169  ! Discussion:
170  !
171  ! This routine permutes the columns of a matrix a, ja, ia.
172  ! The result is written in the output matrix ao, jao, iao.
173  ! cperm computes B = A P, where P is a permutation matrix
174  ! that maps column j into column perm(j), i.e., on return
175  ! a(i,j) becomes a(i,perm(j)) in new matrix
176  !
177  ! if job=1 then ao, iao are not used.
178  ! This routine is in place: ja, jao can be the same.
179  ! If the matrix is initially sorted (by increasing column number)
180  ! then ao,jao,iao may not be on return.
181  !
182  ! Modified:
183  !
184  ! 07 January 2004
185  !
186  ! Author:
187  !
188  ! Youcef Saad
189  !
190  ! Parameters:
191  !
192  ! Input, integer ( kind = 4 ) NROW, the row dimension of the matrix.
193  !
194  ! Input, real A(*), integer ( kind = 4 ) JA(*), IA(NROW+1), the matrix in CSR
195  ! Compressed Sparse Row format.
196  !
197  ! perm = integer ( kind = 4 ) array of length ncol (number of columns of A
198  ! containing the permutation array the columns:
199  ! a(i,j) in the original matrix becomes a(i,perm(j))
200  ! in the output matrix.
201  !
202  ! job = integer ( kind = 4 ) indicating the work to be done:
203  ! job = 1 permute a, ja, ia into ao, jao, iao
204  ! (including the copying of real values ao and
205  ! the array iao).
206  ! job /= 1 : ignore real values ao and ignore iao.
207  !
208  !
209  ! on return:
210  !
211  ! ao, jao, iao = input matrix in a, ja, ia format (array ao not needed)
212  !
213  implicit none
214 
215  integer ( kind = 4 ) nrow
216 
217  real ( kind = 8 ) a(*)
218  real ( kind = 8 ) ao(*)
219  integer ( kind = 4 ) ia(nrow+1)
220  integer ( kind = 4 ) iao(nrow+1)
221  integer ( kind = 4 ) ja(*)
222  integer ( kind = 4 ) jao(*)
223  integer ( kind = 4 ) job
224  integer ( kind = 4 ) k
225  integer ( kind = 4 ) nnz
226  integer ( kind = 4 ) perm(*)
227 
228  nnz = ia(nrow+1)-1
229 
230  do k = 1, nnz
231  jao(k) = perm(ja(k))
232  end do
233  !
234  ! Done with the JA array. Return if no need to touch values.
235  !
236  if ( job /= 1 ) then
237  return
238  end if
239  !
240  ! Else get new pointers, and copy values too.
241  !
242  iao(1:nrow+1) = ia(1:nrow+1)
243 
244  ao(1:nnz) = a(1:nnz)
245 
246  return
247 end subroutine cperm
248 
249 subroutine rperm ( nrow, a, ja, ia, ao, jao, iao, perm, job )
250 
251  !*****************************************************************************80
252  !
253  !! RPERM permutes the rows of a matrix in CSR format.
254  !
255  ! Discussion:
256  !
257  ! This routine computes B = P*A where P is a permutation matrix.
258  ! the permutation P is defined through the array perm: for each j,
259  ! perm(j) represents the destination row number of row number j.
260  !
261  ! Modified:
262  !
263  ! 07 January 2004
264  !
265  ! Author:
266  !
267  ! Youcef Saad
268  !
269  ! Parameters:
270  !
271  ! Input, integer ( kind = 4 ) NROW, the row dimension of the matrix.
272  !
273  ! Input, real A(*), integer ( kind = 4 ) JA(*), IA(NROW+1), the matrix in CSR
274  ! Compressed Sparse Row format.
275  !
276  ! perm = integer ( kind = 4 ) array of length nrow containing the
277  ! permutation arrays
278  ! for the rows: perm(i) is the destination of row i in the
279  ! permuted matrix.
280  ! ---> a(i,j) in the original matrix becomes a(perm(i),j)
281  ! in the output matrix.
282  !
283  ! job = integer ( kind = 4 ) indicating the work to be done:
284  ! job = 1 permute a, ja, ia into ao, jao, iao
285  ! (including the copying of real values ao and
286  ! the array iao).
287  ! job /= 1 : ignore real values.
288  ! (in which case arrays a and ao are not needed nor
289  ! used).
290  !
291  ! Output, real AO(*), integer ( kind = 4 ) JAO(*), IAO(NROW+1), the permuted
292  ! matrix in CSR Compressed Sparse Row format.
293  !
294  ! note :
295  ! if (job/=1) then the arrays a and ao are not used.
296  !
297  implicit none
298 
299  integer ( kind = 4 ) nrow
300 
301  real ( kind = 8 ) a(*)
302  real ( kind = 8 ) ao(*)
303  integer ( kind = 4 ) i
304  integer ( kind = 4 ) ia(nrow+1)
305  integer ( kind = 4 ) iao(nrow+1)
306  integer ( kind = 4 ) ii
307  integer ( kind = 4 ) j
308  integer ( kind = 4 ) ja(*)
309  integer ( kind = 4 ) jao(*)
310  integer ( kind = 4 ) job
311  integer ( kind = 4 ) k
312  integer ( kind = 4 ) ko
313  integer ( kind = 4 ) perm(nrow)
314  logical values
315 
316  values = ( job == 1 )
317  !
