LAPACK  3.10.0
LAPACK: Linear Algebra PACKage

◆ zlasyf()

subroutine zlasyf ( character  UPLO,
integer  N,
integer  NB,
integer  KB,
complex*16, dimension( lda, * )  A,
integer  LDA,
integer, dimension( * )  IPIV,
complex*16, dimension( ldw, * )  W,
integer  LDW,
integer  INFO 
)

ZLASYF computes a partial factorization of a complex symmetric matrix using the Bunch-Kaufman diagonal pivoting method.

Download ZLASYF + dependencies [TGZ] [ZIP] [TXT]

Purpose:
 ZLASYF computes a partial factorization of a complex symmetric matrix
 A using the Bunch-Kaufman diagonal pivoting method. The partial
 factorization has the form:

 A  =  ( I  U12 ) ( A11  0  ) (  I       0    )  if UPLO = 'U', or:
       ( 0  U22 ) (  0   D  ) ( U12**T U22**T )

 A  =  ( L11  0 ) ( D    0  ) ( L11**T L21**T )  if UPLO = 'L'
       ( L21  I ) ( 0   A22 ) (  0       I    )

 where the order of D is at most NB. The actual order is returned in
 the argument KB, and is either NB or NB-1, or N if N <= NB.
 Note that U**T denotes the transpose of U.

 ZLASYF is an auxiliary routine called by ZSYTRF. It uses blocked code
 (calling Level 3 BLAS) to update the submatrix A11 (if UPLO = 'U') or
 A22 (if UPLO = 'L').
Parameters
[in]UPLO
          UPLO is CHARACTER*1
          Specifies whether the upper or lower triangular part of the
          symmetric matrix A is stored:
          = 'U':  Upper triangular
          = 'L':  Lower triangular
[in]N
          N is INTEGER
          The order of the matrix A.  N >= 0.
[in]NB
          NB is INTEGER
          The maximum number of columns of the matrix A that should be
          factored.  NB should be at least 2 to allow for 2-by-2 pivot
          blocks.
[out]KB
          KB is INTEGER
          The number of columns of A that were actually factored.
          KB is either NB-1 or NB, or N if N <= NB.
[in,out]A
          A is COMPLEX*16 array, dimension (LDA,N)
          On entry, the symmetric matrix A.  If UPLO = 'U', the leading
          n-by-n upper triangular part of A contains the upper
          triangular part of the matrix A, and the strictly lower
          triangular part of A is not referenced.  If UPLO = 'L', the
          leading n-by-n lower triangular part of A contains the lower
          triangular part of the matrix A, and the strictly upper
          triangular part of A is not referenced.
          On exit, A contains details of the partial factorization.
[in]LDA
          LDA is INTEGER
          The leading dimension of the array A.  LDA >= max(1,N).
[out]IPIV
          IPIV is INTEGER array, dimension (N)
          Details of the interchanges and the block structure of D.

          If UPLO = 'U':
             Only the last KB elements of IPIV are set.

             If IPIV(k) > 0, then rows and columns k and IPIV(k) were
             interchanged and D(k,k) is a 1-by-1 diagonal block.

             If IPIV(k) = IPIV(k-1) < 0, then rows and columns
             k-1 and -IPIV(k) were interchanged and D(k-1:k,k-1:k)
             is a 2-by-2 diagonal block.

          If UPLO = 'L':
             Only the first KB elements of IPIV are set.

             If IPIV(k) > 0, then rows and columns k and IPIV(k) were
             interchanged and D(k,k) is a 1-by-1 diagonal block.

