001:       SUBROUTINE CHETRD( UPLO, N, A, LDA, D, E, TAU, WORK, LWORK, INFO )
002: *
003: *  -- LAPACK routine (version 3.2) --
004: *     Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd..
005: *     November 2006
006: *
007: *     .. Scalar Arguments ..
008:       CHARACTER          UPLO
009:       INTEGER            INFO, LDA, LWORK, N
010: *     ..
011: *     .. Array Arguments ..
012:       REAL               D( * ), E( * )
013:       COMPLEX            A( LDA, * ), TAU( * ), WORK( * )
014: *     ..
015: *
016: *  Purpose
017: *  =======
018: *
019: *  CHETRD reduces a complex Hermitian matrix A to real symmetric
020: *  tridiagonal form T by a unitary similarity transformation:
021: *  Q**H * A * Q = T.
022: *
023: *  Arguments
024: *  =========
025: *
026: *  UPLO    (input) CHARACTER*1
027: *          = 'U':  Upper triangle of A is stored;
028: *          = 'L':  Lower triangle of A is stored.
029: *
030: *  N       (input) INTEGER
031: *          The order of the matrix A.  N >= 0.
032: *
033: *  A       (input/output) COMPLEX array, dimension (LDA,N)
034: *          On entry, the Hermitian matrix A.  If UPLO = 'U', the leading
035: *          N-by-N upper triangular part of A contains the upper
036: *          triangular part of the matrix A, and the strictly lower
037: *          triangular part of A is not referenced.  If UPLO = 'L', the
038: *          leading N-by-N lower triangular part of A contains the lower
039: *          triangular part of the matrix A, and the strictly upper
040: *          triangular part of A is not referenced.
041: *          On exit, if UPLO = 'U', the diagonal and first superdiagonal
042: *          of A are overwritten by the corresponding elements of the
043: *          tridiagonal matrix T, and the elements above the first
044: *          superdiagonal, with the array TAU, represent the unitary
045: *          matrix Q as a product of elementary reflectors; if UPLO
046: *          = 'L', the diagonal and first subdiagonal of A are over-
047: *          written by the corresponding elements of the tridiagonal
048: *          matrix T, and the elements below the first subdiagonal, with
049: *          the array TAU, represent the unitary matrix Q as a product
050: *          of elementary reflectors. See Further Details.
051: *
052: *  LDA     (input) INTEGER
053: *          The leading dimension of the array A.  LDA >= max(1,N).
054: *
055: *  D       (output) REAL array, dimension (N)
056: *          The diagonal elements of the tridiagonal matrix T:
057: *          D(i) = A(i,i).
058: *
059: *  E       (output) REAL array, dimension (N-1)
060: *          The off-diagonal elements of the tridiagonal matrix T:
061: *          E(i) = A(i,i+1) if UPLO = 'U', E(i) = A(i+1,i) if UPLO = 'L'.
062: *
063: *  TAU     (output) COMPLEX array, dimension (N-1)
064: *          The scalar factors of the elementary reflectors (see Further
065: *          Details).
066: *
067: *  WORK    (workspace/output) COMPLEX array, dimension (MAX(1,LWORK))
068: *          On exit, if INFO = 0, WORK(1) returns the optimal LWORK.
069: *
070: *  LWORK   (input) INTEGER
071: *          The dimension of the array WORK.  LWORK >= 1.
072: *          For optimum performance LWORK >= N*NB, where NB is the
073: *          optimal blocksize.
074: *
075: *          If LWORK = -1, then a workspace query is assumed; the routine
076: *          only calculates the optimal size of the WORK array, returns
077: *          this value as the first entry of the WORK array, and no error
078: *          message related to LWORK is issued by XERBLA.
079: *
080: *  INFO    (output) INTEGER
081: *          = 0:  successful exit
082: *          < 0:  if INFO = -i, the i-th argument had an illegal value
083: *
084: *  Further Details
085: *  ===============
086: *
087: *  If UPLO = 'U', the matrix Q is represented as a product of elementary
088: *  reflectors
089: *
090: *     Q = H(n-1) . . . H(2) H(1).