318  ! Determine pointers for output matrix.
319  !
320  do j = 1, nrow
321  i = perm(j)
322  iao(i+1) = ia(j+1) - ia(j)
323  end do
324  !
325  ! Get pointers from lengths.
326  !
327  iao(1) = 1
328  do j = 1, nrow
329  iao(j+1) = iao(j+1) + iao(j)
330  end do
331  !
332  ! Copying.
333  !
334  do ii = 1, nrow
335  !
336  ! Old row = II is new row IPERM(II), and KO is the new pointer.
337  !
338  ko = iao(perm(ii))
339 
340  do k = ia(ii), ia(ii+1)-1
341  jao(ko) = ja(k)
342  if ( values ) then
343  ao(ko) = a(k)
344  end if
345  ko = ko + 1
346  end do
347 
348  end do
349 
350  return
351 end subroutine rperm
352 
353 subroutine dperm ( nrow, a, ja, ia, ao, jao, iao, perm, qperm, job )
354 
355  !*****************************************************************************80
356  !
357  !! DPERM permutes the rows and columns of a matrix stored in CSR format.
358  !
359  ! Discussion:
360  !
361  ! This routine computes P*A*Q, where P and Q are permutation matrices.
362  ! P maps row i into row perm(i) and Q maps column j into column qperm(j).
363  ! A(I,J) becomes A(perm(i),qperm(j)) in the new matrix.
364  !
365  ! In the particular case where Q is the transpose of P (symmetric
366  ! permutation of A) then qperm is not needed.
367  ! note that qperm should be of length ncol (number of columns) but this
368  ! is not checked.
369  !
370  ! The algorithm is "in place".
371  !
372  ! The column indices may not be sorted on return even if they are
373  ! sorted on entry.
374  !
375  ! In case job == 2 or job == 4, a and ao are never referred to
376  ! and can be dummy arguments.
377  !
378  ! Modified:
379  !
380  ! 07 January 2004
381  !
382  ! Author:
383  !
384  ! Youcef Saad
385  !
386  ! Parameters:
387  !
388  ! Input, integer ( kind = 4 ) NROW, the order of the matrix.
389  !
390  ! Input, real A(*), integer ( kind = 4 ) JA(*), IA(NROW+1), the matrix in CSR
391  ! Compressed Sparse Row format.
392  !
393  ! Input, integer ( kind = 4 ) PERM(NROW), the permutation array for the rows: PERM(I)
394  ! is the destination of row I in the permuted matrix; also the destination
395  ! of column I in case the permutation is symmetric (JOB <= 2).
396  !
397  ! Input, integer ( kind = 4 ) QPERM(NROW), the permutation array for the columns.
398  ! This should be provided only if JOB=3 or JOB=4, that is, only in
399  ! the case of a nonsymmetric permutation of rows and columns.
400  ! Otherwise QPERM is a dummy argument.
401  !
402  ! job = integer ( kind = 4 ) indicating the work to be done:
403  ! * job = 1,2 permutation is symmetric Ao :== P * A * transp(P)
404  ! job = 1 permute a, ja, ia into ao, jao, iao
405  ! job = 2 permute matrix ignoring real values.
406  ! * job = 3,4 permutation is non-symmetric Ao :== P * A * Q
407  ! job = 3 permute a, ja, ia into ao, jao, iao
408  ! job = 4 permute matrix ignoring real values.
409  !
410  ! Output, real AO(*), JAO(*), IAO(NROW+1), the permuted matrix in CSR
411  ! Compressed Sparse Row format.
412  !
413  implicit none
414 
415  integer ( kind = 4 ) nrow
416 
417  real ( kind = 8 ) a(*)
418  real ( kind = 8 ) ao(*)
419  integer ( kind = 4 ) ia(nrow+1)
420  integer ( kind = 4 ) iao(nrow+1)
421  integer ( kind = 4 ) ja(*)
422  integer ( kind = 4 ) jao(*)
423  integer ( kind = 4 ) job
424  integer ( kind = 4 ) locjob
425  integer ( kind = 4 ) perm(nrow)
426  integer ( kind = 4 ) qperm(nrow)
427  !
428  ! LOCJOB indicates whether or not real values must be copied.
429  !
430  locjob = mod( job, 2 )
431  !
432  ! Permute the rows first.
433  !
434  call rperm ( nrow, a, ja, ia, ao, jao, iao, perm, locjob )
435  !
436  ! Permute the columns.
437  !
438  locjob = 0
439 
440  if ( job <= 2 ) then
441  call cperm ( nrow, ao, jao, iao, ao, jao, iao, perm, locjob )
442  else
443  call cperm ( nrow, ao, jao, iao, ao, jao, iao, qperm, locjob )
444  end if
445 
446  return
447 end subroutine dperm
448 
subroutine rperm(nrow, a, ja, ia, ao, jao, iao, perm, job)
Definition: sparsekit.f90:250
subroutine amux(n, x, y, a, ja, ia)
Definition: sparsekit.f90:2
subroutine dvperm(n, x, perm)
Definition: sparsekit.f90:62
subroutine cperm(nrow, a, ja, ia, ao, jao, iao, perm, job)
Definition: sparsekit.f90:164
subroutine dperm(nrow, a, ja, ia, ao, jao, iao, perm, qperm, job)
Definition: sparsekit.f90:354