             If IPIV(k) = IPIV(k+1) < 0, then rows and columns
             k+1 and -IPIV(k) were interchanged and D(k:k+1,k:k+1)
             is a 2-by-2 diagonal block.
[out]W
          W is COMPLEX*16 array, dimension (LDW,NB)
[in]LDW
          LDW is INTEGER
          The leading dimension of the array W.  LDW >= max(1,N).
[out]INFO
          INFO is INTEGER
          = 0: successful exit
          > 0: if INFO = k, D(k,k) is exactly zero.  The factorization
               has been completed, but the block diagonal matrix D is
               exactly singular.
Author
Univ. of Tennessee
Univ. of California Berkeley
Univ. of Colorado Denver
NAG Ltd.
Contributors:
  November 2013,  Igor Kozachenko,
                  Computer Science Division,
                  University of California, Berkeley

Definition at line 176 of file zlasyf.f.

177 *
178 * -- LAPACK computational routine --
179 * -- LAPACK is a software package provided by Univ. of Tennessee, --
180 * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
181 *
182 * .. Scalar Arguments ..
183  CHARACTER UPLO
184  INTEGER INFO, KB, LDA, LDW, N, NB
185 * ..
186 * .. Array Arguments ..
187  INTEGER IPIV( * )
188  COMPLEX*16 A( LDA, * ), W( LDW, * )
189 * ..
190 *
191 * =====================================================================
192 *
193 * .. Parameters ..
194  DOUBLE PRECISION ZERO, ONE
195  parameter( zero = 0.0d+0, one = 1.0d+0 )
196  DOUBLE PRECISION EIGHT, SEVTEN
197  parameter( eight = 8.0d+0, sevten = 17.0d+0 )
198  COMPLEX*16 CONE
199  parameter( cone = ( 1.0d+0, 0.0d+0 ) )
200 * ..
201 * .. Local Scalars ..
202  INTEGER IMAX, J, JB, JJ, JMAX, JP, K, KK, KKW, KP,
203  $ KSTEP, KW
204  DOUBLE PRECISION ABSAKK, ALPHA, COLMAX, ROWMAX
205  COMPLEX*16 D11, D21, D22, R1, T, Z
206 * ..
207 * .. External Functions ..
208  LOGICAL LSAME
209  INTEGER IZAMAX
210  EXTERNAL lsame, izamax
211 * ..
212 * .. External Subroutines ..
213  EXTERNAL zcopy, zgemm, zgemv, zscal, zswap
214 * ..
215 * .. Intrinsic Functions ..
216  INTRINSIC abs, dble, dimag, max, min, sqrt
217 * ..
218 * .. Statement Functions ..
219  DOUBLE PRECISION CABS1
220 * ..
221 * .. Statement Function definitions ..
222  cabs1( z ) = abs( dble( z ) ) + abs( dimag( z ) )
223 * ..
224 * .. Executable Statements ..
225 *
226  info = 0
227 *
228 * Initialize ALPHA for use in choosing pivot block size.
229 *
230  alpha = ( one+sqrt( sevten ) ) / eight
231 *
232  IF( lsame( uplo, 'U' ) ) THEN
233 *
234 * Factorize the trailing columns of A using the upper triangle
235 * of A and working backwards, and compute the matrix W = U12*D
236 * for use in updating A11
237 *
238 * K is the main loop index, decreasing from N in steps of 1 or 2
239 *
240 * KW is the column of W which corresponds to column K of A
241 *
242  k = n
243  10 CONTINUE
244  kw = nb + k - n
245 *
246 * Exit from loop
247 *
248  IF( ( k.LE.n-nb+1 .AND. nb.LT.n ) .OR. k.LT.1 )
249  $ GO TO 30
250 *
251 * Copy column K of A to column KW of W and update it
252 *
253  CALL zcopy( k, a( 1, k ), 1, w( 1, kw ), 1 )
254  IF( k.LT.n )
255  $ CALL zgemv( 'No transpose', k, n-k, -cone, a( 1, k+1 ), lda,
256  $ w( k, kw+1 ), ldw, cone, w( 1, kw ), 1 )
257 *
258  kstep = 1
259 *
260 * Determine rows and columns to be interchanged and whether
261 * a 1-by-1 or 2-by-2 pivot block will be used
262 *
263  absakk = cabs1( w( k, kw ) )
264 *
265 * IMAX is the row-index of the largest off-diagonal element in
266 
267 *
268  IF( k.GT.1 ) THEN
269  imax = izamax( k-1, w( 1, kw ), 1 )
270  colmax = cabs1( w( imax, kw ) )
271  ELSE
272  colmax = zero
273  END IF
274 *
275  IF( max( absakk, colmax ).EQ.zero ) THEN
276 *
277 * Column K is zero or underflow: set INFO and continue
278 *
279  IF( info.EQ.0 )
280  $ info = k
281  kp = k
282  ELSE
283  IF( absakk.GE.alpha*colmax ) THEN
284 *
285 * no interchange, use 1-by-1 pivot block
286 *
287  kp = k
288  ELSE
289 *
290 * Copy column IMAX to column KW-1 of W and update it
291 *
292  CALL zcopy( imax, a( 1, imax ), 1, w( 1, kw-1 ), 1 )
293  CALL zcopy( k-imax, a( imax, imax+1 ), lda,
294  $ w( imax+1, kw-1 ), 1 )
295  IF( k.LT.n )
296  $ CALL zgemv( 'No transpose', k, n-k, -cone,
297  $ a( 1, k+1 ), lda, w( imax, kw+1 ), ldw,
298  $ cone, w( 1, kw-1 ), 1 )
299 *
300 * JMAX is the column-index of the largest off-diagonal
301 * element in row IMAX, and ROWMAX is its absolute value
302 *
303  jmax = imax + izamax( k-imax, w( imax+1, kw-1 ), 1 )
304  rowmax = cabs1( w( jmax, kw-1 ) )
305  IF( imax.GT.1 ) THEN
306  jmax = izamax( imax-1, w( 1, kw-1 ), 1 )
307  rowmax = max( rowmax, cabs1( w( jmax, kw-1 ) ) )
308  END IF
309 *
310  IF( absakk.GE.alpha*colmax*( colmax / rowmax ) ) THEN
311 *
312 * no interchange, use 1-by-1 pivot block
313 *
314  kp = k
315  ELSE IF( cabs1( w( imax, kw-1 ) ).GE.alpha*rowmax ) THEN
316 *
317 * interchange rows and columns K and IMAX, use 1-by-1
318 * pivot block
319 *
320  kp = imax
321 *
322 * copy column KW-1 of W to column KW of W
323 *
324  CALL zcopy( k, w( 1, kw-1 ), 1, w( 1, kw ), 1 )
325  ELSE
326 *
327 * interchange rows and columns K-1 and IMAX, use 2-by-2
328 * pivot block
329 *
330  kp = imax
331  kstep = 2
332  END IF
333  END IF
334 *
335 * ============================================================
336 *
337 * KK is the column of A where pivoting step stopped
338 *
339  kk = k - kstep + 1
340 *
341 * KKW is the column of W which corresponds to column KK of A
342 *
343  kkw = nb + kk - n
344 *
345 * Interchange rows and columns KP and KK.
346 * Updated column KP is already stored in column KKW of W.
347 *
348  IF( kp.NE.kk ) THEN
349 *
350 * Copy non-updated column KK to column KP of submatrix A
351 * at step K. No need to copy element into column K
352 * (or K and K-1 for 2-by-2 pivot) of A, since these columns
353 * will be later overwritten.
354 *
355  a( kp, kp ) = a( kk, kk )
356  CALL zcopy( kk-1-kp, a( kp+1, kk ), 1, a( kp, kp+1 ),
357  $ lda )
358  IF( kp.GT.1 )
359  $ CALL zcopy( kp-1, a( 1, kk ), 1, a( 1, kp ), 1 )
360 *
361 * Interchange rows KK and KP in last K+1 to N columns of A
362 * (columns K (or K and K-1 for 2-by-2 pivot) of A will be
363 * later overwritten). Interchange rows KK and KP
364 * in last KKW to NB columns of W.
365 *
366  IF( k.LT.n )
367  $ CALL zswap( n-k, a( kk, k+1 ), lda, a( kp, k+1 ),
368  $ lda )
369  CALL zswap( n-kk+1, w( kk, kkw ), ldw, w( kp, kkw ),
370  $ ldw )
371  END IF
372 *
373  IF( kstep.EQ.1 ) THEN
374 *
375 * 1-by-1 pivot block D(k): column kw of W now holds
376 *
377 * W(kw) = U(k)*D(k),
378 *
379 * where U(k) is the k-th column of U
380 *
381 * Store subdiag. elements of column U(k)
382 * and 1-by-1 block D(k) in column k of A.
383 * NOTE: Diagonal element U(k,k) is a UNIT element
384 * and not stored.
385 * A(k,k) := D(k,k) = W(k,kw)
386 * A(1:k-1,k) := U(1:k-1,k) = W(1:k-1,kw)/D(k,k)
387 *
388  CALL zcopy( k, w( 1, kw ), 1, a( 1, k ), 1 )
389  r1 = cone / a( k, k )
390  CALL zscal( k-1, r1, a( 1, k ), 1 )
391 *
392  ELSE
393 *
394 * 2-by-2 pivot block D(k): columns kw and kw-1 of W now hold
395 *
396 * ( W(kw-1) W(kw) ) = ( U(k-1) U(k) )*D(k)
397 *
398 * where U(k) and U(k-1) are the k-th and (k-1)-th columns
399 * of U
400 *
401 * Store U(1:k-2,k-1) and U(1:k-2,k) and 2-by-2
402 * block D(k-1:k,k-1:k) in columns k-1 and k of A.
403 * NOTE: 2-by-2 diagonal block U(k-1:k,k-1:k) is a UNIT
404 * block and not stored.
405 * A(k-1:k,k-1:k) := D(k-1:k,k-1:k) = W(k-1:k,kw-1:kw)
406 * A(1:k-2,k-1:k) := U(1:k-2,k:k-1:k) =
407 * = W(1:k-2,kw-1:kw) * ( D(k-1:k,k-1:k)**(-1) )
408 *
409  IF( k.GT.2 ) THEN
410 *
411 * Compose the columns of the inverse of 2-by-2 pivot
412 * block D in the following way to reduce the number
413 * of FLOPS when we myltiply panel ( W(kw-1) W(kw) ) by
414 * this inverse
415 *
416 * D**(-1) = ( d11 d21 )**(-1) =
417 * ( d21 d22 )
418 *
419 * = 1/(d11*d22-d21**2) * ( ( d22 ) (-d21 ) ) =
420 * ( (-d21 ) ( d11 ) )
421 *
422 * = 1/d21 * 1/((d11/d21)*(d22/d21)-1) *
423 *
424 * * ( ( d22/d21 ) ( -1 ) ) =
425 * ( ( -1 ) ( d11/d21 ) )
426 *
427 * = 1/d21 * 1/(D22*D11-1) * ( ( D11 ) ( -1 ) ) =
428 * ( ( -1 ) ( D22 ) )
429 *
430 * = 1/d21 * T * ( ( D11 ) ( -1 ) )
431 * ( ( -1 ) ( D22 ) )
432 *
433 * = D21 * ( ( D11 ) ( -1 ) )
434 * ( ( -1 ) ( D22 ) )
435 *
436  d21 = w( k-1, kw )
437  d11 = w( k, kw ) / d21
438  d22 = w( k-1, kw-1 ) / d21
439  t = cone / ( d11*d22-cone )
440  d21 = t / d21
441 *
442 * Update elements in columns A(k-1) and A(k) as
443 * dot products of rows of ( W(kw-1) W(kw) ) and columns
444 * of D**(-1)
445 *
446  DO 20 j = 1, k - 2
447  a( j, k-1 ) = d21*( d11*w( j, kw-1 )-w( j, kw ) )
448  a( j, k ) = d21*( d22*w( j, kw )-w( j, kw-1 ) )
449  20 CONTINUE
450  END IF
451 *
452 * Copy D(k) to A
453 *
454  a( k-1, k-1 ) = w( k-1, kw-1 )
455  a( k-1, k ) = w( k-1, kw )
456  a( k, k ) = w( k, kw )
457 *
458  END IF
459 *
460  END IF
461 *
462 * Store details of the interchanges in IPIV
463 *
464  IF( kstep.EQ.1 ) THEN
465  ipiv( k ) = kp
466  ELSE
467  ipiv( k ) = -kp
468  ipiv( k-1 ) = -kp
469  END IF
470 *
471 * Decrease K and return to the start of the main loop
472 *
473  k = k - kstep
474  GO TO 10
475 *
476  30 CONTINUE
477 *
478 * Update the upper triangle of A11 (= A(1:k,1:k)) as
479 *
480 * A11 := A11 - U12*D*U12**T = A11 - U12*W**T
481 *
482 * computing blocks of NB columns at a time
483 *
484  DO 50 j = ( ( k-1 ) / nb )*nb + 1, 1, -nb
485  jb = min( nb, k-j+1 )
486 *
487 * Update the upper triangle of the diagonal block
488 *
489  DO 40 jj = j, j + jb - 1
490  CALL zgemv( 'No transpose', jj-j+1, n-k, -cone,
491  $ a( j, k+1 ), lda, w( jj, kw+1 ), ldw, cone,
492  $ a( j, jj ), 1 )
493  40 CONTINUE
494 *
495 * Update the rectangular superdiagonal block
496 *
497  CALL zgemm( 'No transpose', 'Transpose', j-1, jb, n-k,
498  $ -cone, a( 1, k+1 ), lda, w( j, kw+1 ), ldw,
499  $ cone, a( 1, j ), lda )
500  50 CONTINUE
501 *
502 * Put U12 in standard form by partially undoing the interchanges
503 * in columns k+1:n looping backwards from k+1 to n
504 *
505  j = k + 1
506  60 CONTINUE
507 *
508 * Undo the interchanges (if any) of rows JJ and JP at each
509 * step J
510 *
511 * (Here, J is a diagonal index)
512  jj = j
513  jp = ipiv( j )
514  IF( jp.LT.0 ) THEN
515  jp = -jp
516 * (Here, J is a diagonal index)
517  j = j + 1
518  END IF
519 * (NOTE: Here, J is used to determine row length. Length N-J+1
520 * of the rows to swap back doesn't include diagonal element)
521  j = j + 1
522  IF( jp.NE.jj .AND. j.LE.n )
523  $ CALL zswap( n-j+1, a( jp, j ), lda, a( jj, j ), lda )
524  IF( j.LT.n )
525  $ GO TO 60
526 *
527 * Set KB to the number of columns factorized
528 *
529  kb = n - k
530 *
531  ELSE
532 *
533 * Factorize the leading columns of A using the lower triangle
534 * of A and working forwards, and compute the matrix W = L21*D
535 * for use in updating A22
536 *
537 * K is the main loop index, increasing from 1 in steps of 1 or 2
538 *
539  k = 1
540  70 CONTINUE
541 *
542 * Exit from loop
543 *
544  IF( ( k.GE.nb .AND. nb.LT.n ) .OR. k.GT.n )