091: *
092: *  Each H(i) has the form
093: *
094: *     H(i) = I - tau * v * v'
095: *
096: *  where tau is a complex scalar, and v is a complex vector with
097: *  v(i+1:n) = 0 and v(i) = 1; v(1:i-1) is stored on exit in
098: *  A(1:i-1,i+1), and tau in TAU(i).
099: *
100: *  If UPLO = 'L', the matrix Q is represented as a product of elementary
101: *  reflectors
102: *
103: *     Q = H(1) H(2) . . . H(n-1).
104: *
105: *  Each H(i) has the form
106: *
107: *     H(i) = I - tau * v * v'
108: *
109: *  where tau is a complex scalar, and v is a complex vector with
110: *  v(1:i) = 0 and v(i+1) = 1; v(i+2:n) is stored on exit in A(i+2:n,i),
111: *  and tau in TAU(i).
112: *
113: *  The contents of A on exit are illustrated by the following examples
114: *  with n = 5:
115: *
116: *  if UPLO = 'U':                       if UPLO = 'L':
117: *
118: *    (  d   e   v2  v3  v4 )              (  d                  )
119: *    (      d   e   v3  v4 )              (  e   d              )
120: *    (          d   e   v4 )              (  v1  e   d          )
121: *    (              d   e  )              (  v1  v2  e   d      )
122: *    (                  d  )              (  v1  v2  v3  e   d  )
123: *
124: *  where d and e denote diagonal and off-diagonal elements of T, and vi
125: *  denotes an element of the vector defining H(i).
126: *
127: *  =====================================================================
128: *
129: *     .. Parameters ..
130:       REAL               ONE
131:       PARAMETER          ( ONE = 1.0E+0 )
132:       COMPLEX            CONE
133:       PARAMETER          ( CONE = ( 1.0E+0, 0.0E+0 ) )
134: *     ..
135: *     .. Local Scalars ..
136:       LOGICAL            LQUERY, UPPER
137:       INTEGER            I, IINFO, IWS, J, KK, LDWORK, LWKOPT, NB,
138:      $                   NBMIN, NX
139: *     ..
140: *     .. External Subroutines ..
141:       EXTERNAL           CHER2K, CHETD2, CLATRD, XERBLA
142: *     ..
143: *     .. Intrinsic Functions ..
144:       INTRINSIC          MAX
145: *     ..
146: *     .. External Functions ..
147:       LOGICAL            LSAME
148:       INTEGER            ILAENV
149:       EXTERNAL           LSAME, ILAENV
150: *     ..
151: *     .. Executable Statements ..
152: *
153: *     Test the input parameters
154: *
155:       INFO = 0
156:       UPPER = LSAME( UPLO, 'U' )
157:       LQUERY = ( LWORK.EQ.-1 )
158:       IF( .NOT.UPPER .AND. .NOT.LSAME( UPLO, 'L' ) ) THEN
159:          INFO = -1
160:       ELSE IF( N.LT.0 ) THEN
161:          INFO = -2
162:       ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
163:          INFO = -4
164:       ELSE IF( LWORK.LT.1 .AND. .NOT.LQUERY ) THEN
165:          INFO = -9
166:       END IF
167: *
168:       IF( INFO.EQ.0 ) THEN
169: *
170: *        Determine the block size.
171: *
172:          NB = ILAENV( 1, 'CHETRD', UPLO, N, -1, -1, -1 )
173:          LWKOPT = N*NB
174:          WORK( 1 ) = LWKOPT
175:       END IF
176: *
177:       IF( INFO.NE.0 ) THEN
178:          CALL XERBLA( 'CHETRD', -INFO )
179:          RETURN
180:       ELSE IF( LQUERY ) THEN
181:          RETURN
182:       END IF
183: *
184: *     Quick return if possible
185: *
186:       IF( N.EQ.0 ) THEN
187:          WORK( 1 ) = 1
188:          RETURN
189:       END IF
190: *
191:       NX = N
192:       IWS = 1
193:       IF( NB.GT.1 .AND. NB.LT.N ) THEN
194: *
195: *        Determine when to cross over from blocked to unblocked code
196: *        (last block is always handled by unblocked code).
197: *
198:          NX = MAX( NB, ILAENV( 3, 'CHETRD', UPLO, N, -1, -1, -1 ) )
199:          IF( NX.LT.N ) THEN
200: *
201: *           Determine if workspace is large enough for blocked code.
202: *
203:             LDWORK = N
204:             IWS = LDWORK*NB
205:             IF( LWORK.LT.IWS ) THEN
206: *
207: *              Not enough workspace to use optimal NB:  determine the
208: *              minimum value of NB, and reduce NB or force use of
209: *              unblocked code by setting NX = N.
210: *
211:                NB = MAX( LWORK / LDWORK, 1 )
212:                NBMIN = ILAENV( 2, 'CHETRD', UPLO, N, -1, -1, -1 )
213:                IF( NB.LT.NBMIN )
214:      $            NX = N
215:             END IF
216:          ELSE
217:             NX = N
218:          END IF
219:       ELSE
220:          NB = 1
221:       END IF
222: *
223:       IF( UPPER ) THEN
224: *
225: *        Reduce the upper triangle of A.
226: *        Columns 1:kk are handled by the unblocked method.
227: *
228:          KK = N - ( ( N-NX+NB-1 ) / NB )*NB
229:          DO 20 I = N - NB + 1, KK + 1, -NB
230: *
231: *           Reduce columns i:i+nb-1 to tridiagonal form and form the
232: *           matrix W which is needed to update the unreduced part of
233: *           the matrix
234: *
235:             CALL CLATRD( UPLO, I+NB-1, NB, A, LDA, E, TAU, WORK,
236:      $                   LDWORK )
237: *
238: *           Update the unreduced submatrix A(1:i-1,1:i-1), using an
239: *           update of the form:  A := A - V*W' - W*V'
240: *
241:             CALL CHER2K( UPLO, 'No transpose', I-1, NB, -CONE,
242:      $                   A( 1, I ), LDA, WORK, LDWORK, ONE, A, LDA )
243: *
244: *           Copy superdiagonal elements back into A, and diagonal
245: *           elements into D
246: *
247:             DO 10 J = I, I + NB - 1
248:                A( J-1, J ) = E( J-1 )
249:                D( J ) = A( J, J )
250:    10       CONTINUE
251:    20    CONTINUE
252: *
253: *        Use unblocked code to reduce the last or only block
254: *
255:          CALL CHETD2( UPLO, KK, A, LDA, D, E, TAU, IINFO )
256:       ELSE
257: *
258: *        Reduce the lower triangle of A
259: *
260:          DO 40 I = 1, N - NX, NB
261: *
262: *           Reduce columns i:i+nb-1 to tridiagonal form and form the
263: *           matrix W which is needed to update the unreduced part of
264: *           the matrix
265: *
266:             CALL CLATRD( UPLO, N-I+1, NB, A( I, I ), LDA, E( I ),
267:      $                   TAU( I ), WORK, LDWORK )
268: *
269: *           Update the unreduced submatrix A(i+nb:n,i+nb:n), using
270: *           an update of the form:  A := A - V*W' - W*V'
271: *
272:             CALL CHER2K( UPLO, 'No transpose', N-I-NB+1, NB, -CONE,
273:      $                   A( I+NB, I ), LDA, WORK( NB+1 ), LDWORK, ONE,
274:      $                   A( I+NB, I+NB ), LDA )
275: *
276: *           Copy subdiagonal elements back into A, and diagonal
277: *           elements into D
278: *
279:             DO 30 J = I, I + NB - 1
280:                A( J+1, J ) = E( J )
281:                D( J ) = A( J, J )
282:    30       CONTINUE
283:    40    CONTINUE
284: *
285: *        Use unblocked code to reduce the last or only block
286: *
287:          CALL CHETD2( UPLO, N-I+1, A( I, I ), LDA, D( I ), E( I ),
288:      $                TAU( I ), IINFO )
289:       END IF
290: *
291:       WORK( 1 ) = LWKOPT
292:       RETURN
293: *
294: *     End of CHETRD
295: *
296:       END
297: