1       SUBROUTINE DLAR1V( N, B1, BN, LAMBDA, D, L, LD, LLD,
  2      $           PIVMIN, GAPTOL, Z, WANTNC, NEGCNT, ZTZ, MINGMA,
  3      $           R, ISUPPZ, NRMINV, RESID, RQCORR, WORK )
  4 *
  5 *  -- LAPACK auxiliary routine (version 3.3.1) --
  6 *  -- LAPACK is a software package provided by Univ. of Tennessee,    --
  7 *  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
  8 *  -- April 2011                                                      --
  9 *
 10 *     .. Scalar Arguments ..
 11       LOGICAL            WANTNC
 12       INTEGER   B1, BN, N, NEGCNT, R
 13       DOUBLE PRECISION   GAPTOL, LAMBDA, MINGMA, NRMINV, PIVMIN, RESID,
 14      $                   RQCORR, ZTZ
 15 *     ..
 16 *     .. Array Arguments ..
 17       INTEGER            ISUPPZ( * )
 18       DOUBLE PRECISION   D( * ), L( * ), LD( * ), LLD( * ),
 19      $                  WORK( * )
 20       DOUBLE PRECISION Z( * )
 21 *     ..
 22 *
 23 *  Purpose
 24 *  =======
 25 *
 26 *  DLAR1V computes the (scaled) r-th column of the inverse of
 27 *  the sumbmatrix in rows B1 through BN of the tridiagonal matrix
 28 *  L D L**T - sigma I. When sigma is close to an eigenvalue, the
 29 *  computed vector is an accurate eigenvector. Usually, r corresponds
 30 *  to the index where the eigenvector is largest in magnitude.
 31 *  The following steps accomplish this computation :
 32 *  (a) Stationary qd transform,  L D L**T - sigma I = L(+) D(+) L(+)**T,
 33 *  (b) Progressive qd transform, L D L**T - sigma I = U(-) D(-) U(-)**T,
 34 *  (c) Computation of the diagonal elements of the inverse of
 35 *      L D L**T - sigma I by combining the above transforms, and choosing
 36 *      r as the index where the diagonal of the inverse is (one of the)
 37 *      largest in magnitude.
 38 *  (d) Computation of the (scaled) r-th column of the inverse using the
 39 *      twisted factorization obtained by combining the top part of the
 40 *      the stationary and the bottom part of the progressive transform.
 41 *
 42 *  Arguments
 43 *  =========
 44 *
 45 *  N        (input) INTEGER
 46 *           The order of the matrix L D L**T.
 47 *
 48 *  B1       (input) INTEGER
 49 *           First index of the submatrix of L D L**T.
 50 *
 51 *  BN       (input) INTEGER
 52 *           Last index of the submatrix of L D L**T.
 53 *
 54 *  LAMBDA    (input) DOUBLE PRECISION
 55 *           The shift. In order to compute an accurate eigenvector,
 56 *           LAMBDA should be a good approximation to an eigenvalue
 57 *           of L D L**T.
 58 *
 59 *  L        (input) DOUBLE PRECISION array, dimension (N-1)
 60 *           The (n-1) subdiagonal elements of the unit bidiagonal matrix
 61 *           L, in elements 1 to N-1.
 62 *
 63 *  D        (input) DOUBLE PRECISION array, dimension (N)
 64 *           The n diagonal elements of the diagonal matrix D.
 65 *
 66 *  LD       (input) DOUBLE PRECISION array, dimension (N-1)
 67 *           The n-1 elements L(i)*D(i).
 68 *
 69 *  LLD      (input) DOUBLE PRECISION array, dimension (N-1)
 70 *           The n-1 elements L(i)*L(i)*D(i).
 71 *
 72 *  PIVMIN   (input) DOUBLE PRECISION
 73 *           The minimum pivot in the Sturm sequence.
 74 *
 75 *  GAPTOL   (input) DOUBLE PRECISION
 76 *           Tolerance that indicates when eigenvector entries are negligible
 77 *           w.r.t. their contribution to the residual.
 78 *
 79 *  Z        (input/output) DOUBLE PRECISION array, dimension (N)
 80 *           On input, all entries of Z must be set to 0.
 81 *           On output, Z contains the (scaled) r-th column of the
 82 *           inverse. The scaling is such that Z(R) equals 1.
 83 *
 84 *  WANTNC   (input) LOGICAL
 85 *           Specifies whether NEGCNT has to be computed.
 86 *
 87 *  NEGCNT   (output) INTEGER
 88 *           If WANTNC is .TRUE. then NEGCNT = the number of pivots < pivmin
 89 *           in the  matrix factorization L D L**T, and NEGCNT = -1 otherwise.
 90 *
 91 *  ZTZ      (output) DOUBLE PRECISION
 92 *           The square of the 2-norm of Z.
 93 *
 94 *  MINGMA   (output) DOUBLE PRECISION
 95 *           The reciprocal of the largest (in magnitude) diagonal
 96 *           element of the inverse of L D L**T - sigma I.
 97 *
 98 *  R        (input/output) INTEGER
 99 *           The twist index for the twisted factorization used to
100 *           compute Z.
101 *           On input, 0 <= R <= N. If R is input as 0, R is set to
102 *           the index where (L D L**T - sigma I)^{-1} is largest
103 *           in magnitude. If 1 <= R <= N, R is unchanged.
104 *           On output, R contains the twist index used to compute Z.
105 *           Ideally, R designates the position of the maximum entry in the
106 *           eigenvector.
107 *
108 *  ISUPPZ   (output) INTEGER array, dimension (2)
109 *           The support of the vector in Z, i.e., the vector Z is
110 *           nonzero only in elements ISUPPZ(1) through ISUPPZ( 2 ).
111 *
112 *  NRMINV   (output) DOUBLE PRECISION
113 *           NRMINV = 1/SQRT( ZTZ )
114 *
115 *  RESID    (output) DOUBLE PRECISION
116 *           The residual of the FP vector.
117 *           RESID = ABS( MINGMA )/SQRT( ZTZ )
118 *
119 *  RQCORR   (output) DOUBLE PRECISION
120 *           The Rayleigh Quotient correction to LAMBDA.
121 *           RQCORR = MINGMA*TMP
122 *
123 *  WORK     (workspace) DOUBLE PRECISION array, dimension (4*N)
124 *
125 *  Further Details
126 *  ===============
127 *
128 *  Based on contributions by
129 *     Beresford Parlett, University of California, Berkeley, USA
130 *     Jim Demmel, University of California, Berkeley, USA
131 *     Inderjit Dhillon, University of Texas, Austin, USA
132 *     Osni Marques, LBNL/NERSC, USA
133 *     Christof Voemel, University of California, Berkeley, USA
134 *
135 *  =====================================================================
136 *
137 *     .. Parameters ..
138       DOUBLE PRECISION   ZERO, ONE
139       PARAMETER          ( ZERO = 0.0D0, ONE = 1.0D0 )
140 
141 *     ..
142 *     .. Local Scalars ..
143       LOGICAL            SAWNAN1, SAWNAN2
144       INTEGER            I, INDLPL, INDP, INDS, INDUMN, NEG1, NEG2, R1,
145      $                   R2
146       DOUBLE PRECISION   DMINUS, DPLUS, EPS, S, TMP
147 *     ..
148 *     .. External Functions ..
149       LOGICAL DISNAN
150       DOUBLE PRECISION   DLAMCH
151       EXTERNAL           DISNAN, DLAMCH
152 *     ..
153 *     .. Intrinsic Functions ..
154       INTRINSIC          ABS
155 *     ..
156 *     .. Executable Statements ..
157 *
158       EPS = DLAMCH( 'Precision' )
159 
160 
161       IF( R.EQ.0 ) THEN
162          R1 = B1
163          R2 = BN
164       ELSE
165          R1 = R
166          R2 = R
167       END IF
168 
169 *     Storage for LPLUS
170       INDLPL = 0
171 *     Storage for UMINUS
172       INDUMN = N
173       INDS = 2*+ 1
174       INDP = 3*+ 1
175 
176       IF( B1.EQ.1 ) THEN
177          WORK( INDS ) = ZERO
178       ELSE
179          WORK( INDS+B1-1 ) = LLD( B1-1 )
180       END IF
181 
182 *
183 *     Compute the stationary transform (using the differential form)
184 *     until the index R2.
185 *
186       SAWNAN1 = .FALSE.
187       NEG1 = 0
188       S = WORK( INDS+B1-1 ) - LAMBDA
189       DO 50 I = B1, R1 - 1
190          DPLUS = D( I ) + S
191          WORK( INDLPL+I ) = LD( I ) / DPLUS
192          IF(DPLUS.LT.ZERO) NEG1 = NEG1 + 1
193          WORK( INDS+I ) = S*WORK( INDLPL+I )*L( I )
194          S = WORK( INDS+I ) - LAMBDA
195  50   CONTINUE
196       SAWNAN1 = DISNAN( S )
197       IF( SAWNAN1 ) GOTO 60
198       DO 51 I = R1, R2 - 1
199          DPLUS = D( I ) + S
200          WORK( INDLPL+I ) = LD( I ) / DPLUS
201          WORK( INDS+I ) = S*WORK( INDLPL+I )*L( I )
202          S = WORK( INDS+I ) - LAMBDA
203  51   CONTINUE
204       SAWNAN1 = DISNAN( S )
205 *
206  60   CONTINUE
207       IF( SAWNAN1 ) THEN
208 *        Runs a slower version of the above loop if a NaN is detected
209          NEG1 = 0
210          S = WORK( INDS+B1-1 ) - LAMBDA
211          DO 70 I = B1, R1 - 1
212             DPLUS = D( I ) + S
213             IF(ABS(DPLUS).LT.PIVMIN) DPLUS = -PIVMIN
214             WORK( INDLPL+I ) = LD( I ) / DPLUS
215             IF(DPLUS.LT.ZERO) NEG1 = NEG1 + 1
216             WORK( INDS+I ) = S*WORK( INDLPL+I )*L( I )
217             IF( WORK( INDLPL+I ).EQ.ZERO )
218      $                      WORK( INDS+I ) = LLD( I )
219             S = WORK( INDS+I ) - LAMBDA
220  70      CONTINUE
221          DO 71 I = R1, R2 - 1
222             DPLUS = D( I ) + S
223             IF(ABS(DPLUS).LT.PIVMIN) DPLUS = -PIVMIN
224             WORK( INDLPL+I ) = LD( I ) / DPLUS
225             WORK( INDS+I ) = S*WORK( INDLPL+I )*L( I )
226             IF( WORK( INDLPL+I ).EQ.ZERO )
227      $                      WORK( INDS+I ) = LLD( I )
228             S = WORK( INDS+I ) - LAMBDA
229  71      CONTINUE
230       END IF
231 *
232 *     Compute the progressive transform (using the differential form)
233 *     until the index R1
234 *
235       SAWNAN2 = .FALSE.
236       NEG2 = 0
237       WORK( INDP+BN-1 ) = D( BN ) - LAMBDA
238       DO 80 I = BN - 1, R1, -1
239          DMINUS = LLD( I ) + WORK( INDP+I )
240          TMP = D( I ) / DMINUS
241          IF(DMINUS.LT.ZERO) NEG2 = NEG2 + 1
242          WORK( INDUMN+I ) = L( I )*TMP
243          WORK( INDP+I-1 ) = WORK( INDP+I )*TMP - LAMBDA
244  80   CONTINUE
245       TMP = WORK( INDP+R1-1 )
246       SAWNAN2 = DISNAN( TMP )
247 
248       IF( SAWNAN2 ) THEN
249 *        Runs a slower version of the above loop if a NaN is detected
250          NEG2 = 0
251          DO 100 I = BN-1, R1, -1
252             DMINUS = LLD( I ) + WORK( INDP+I )
253             IF(ABS(DMINUS).LT.PIVMIN) DMINUS = -PIVMIN
254             TMP = D( I ) / DMINUS
255             IF(DMINUS.LT.ZERO) NEG2 = NEG2 + 1
256             WORK( INDUMN+I ) = L( I )*TMP
257             WORK( INDP+I-1 ) = WORK( INDP+I )*TMP - LAMBDA
258             IF( TMP.EQ.ZERO )
259      $          WORK( INDP+I-1 ) = D( I ) - LAMBDA
260  100     CONTINUE
261       END IF
262 *
263 *     Find the index (from R1 to R2) of the largest (in magnitude)
264 *     diagonal element of the inverse
265 *
266       MINGMA = WORK( INDS+R1-1 ) + WORK( INDP+R1-1 )
267       IF( MINGMA.LT.ZERO ) NEG1 = NEG1 + 1
268       IF( WANTNC ) THEN
269          NEGCNT = NEG1 + NEG2
270       ELSE
271          NEGCNT = -1
272       ENDIF
273       IFABS(MINGMA).EQ.ZERO )
274      $   MINGMA = EPS*WORK( INDS+R1-1 )
275       R = R1
276       DO 110 I = R1, R2 - 1
277          TMP = WORK( INDS+I ) + WORK( INDP+I )
278          IF( TMP.EQ.ZERO )
279      $      TMP = EPS*WORK( INDS+I )
280          IFABS( TMP ).LE.ABS( MINGMA ) ) THEN
281             MINGMA = TMP
282             R = I + 1
283          END IF
284  110  CONTINUE
285 *
286 *     Compute the FP vector: solve N^T v = e_r
287 *
288       ISUPPZ( 1 ) = B1
289       ISUPPZ( 2 ) = BN
290       Z( R ) = ONE
291       ZTZ = ONE
292 *
293 *     Compute the FP vector upwards from R
294 *
295       IF.NOT.SAWNAN1 .AND. .NOT.SAWNAN2 ) THEN
296          DO 210 I = R-1, B1, -1
297             Z( I ) = -( WORK( INDLPL+I )*Z( I+1 ) )
298             IF( (ABS(Z(I))+ABS(Z(I+1)))* ABS(LD(I)).LT.GAPTOL )
299      $           THEN
300                Z( I ) = ZERO
301                ISUPPZ( 1 ) = I + 1
302                GOTO 220
303             ENDIF
304             ZTZ = ZTZ + Z( I )*Z( I )
305  210     CONTINUE
306  220     CONTINUE
307       ELSE
308 *        Run slower loop if NaN occurred.
309          DO 230 I = R - 1, B1, -1
310             IF( Z( I+1 ).EQ.ZERO ) THEN
311                Z( I ) = -( LD( I+1 ) / LD( I ) )*Z( I+2 )
312             ELSE
313                Z( I ) = -( WORK( INDLPL+I )*Z( I+1 ) )
314             END IF
315             IF( (ABS(Z(I))+ABS(Z(I+1)))* ABS(LD(I)).LT.GAPTOL )
316      $           THEN
317                Z( I ) = ZERO
318                ISUPPZ( 1 ) = I + 1
319                GO TO 240
320             END IF
321             ZTZ = ZTZ + Z( I )*Z( I )
322  230     CONTINUE
323  240     CONTINUE
324       ENDIF
325 
326 *     Compute the FP vector downwards from R in blocks of size BLKSIZ
327       IF.NOT.SAWNAN1 .AND. .NOT.SAWNAN2 ) THEN
328          DO 250 I = R, BN-1
329             Z( I+1 ) = -( WORK( INDUMN+I )*Z( I ) )
330             IF( (ABS(Z(I))+ABS(Z(I+1)))* ABS(LD(I)).LT.GAPTOL )
331      $         THEN
332                Z( I+1 ) = ZERO
333                ISUPPZ( 2 ) = I
334                GO TO 260
335             END IF
336             ZTZ = ZTZ + Z( I+1 )*Z( I+1 )
337  250     CONTINUE
338  260     CONTINUE
339       ELSE
340 *        Run slower loop if NaN occurred.
341          DO 270 I = R, BN - 1
342             IF( Z( I ).EQ.ZERO ) THEN
343                Z( I+1 ) = -( LD( I-1 ) / LD( I ) )*Z( I-1 )
344             ELSE
345                Z( I+1 ) = -( WORK( INDUMN+I )*Z( I ) )
346             END IF
347             IF( (ABS(Z(I))+ABS(Z(I+1)))* ABS(LD(I)).LT.GAPTOL )
348      $           THEN
349                Z( I+1 ) = ZERO
350                ISUPPZ( 2 ) = I
351                GO TO 280
352             END IF
353             ZTZ = ZTZ + Z( I+1 )*Z( I+1 )
354  270     CONTINUE
355  280     CONTINUE
356       END IF
357 *
358 *     Compute quantities for convergence test
359 *
360       TMP = ONE / ZTZ
361       NRMINV = SQRT( TMP )
362       RESID = ABS( MINGMA )*NRMINV
363       RQCORR = MINGMA*TMP
364 *
365 *
366       RETURN
367 *
368 *     End of DLAR1V
369 *
370       END