545  $ GO TO 90
546 *
547 * Copy column K of A to column K of W and update it
548 *
549  CALL zcopy( n-k+1, a( k, k ), 1, w( k, k ), 1 )
550  CALL zgemv( 'No transpose', n-k+1, k-1, -cone, a( k, 1 ), lda,
551  $ w( k, 1 ), ldw, cone, w( k, k ), 1 )
552 *
553  kstep = 1
554 *
555 * Determine rows and columns to be interchanged and whether
556 * a 1-by-1 or 2-by-2 pivot block will be used
557 *
558  absakk = cabs1( w( k, k ) )
559 *
560 * IMAX is the row-index of the largest off-diagonal element in
561 
562 *
563  IF( k.LT.n ) THEN
564  imax = k + izamax( n-k, w( k+1, k ), 1 )
565  colmax = cabs1( w( imax, k ) )
566  ELSE
567  colmax = zero
568  END IF
569 *
570  IF( max( absakk, colmax ).EQ.zero ) THEN
571 *
572 * Column K is zero or underflow: set INFO and continue
573 *
574  IF( info.EQ.0 )
575  $ info = k
576  kp = k
577  ELSE
578  IF( absakk.GE.alpha*colmax ) THEN
579 *
580 * no interchange, use 1-by-1 pivot block
581 *
582  kp = k
583  ELSE
584 *
585 * Copy column IMAX to column K+1 of W and update it
586 *
587  CALL zcopy( imax-k, a( imax, k ), lda, w( k, k+1 ), 1 )
588  CALL zcopy( n-imax+1, a( imax, imax ), 1, w( imax, k+1 ),
589  $ 1 )
590  CALL zgemv( 'No transpose', n-k+1, k-1, -cone, a( k, 1 ),
591  $ lda, w( imax, 1 ), ldw, cone, w( k, k+1 ),
592  $ 1 )
593 *
594 * JMAX is the column-index of the largest off-diagonal
595 * element in row IMAX, and ROWMAX is its absolute value
596 *
597  jmax = k - 1 + izamax( imax-k, w( k, k+1 ), 1 )
598  rowmax = cabs1( w( jmax, k+1 ) )
599  IF( imax.LT.n ) THEN
600  jmax = imax + izamax( n-imax, w( imax+1, k+1 ), 1 )
601  rowmax = max( rowmax, cabs1( w( jmax, k+1 ) ) )
602  END IF
603 *
604  IF( absakk.GE.alpha*colmax*( colmax / rowmax ) ) THEN
605 *
606 * no interchange, use 1-by-1 pivot block
607 *
608  kp = k
609  ELSE IF( cabs1( w( imax, k+1 ) ).GE.alpha*rowmax ) THEN
610 *
611 * interchange rows and columns K and IMAX, use 1-by-1
612 * pivot block
613 *
614  kp = imax
615 *
616 * copy column K+1 of W to column K of W
617 *
618  CALL zcopy( n-k+1, w( k, k+1 ), 1, w( k, k ), 1 )
619  ELSE
620 *
621 * interchange rows and columns K+1 and IMAX, use 2-by-2
622 * pivot block
623 *
624  kp = imax
625  kstep = 2
626  END IF
627  END IF
628 *
629 * ============================================================
630 *
631 * KK is the column of A where pivoting step stopped
632 *
633  kk = k + kstep - 1
634 *
635 * Interchange rows and columns KP and KK.
636 * Updated column KP is already stored in column KK of W.
637 *
638  IF( kp.NE.kk ) THEN
639 *
640 * Copy non-updated column KK to column KP of submatrix A
641 * at step K. No need to copy element into column K
642 * (or K and K+1 for 2-by-2 pivot) of A, since these columns
643 * will be later overwritten.
