1: *> \brief \b DSYRFS
2: *
3: * =========== DOCUMENTATION ===========
4: *
5: * Online html documentation available at
6: * http://www.netlib.org/lapack/explore-html/
7: *
8: *> \htmlonly
9: *> Download DSYRFS + dependencies
10: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/dsyrfs.f">
11: *> [TGZ]</a>
12: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/dsyrfs.f">
13: *> [ZIP]</a>
14: *> <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/dsyrfs.f">
15: *> [TXT]</a>
16: *> \endhtmlonly
17: *
18: * Definition:
19: * ===========
20: *
21: * SUBROUTINE DSYRFS( UPLO, N, NRHS, A, LDA, AF, LDAF, IPIV, B, LDB,
22: * X, LDX, FERR, BERR, WORK, IWORK, INFO )
23: *
24: * .. Scalar Arguments ..
25: * CHARACTER UPLO
26: * INTEGER INFO, LDA, LDAF, LDB, LDX, N, NRHS
27: * ..
28: * .. Array Arguments ..
29: * INTEGER IPIV( * ), IWORK( * )
30: * DOUBLE PRECISION A( LDA, * ), AF( LDAF, * ), B( LDB, * ),
31: * $ BERR( * ), FERR( * ), WORK( * ), X( LDX, * )
32: * ..
33: *
34: *
35: *> \par Purpose:
36: * =============
37: *>
38: *> \verbatim
39: *>
40: *> DSYRFS improves the computed solution to a system of linear
41: *> equations when the coefficient matrix is symmetric indefinite, and
42: *> provides error bounds and backward error estimates for the solution.
43: *> \endverbatim
44: *
45: * Arguments:
46: * ==========
47: *
48: *> \param[in] UPLO
49: *> \verbatim
50: *> UPLO is CHARACTER*1
51: *> = 'U': Upper triangle of A is stored;
52: *> = 'L': Lower triangle of A is stored.
53: *> \endverbatim
54: *>
55: *> \param[in] N
56: *> \verbatim
57: *> N is INTEGER
58: *> The order of the matrix A. N >= 0.
59: *> \endverbatim
60: *>
61: *> \param[in] NRHS
62: *> \verbatim
63: *> NRHS is INTEGER
64: *> The number of right hand sides, i.e., the number of columns
65: *> of the matrices B and X. NRHS >= 0.
66: *> \endverbatim
67: *>
68: *> \param[in] A
69: *> \verbatim
70: *> A is DOUBLE PRECISION array, dimension (LDA,N)
71: *> The symmetric matrix A. If UPLO = 'U', the leading N-by-N
72: *> upper triangular part of A contains the upper triangular part
73: *> of the matrix A, and the strictly lower triangular part of A
74: *> is not referenced. If UPLO = 'L', the leading N-by-N lower
75: *> triangular part of A contains the lower triangular part of
76: *> the matrix A, and the strictly upper triangular part of A is
77: *> not referenced.
78: *> \endverbatim
79: *>
80: *> \param[in] LDA
81: *> \verbatim
82: *> LDA is INTEGER
83: *> The leading dimension of the array A. LDA >= max(1,N).
84: *> \endverbatim
85: *>
86: *> \param[in] AF
87: *> \verbatim
88: *> AF is DOUBLE PRECISION array, dimension (LDAF,N)
89: *> The factored form of the matrix A. AF contains the block
90: *> diagonal matrix D and the multipliers used to obtain the
91: *> factor U or L from the factorization A = U*D*U**T or
92: *> A = L*D*L**T as computed by DSYTRF.
93: *> \endverbatim
94: *>
95: *> \param[in] LDAF
96: *> \verbatim
97: *> LDAF is INTEGER
98: *> The leading dimension of the array AF. LDAF >= max(1,N).
99: *> \endverbatim
100: *>
101: *> \param[in] IPIV
102: *> \verbatim
103: *> IPIV is INTEGER array, dimension (N)
104: *> Details of the interchanges and the block structure of D
105: *> as determined by DSYTRF.
106: *> \endverbatim
107: *>
108: *> \param[in] B
109: *> \verbatim
110: *> B is DOUBLE PRECISION array, dimension (LDB,NRHS)
111: *> The right hand side matrix B.
112: *> \endverbatim
113: *>
114: *> \param[in] LDB
115: *> \verbatim
116: *> LDB is INTEGER
117: *> The leading dimension of the array B. LDB >= max(1,N).
118: *> \endverbatim
119: *>
120: *> \param[in,out] X
121: *> \verbatim
122: *> X is DOUBLE PRECISION array, dimension (LDX,NRHS)
123: *> On entry, the solution matrix X, as computed by DSYTRS.
124: *> On exit, the improved solution matrix X.
125: *> \endverbatim
126: *>
127: *> \param[in] LDX
128: *> \verbatim
129: *> LDX is INTEGER
130: *> The leading dimension of the array X. LDX >= max(1,N).
131: *> \endverbatim
132: *>
133: *> \param[out] FERR
134: *> \verbatim
135: *> FERR is DOUBLE PRECISION array, dimension (NRHS)
136: *> The estimated forward error bound for each solution vector
137: *> X(j) (the j-th column of the solution matrix X).
138: *> If XTRUE is the true solution corresponding to X(j), FERR(j)
139: *> is an estimated upper bound for the magnitude of the largest
140: *> element in (X(j) - XTRUE) divided by the magnitude of the
141: *> largest element in X(j). The estimate is as reliable as
142: *> the estimate for RCOND, and is almost always a slight
143: *> overestimate of the true error.
144: *> \endverbatim
145: *>
146: *> \param[out] BERR
147: *> \verbatim
148: *> BERR is DOUBLE PRECISION array, dimension (NRHS)
149: *> The componentwise relative backward error of each solution
150: *> vector X(j) (i.e., the smallest relative change in
151: *> any element of A or B that makes X(j) an exact solution).
152: *> \endverbatim
153: *>
154: *> \param[out] WORK
155: *> \verbatim
156: *> WORK is DOUBLE PRECISION array, dimension (3*N)
157: *> \endverbatim
158: *>
159: *> \param[out] IWORK
160: *> \verbatim
161: *> IWORK is INTEGER array, dimension (N)
162: *> \endverbatim
163: *>
164: *> \param[out] INFO
165: *> \verbatim
166: *> INFO is INTEGER
167: *> = 0: successful exit
168: *> < 0: if INFO = -i, the i-th argument had an illegal value
169: *> \endverbatim
170: *
171: *> \par Internal Parameters:
172: * =========================
173: *>
174: *> \verbatim
175: *> ITMAX is the maximum number of steps of iterative refinement.
176: *> \endverbatim
177: *
178: * Authors:
179: * ========
180: *
181: *> \author Univ. of Tennessee
182: *> \author Univ. of California Berkeley
183: *> \author Univ. of Colorado Denver
184: *> \author NAG Ltd.
185: *
186: *> \date November 2011
187: *
188: *> \ingroup doubleSYcomputational
189: *
190: * =====================================================================
191: SUBROUTINE DSYRFS( UPLO, N, NRHS, A, LDA, AF, LDAF, IPIV, B, LDB,
192: $ X, LDX, FERR, BERR, WORK, IWORK, INFO )
193: *
194: * -- LAPACK computational routine (version 3.4.0) --
195: * -- LAPACK is a software package provided by Univ. of Tennessee, --
196: * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
197: * November 2011
198: *
199: * .. Scalar Arguments ..
200: CHARACTER UPLO
201: INTEGER INFO, LDA, LDAF, LDB, LDX, N, NRHS
202: * ..
203: * .. Array Arguments ..
204: INTEGER IPIV( * ), IWORK( * )
205: DOUBLE PRECISION A( LDA, * ), AF( LDAF, * ), B( LDB, * ),
206: $ BERR( * ), FERR( * ), WORK( * ), X( LDX, * )
207: * ..
208: *
209: * =====================================================================
210: *
211: * .. Parameters ..
212: INTEGER ITMAX
213: PARAMETER ( ITMAX = 5 )
214: DOUBLE PRECISION ZERO
215: PARAMETER ( ZERO = 0.0D+0 )
216: DOUBLE PRECISION ONE
217: PARAMETER ( ONE = 1.0D+0 )
218: DOUBLE PRECISION TWO
219: PARAMETER ( TWO = 2.0D+0 )
220: DOUBLE PRECISION THREE
221: PARAMETER ( THREE = 3.0D+0 )
222: * ..
223: * .. Local Scalars ..
224: LOGICAL UPPER
225: INTEGER COUNT, I, J, K, KASE, NZ
226: DOUBLE PRECISION EPS, LSTRES, S, SAFE1, SAFE2, SAFMIN, XK
227: * ..
228: * .. Local Arrays ..
229: INTEGER ISAVE( 3 )
230: * ..
231: * .. External Subroutines ..
232: EXTERNAL DAXPY, DCOPY, DLACN2, DSYMV, DSYTRS, XERBLA
233: * ..
234: * .. Intrinsic Functions ..
235: INTRINSIC ABS, MAX
236: * ..
237: * .. External Functions ..
238: LOGICAL LSAME
239: DOUBLE PRECISION DLAMCH
240: EXTERNAL LSAME, DLAMCH
241: * ..
242: * .. Executable Statements ..
243: *
244: * Test the input parameters.
245: *
246: INFO = 0
247: UPPER = LSAME( UPLO, 'U' )
248: IF( .NOT.UPPER .AND. .NOT.LSAME( UPLO, 'L' ) ) THEN
249: INFO = -1
250: ELSE IF( N.LT.0 ) THEN
251: INFO = -2
252: ELSE IF( NRHS.LT.0 ) THEN
253: INFO = -3
254: ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
255: INFO = -5
256: ELSE IF( LDAF.LT.MAX( 1, N ) ) THEN
257: INFO = -7
258: ELSE IF( LDB.LT.MAX( 1, N ) ) THEN
259: INFO = -10
260: ELSE IF( LDX.LT.MAX( 1, N ) ) THEN
261: INFO = -12
262: END IF
263: IF( INFO.NE.0 ) THEN
264: CALL XERBLA( 'DSYRFS', -INFO )
265: RETURN
266: END IF
267: *
268: * Quick return if possible
269: *
270: IF( N.EQ.0 .OR. NRHS.EQ.0 ) THEN
271: DO 10 J = 1, NRHS
272: FERR( J ) = ZERO
273: BERR( J ) = ZERO
274: 10 CONTINUE
275: RETURN
276: END IF
277: *
278: * NZ = maximum number of nonzero elements in each row of A, plus 1
279: *
280: NZ = N + 1
281: EPS = DLAMCH( 'Epsilon' )
282: SAFMIN = DLAMCH( 'Safe minimum' )
283: SAFE1 = NZ*SAFMIN
284: SAFE2 = SAFE1 / EPS
285: *
286: * Do for each right hand side
287: *
288: DO 140 J = 1, NRHS
289: *
290: COUNT = 1
291: LSTRES = THREE
292: 20 CONTINUE
293: *
294: * Loop until stopping criterion is satisfied.
295: *
296: * Compute residual R = B - A * X
297: *
298: CALL DCOPY( N, B( 1, J ), 1, WORK( N+1 ), 1 )
299: CALL DSYMV( UPLO, N, -ONE, A, LDA, X( 1, J ), 1, ONE,
300: $ WORK( N+1 ), 1 )
301: *
302: * Compute componentwise relative backward error from formula
303: *
304: * max(i) ( abs(R(i)) / ( abs(A)*abs(X) + abs(B) )(i) )
305: *
306: * where abs(Z) is the componentwise absolute value of the matrix
307: * or vector Z. If the i-th component of the denominator is less
308: * than SAFE2, then SAFE1 is added to the i-th components of the
309: * numerator and denominator before dividing.
310: *
311: DO 30 I = 1, N
312: WORK( I ) = ABS( B( I, J ) )
313: 30 CONTINUE
314: *
315: * Compute abs(A)*abs(X) + abs(B).
316: *
317: IF( UPPER ) THEN
318: DO 50 K = 1, N
319: S = ZERO
320: XK = ABS( X( K, J ) )
321: DO 40 I = 1, K - 1
322: WORK( I ) = WORK( I ) + ABS( A( I, K ) )*XK
323: S = S + ABS( A( I, K ) )*ABS( X( I, J ) )
324: 40 CONTINUE
325: WORK( K ) = WORK( K ) + ABS( A( K, K ) )*XK + S
326: 50 CONTINUE
327: ELSE
328: DO 70 K = 1, N
329: S = ZERO
330: XK = ABS( X( K, J ) )
331: WORK( K ) = WORK( K ) + ABS( A( K, K ) )*XK
332: DO 60 I = K + 1, N
333: WORK( I ) = WORK( I ) + ABS( A( I, K ) )*XK
334: S = S + ABS( A( I, K ) )*ABS( X( I, J ) )
335: 60 CONTINUE
336: WORK( K ) = WORK( K ) + S
337: 70 CONTINUE
338: END IF
339: S = ZERO
340: DO 80 I = 1, N
341: IF( WORK( I ).GT.SAFE2 ) THEN
342: S = MAX( S, ABS( WORK( N+I ) ) / WORK( I ) )
343: ELSE
344: S = MAX( S, ( ABS( WORK( N+I ) )+SAFE1 ) /
345: $ ( WORK( I )+SAFE1 ) )
346: END IF
347: 80 CONTINUE
348: BERR( J ) = S
349: *
350: * Test stopping criterion. Continue iterating if
351: * 1) The residual BERR(J) is larger than machine epsilon, and
352: * 2) BERR(J) decreased by at least a factor of 2 during the
353: * last iteration, and
354: * 3) At most ITMAX iterations tried.
355: *
356: IF( BERR( J ).GT.EPS .AND. TWO*BERR( J ).LE.LSTRES .AND.
357: $ COUNT.LE.ITMAX ) THEN
358: *
359: * Update solution and try again.
360: *
361: CALL DSYTRS( UPLO, N, 1, AF, LDAF, IPIV, WORK( N+1 ), N,
362: $ INFO )
363: CALL DAXPY( N, ONE, WORK( N+1 ), 1, X( 1, J ), 1 )
364: LSTRES = BERR( J )
365: COUNT = COUNT + 1
366: GO TO 20
367: END IF
368: *
369: * Bound error from formula
370: *
371: * norm(X - XTRUE) / norm(X) .le. FERR =
372: * norm( abs(inv(A))*
373: * ( abs(R) + NZ*EPS*( abs(A)*abs(X)+abs(B) ))) / norm(X)
374: *
375: * where
376: * norm(Z) is the magnitude of the largest component of Z
377: * inv(A) is the inverse of A
378: * abs(Z) is the componentwise absolute value of the matrix or
379: * vector Z
380: * NZ is the maximum number of nonzeros in any row of A, plus 1
381: * EPS is machine epsilon
382: *
383: * The i-th component of abs(R)+NZ*EPS*(abs(A)*abs(X)+abs(B))
384: * is incremented by SAFE1 if the i-th component of
385: * abs(A)*abs(X) + abs(B) is less than SAFE2.
386: *
387: * Use DLACN2 to estimate the infinity-norm of the matrix
388: * inv(A) * diag(W),
389: * where W = abs(R) + NZ*EPS*( abs(A)*abs(X)+abs(B) )))
390: *
391: DO 90 I = 1, N
392: IF( WORK( I ).GT.SAFE2 ) THEN
393: WORK( I ) = ABS( WORK( N+I ) ) + NZ*EPS*WORK( I )
394: ELSE
395: WORK( I ) = ABS( WORK( N+I ) ) + NZ*EPS*WORK( I ) + SAFE1
396: END IF
397: 90 CONTINUE
398: *
399: KASE = 0
400: 100 CONTINUE
401: CALL DLACN2( N, WORK( 2*N+1 ), WORK( N+1 ), IWORK, FERR( J ),
402: $ KASE, ISAVE )
403: IF( KASE.NE.0 ) THEN
404: IF( KASE.EQ.1 ) THEN
405: *
406: * Multiply by diag(W)*inv(A**T).
407: *
408: CALL DSYTRS( UPLO, N, 1, AF, LDAF, IPIV, WORK( N+1 ), N,
409: $ INFO )
410: DO 110 I = 1, N
411: WORK( N+I ) = WORK( I )*WORK( N+I )
412: 110 CONTINUE
413: ELSE IF( KASE.EQ.2 ) THEN
414: *
415: * Multiply by inv(A)*diag(W).
416: *
417: DO 120 I = 1, N
418: WORK( N+I ) = WORK( I )*WORK( N+I )
419: 120 CONTINUE
420: CALL DSYTRS( UPLO, N, 1, AF, LDAF, IPIV, WORK( N+1 ), N,
421: $ INFO )
422: END IF
423: GO TO 100
424: END IF
425: *
426: * Normalize error.
427: *
428: LSTRES = ZERO
429: DO 130 I = 1, N
430: LSTRES = MAX( LSTRES, ABS( X( I, J ) ) )
431: 130 CONTINUE
432: IF( LSTRES.NE.ZERO )
433: $ FERR( J ) = FERR( J ) / LSTRES
434: *
435: 140 CONTINUE
436: *
437: RETURN
438: *
439: * End of DSYRFS
440: *
441: END
CVSweb interface <joel.bertrand@systella.fr>