644 *
645  a( kp, kp ) = a( kk, kk )
646  CALL zcopy( kp-kk-1, a( kk+1, kk ), 1, a( kp, kk+1 ),
647  $ lda )
648  IF( kp.LT.n )
649  $ CALL zcopy( n-kp, a( kp+1, kk ), 1, a( kp+1, kp ), 1 )
650 *
651 * Interchange rows KK and KP in first K-1 columns of A
652 * (columns K (or K and K+1 for 2-by-2 pivot) of A will be
653 * later overwritten). Interchange rows KK and KP
654 * in first KK columns of W.
655 *
656  IF( k.GT.1 )
657  $ CALL zswap( k-1, a( kk, 1 ), lda, a( kp, 1 ), lda )
658  CALL zswap( kk, w( kk, 1 ), ldw, w( kp, 1 ), ldw )
659  END IF
660 *
661  IF( kstep.EQ.1 ) THEN
662 *
663 * 1-by-1 pivot block D(k): column k of W now holds
664 *
665 * W(k) = L(k)*D(k),
666 *
667 * where L(k) is the k-th column of L
668 *
669 * Store subdiag. elements of column L(k)
670 * and 1-by-1 block D(k) in column k of A.
671 * (NOTE: Diagonal element L(k,k) is a UNIT element
672 * and not stored)
673 * A(k,k) := D(k,k) = W(k,k)
674 * A(k+1:N,k) := L(k+1:N,k) = W(k+1:N,k)/D(k,k)
675 *
676  CALL zcopy( n-k+1, w( k, k ), 1, a( k, k ), 1 )
677  IF( k.LT.n ) THEN
678  r1 = cone / a( k, k )
679  CALL zscal( n-k, r1, a( k+1, k ), 1 )
680  END IF
681 *
682  ELSE
683 *
684 * 2-by-2 pivot block D(k): columns k and k+1 of W now hold
685 *
686 * ( W(k) W(k+1) ) = ( L(k) L(k+1) )*D(k)
687 *
688 * where L(k) and L(k+1) are the k-th and (k+1)-th columns
689 * of L
690 *
691 * Store L(k+2:N,k) and L(k+2:N,k+1) and 2-by-2
692 * block D(k:k+1,k:k+1) in columns k and k+1 of A.
693 * (NOTE: 2-by-2 diagonal block L(k:k+1,k:k+1) is a UNIT
694 * block and not stored)
695 * A(k:k+1,k:k+1) := D(k:k+1,k:k+1) = W(k:k+1,k:k+1)
696 * A(k+2:N,k:k+1) := L(k+2:N,k:k+1) =
697 * = W(k+2:N,k:k+1) * ( D(k:k+1,k:k+1)**(-1) )
698 *
699  IF( k.LT.n-1 ) THEN
700 *
701 * Compose the columns of the inverse of 2-by-2 pivot
702 * block D in the following way to reduce the number
703 * of FLOPS when we myltiply panel ( W(k) W(k+1) ) by
704 * this inverse
705 *
706 * D**(-1) = ( d11 d21 )**(-1) =
707 * ( d21 d22 )
708 *
709 * = 1/(d11*d22-d21**2) * ( ( d22 ) (-d21 ) ) =
710 * ( (-d21 ) ( d11 ) )
711 *
712 * = 1/d21 * 1/((d11/d21)*(d22/d21)-1) *
713 *
714 * * ( ( d22/d21 ) ( -1 ) ) =
715 * ( ( -1 ) ( d11/d21 ) )
716 *
717 * = 1/d21 * 1/(D22*D11-1) * ( ( D11 ) ( -1 ) ) =
718 * ( ( -1 ) ( D22 ) )
719 *
720 * = 1/d21 * T * ( ( D11 ) ( -1 ) )
721 * ( ( -1 ) ( D22 ) )
722 *
723 * = D21 * ( ( D11 ) ( -1 ) )
724 * ( ( -1 ) ( D22 ) )
725 *
726  d21 = w( k+1, k )
727  d11 = w( k+1, k+1 ) / d21
728  d22 = w( k, k ) / d21
729  t = cone / ( d11*d22-cone )
730  d21 = t / d21
731 *
732 * Update elements in columns A(k) and A(k+1) as
733 * dot products of rows of ( W(k) W(k+1) ) and columns
734 * of D**(-1)
735 *
736  DO 80 j = k + 2, n
737  a( j, k ) = d21*( d11*w( j, k )-w( j, k+1 ) )
738  a( j, k+1 ) = d21*( d22*w( j, k+1 )-w( j, k ) )
739  80 CONTINUE
740  END IF
741 *
742 * Copy D(k) to A
743 *
744  a( k, k ) = w( k, k )
745  a( k+1, k ) = w( k+1, k )
746  a( k+1, k+1 ) = w( k+1, k+1 )
747 *
748  END IF
749 *
750  END IF
751 *
752 * Store details of the interchanges in IPIV
753 *
754  IF( kstep.EQ.1 ) THEN
755  ipiv( k ) = kp
756  ELSE
757  ipiv( k ) = -kp
758  ipiv( k+1 ) = -kp
759  END IF
760 *
761 * Increase K and return to the start of the main loop
762 *
763  k = k + kstep
764  GO TO 70
765 *
766  90 CONTINUE
767 *
768 * Update the lower triangle of A22 (= A(k:n,k:n)) as
769 *
770 * A22 := A22 - L21*D*L21**T = A22 - L21*W**T
771 *
772 * computing blocks of NB columns at a time
773 *
774  DO 110 j = k, n, nb
775  jb = min( nb, n-j+1 )
776 *
777 * Update the lower triangle of the diagonal block
778 *
779  DO 100 jj = j, j + jb - 1
780  CALL zgemv( 'No transpose', j+jb-jj, k-1, -cone,
781  $ a( jj, 1 ), lda, w( jj, 1 ), ldw, cone,
782  $ a( jj, jj ), 1 )
783  100 CONTINUE
784 *
785 * Update the rectangular subdiagonal block
786 *
787  IF( j+jb.LE.n )
788  $ CALL zgemm( 'No transpose', 'Transpose', n-j-jb+1, jb,
789  $ k-1, -cone, a( j+jb, 1 ), lda, w( j, 1 ),
790  $ ldw, cone, a( j+jb, j ), lda )
791  110 CONTINUE
792 *
793 * Put L21 in standard form by partially undoing the interchanges
794 * of rows in columns 1:k-1 looping backwards from k-1 to 1
795 *
796  j = k - 1
797  120 CONTINUE
798 *
799 * Undo the interchanges (if any) of rows JJ and JP at each
800 * step J
801 *
802 * (Here, J is a diagonal index)
803  jj = j
804  jp = ipiv( j )
805  IF( jp.LT.0 ) THEN
806  jp = -jp
807 * (Here, J is a diagonal index)
808  j = j - 1
809  END IF
810 * (NOTE: Here, J is used to determine row length. Length J
811 * of the rows to swap back doesn't include diagonal element)
812  j = j - 1
813  IF( jp.NE.jj .AND. j.GE.1 )
814  $ CALL zswap( j, a( jp, 1 ), lda, a( jj, 1 ), lda )
815  IF( j.GT.1 )
816  $ GO TO 120
817 *
818 * Set KB to the number of columns factorized
819 *
820  kb = k - 1
821 *
822  END IF
823  RETURN
824 *
825 * End of ZLASYF
826 *
integer function izamax(N, ZX, INCX)
IZAMAX
Definition: izamax.f:71
logical function lsame(CA, CB)
LSAME
Definition: lsame.f:53
subroutine zswap(N, ZX, INCX, ZY, INCY)
ZSWAP
Definition: zswap.f:81
subroutine zscal(N, ZA, ZX, INCX)
ZSCAL
Definition: zscal.f:78
subroutine zcopy(N, ZX, INCX, ZY, INCY)
ZCOPY
Definition: zcopy.f:81
subroutine zgemv(TRANS, M, N, ALPHA, A, LDA, X, INCX, BETA, Y, INCY)
ZGEMV
Definition: zgemv.f:158
subroutine zgemm(TRANSA, TRANSB, M, N, K, ALPHA, A, LDA, B, LDB, BETA, C, LDC)
ZGEMM
Definition: zgemm.f:187
Here is the call graph for this function:
Here is the caller graph for this function: