1: SUBROUTINE DPOSVX( FACT, UPLO, N, NRHS, A, LDA, AF, LDAF, EQUED,
2: $ S, B, LDB, X, LDX, RCOND, FERR, BERR, WORK,
3: $ IWORK, INFO )
4: *
5: * -- LAPACK driver routine (version 3.2) --
6: * -- LAPACK is a software package provided by Univ. of Tennessee, --
7: * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
8: * November 2006
9: *
10: * .. Scalar Arguments ..
11: CHARACTER EQUED, FACT, UPLO
12: INTEGER INFO, LDA, LDAF, LDB, LDX, N, NRHS
13: DOUBLE PRECISION RCOND
14: * ..
15: * .. Array Arguments ..
16: INTEGER IWORK( * )
17: DOUBLE PRECISION A( LDA, * ), AF( LDAF, * ), B( LDB, * ),
18: $ BERR( * ), FERR( * ), S( * ), WORK( * ),
19: $ X( LDX, * )
20: * ..
21: *
22: * Purpose
23: * =======
24: *
25: * DPOSVX uses the Cholesky factorization A = U**T*U or A = L*L**T to
26: * compute the solution to a real system of linear equations
27: * A * X = B,
28: * where A is an N-by-N symmetric positive definite matrix and X and B
29: * are N-by-NRHS matrices.
30: *
31: * Error bounds on the solution and a condition estimate are also
32: * provided.
33: *
34: * Description
35: * ===========
36: *
37: * The following steps are performed:
38: *
39: * 1. If FACT = 'E', real scaling factors are computed to equilibrate
40: * the system:
41: * diag(S) * A * diag(S) * inv(diag(S)) * X = diag(S) * B
42: * Whether or not the system will be equilibrated depends on the
43: * scaling of the matrix A, but if equilibration is used, A is
44: * overwritten by diag(S)*A*diag(S) and B by diag(S)*B.
45: *
46: * 2. If FACT = 'N' or 'E', the Cholesky decomposition is used to
47: * factor the matrix A (after equilibration if FACT = 'E') as
48: * A = U**T* U, if UPLO = 'U', or
49: * A = L * L**T, if UPLO = 'L',
50: * where U is an upper triangular matrix and L is a lower triangular
51: * matrix.
52: *
53: * 3. If the leading i-by-i principal minor is not positive definite,
54: * then the routine returns with INFO = i. Otherwise, the factored
55: * form of A is used to estimate the condition number of the matrix
56: * A. If the reciprocal of the condition number is less than machine
57: * precision, INFO = N+1 is returned as a warning, but the routine
58: * still goes on to solve for X and compute error bounds as
59: * described below.
60: *
61: * 4. The system of equations is solved for X using the factored form
62: * of A.
63: *
64: * 5. Iterative refinement is applied to improve the computed solution
65: * matrix and calculate error bounds and backward error estimates
66: * for it.
67: *
68: * 6. If equilibration was used, the matrix X is premultiplied by
69: * diag(S) so that it solves the original system before
70: * equilibration.
71: *
72: * Arguments
73: * =========
74: *
75: * FACT (input) CHARACTER*1
76: * Specifies whether or not the factored form of the matrix A is
77: * supplied on entry, and if not, whether the matrix A should be
78: * equilibrated before it is factored.
79: * = 'F': On entry, AF contains the factored form of A.
80: * If EQUED = 'Y', the matrix A has been equilibrated
81: * with scaling factors given by S. A and AF will not
82: * be modified.
83: * = 'N': The matrix A will be copied to AF and factored.
84: * = 'E': The matrix A will be equilibrated if necessary, then
85: * copied to AF and factored.
86: *
87: * UPLO (input) CHARACTER*1
88: * = 'U': Upper triangle of A is stored;
89: * = 'L': Lower triangle of A is stored.
90: *
91: * N (input) INTEGER
92: * The number of linear equations, i.e., the order of the
93: * matrix A. N >= 0.
94: *
95: * NRHS (input) INTEGER
96: * The number of right hand sides, i.e., the number of columns
97: * of the matrices B and X. NRHS >= 0.
98: *
99: * A (input/output) DOUBLE PRECISION array, dimension (LDA,N)
100: * On entry, the symmetric matrix A, except if FACT = 'F' and
101: * EQUED = 'Y', then A must contain the equilibrated matrix
102: * diag(S)*A*diag(S). If UPLO = 'U', the leading
103: * N-by-N upper triangular part of A contains the upper
104: * triangular part of the matrix A, and the strictly lower
105: * triangular part of A is not referenced. If UPLO = 'L', the
106: * leading N-by-N lower triangular part of A contains the lower
107: * triangular part of the matrix A, and the strictly upper
108: * triangular part of A is not referenced. A is not modified if
109: * FACT = 'F' or 'N', or if FACT = 'E' and EQUED = 'N' on exit.
110: *
111: * On exit, if FACT = 'E' and EQUED = 'Y', A is overwritten by
112: * diag(S)*A*diag(S).
113: *
114: * LDA (input) INTEGER
115: * The leading dimension of the array A. LDA >= max(1,N).
116: *
117: * AF (input or output) DOUBLE PRECISION array, dimension (LDAF,N)
118: * If FACT = 'F', then AF is an input argument and on entry
119: * contains the triangular factor U or L from the Cholesky
120: * factorization A = U**T*U or A = L*L**T, in the same storage
121: * format as A. If EQUED .ne. 'N', then AF is the factored form
122: * of the equilibrated matrix diag(S)*A*diag(S).
123: *
124: * If FACT = 'N', then AF is an output argument and on exit
125: * returns the triangular factor U or L from the Cholesky
126: * factorization A = U**T*U or A = L*L**T of the original
127: * matrix A.
128: *
129: * If FACT = 'E', then AF is an output argument and on exit
130: * returns the triangular factor U or L from the Cholesky
131: * factorization A = U**T*U or A = L*L**T of the equilibrated
132: * matrix A (see the description of A for the form of the
133: * equilibrated matrix).
134: *
135: * LDAF (input) INTEGER
136: * The leading dimension of the array AF. LDAF >= max(1,N).
137: *
138: * EQUED (input or output) CHARACTER*1
139: * Specifies the form of equilibration that was done.
140: * = 'N': No equilibration (always true if FACT = 'N').
141: * = 'Y': Equilibration was done, i.e., A has been replaced by
142: * diag(S) * A * diag(S).
143: * EQUED is an input argument if FACT = 'F'; otherwise, it is an
144: * output argument.
145: *
146: * S (input or output) DOUBLE PRECISION array, dimension (N)
147: * The scale factors for A; not accessed if EQUED = 'N'. S is
148: * an input argument if FACT = 'F'; otherwise, S is an output
149: * argument. If FACT = 'F' and EQUED = 'Y', each element of S
150: * must be positive.
151: *
152: * B (input/output) DOUBLE PRECISION array, dimension (LDB,NRHS)
153: * On entry, the N-by-NRHS right hand side matrix B.
154: * On exit, if EQUED = 'N', B is not modified; if EQUED = 'Y',
155: * B is overwritten by diag(S) * B.
156: *
157: * LDB (input) INTEGER
158: * The leading dimension of the array B. LDB >= max(1,N).
159: *
160: * X (output) DOUBLE PRECISION array, dimension (LDX,NRHS)
161: * If INFO = 0 or INFO = N+1, the N-by-NRHS solution matrix X to
162: * the original system of equations. Note that if EQUED = 'Y',
163: * A and B are modified on exit, and the solution to the
164: * equilibrated system is inv(diag(S))*X.
165: *
166: * LDX (input) INTEGER
167: * The leading dimension of the array X. LDX >= max(1,N).
168: *
169: * RCOND (output) DOUBLE PRECISION
170: * The estimate of the reciprocal condition number of the matrix
171: * A after equilibration (if done). If RCOND is less than the
172: * machine precision (in particular, if RCOND = 0), the matrix
173: * is singular to working precision. This condition is
174: * indicated by a return code of INFO > 0.
175: *
176: * FERR (output) DOUBLE PRECISION array, dimension (NRHS)
177: * The estimated forward error bound for each solution vector
178: * X(j) (the j-th column of the solution matrix X).
179: * If XTRUE is the true solution corresponding to X(j), FERR(j)
180: * is an estimated upper bound for the magnitude of the largest
181: * element in (X(j) - XTRUE) divided by the magnitude of the
182: * largest element in X(j). The estimate is as reliable as
183: * the estimate for RCOND, and is almost always a slight
184: * overestimate of the true error.
185: *
186: * BERR (output) DOUBLE PRECISION array, dimension (NRHS)
187: * The componentwise relative backward error of each solution
188: * vector X(j) (i.e., the smallest relative change in
189: * any element of A or B that makes X(j) an exact solution).
190: *
191: * WORK (workspace) DOUBLE PRECISION array, dimension (3*N)
192: *
193: * IWORK (workspace) INTEGER array, dimension (N)
194: *
195: * INFO (output) INTEGER
196: * = 0: successful exit
197: * < 0: if INFO = -i, the i-th argument had an illegal value
198: * > 0: if INFO = i, and i is
199: * <= N: the leading minor of order i of A is
200: * not positive definite, so the factorization
201: * could not be completed, and the solution has not
202: * been computed. RCOND = 0 is returned.
203: * = N+1: U is nonsingular, but RCOND is less than machine
204: * precision, meaning that the matrix is singular
205: * to working precision. Nevertheless, the
206: * solution and error bounds are computed because
207: * there are a number of situations where the
208: * computed solution can be more accurate than the
209: * value of RCOND would suggest.
210: *
211: * =====================================================================
212: *
213: * .. Parameters ..
214: DOUBLE PRECISION ZERO, ONE
215: PARAMETER ( ZERO = 0.0D+0, ONE = 1.0D+0 )
216: * ..
217: * .. Local Scalars ..
218: LOGICAL EQUIL, NOFACT, RCEQU
219: INTEGER I, INFEQU, J
220: DOUBLE PRECISION AMAX, ANORM, BIGNUM, SCOND, SMAX, SMIN, SMLNUM
221: * ..
222: * .. External Functions ..
223: LOGICAL LSAME
224: DOUBLE PRECISION DLAMCH, DLANSY
225: EXTERNAL LSAME, DLAMCH, DLANSY
226: * ..
227: * .. External Subroutines ..
228: EXTERNAL DLACPY, DLAQSY, DPOCON, DPOEQU, DPORFS, DPOTRF,
229: $ DPOTRS, XERBLA
230: * ..
231: * .. Intrinsic Functions ..
232: INTRINSIC MAX, MIN
233: * ..
234: * .. Executable Statements ..
235: *
236: INFO = 0
237: NOFACT = LSAME( FACT, 'N' )
238: EQUIL = LSAME( FACT, 'E' )
239: IF( NOFACT .OR. EQUIL ) THEN
240: EQUED = 'N'
241: RCEQU = .FALSE.
242: ELSE
243: RCEQU = LSAME( EQUED, 'Y' )
244: SMLNUM = DLAMCH( 'Safe minimum' )
245: BIGNUM = ONE / SMLNUM
246: END IF
247: *
248: * Test the input parameters.
249: *
250: IF( .NOT.NOFACT .AND. .NOT.EQUIL .AND. .NOT.LSAME( FACT, 'F' ) )
251: $ THEN
252: INFO = -1
253: ELSE IF( .NOT.LSAME( UPLO, 'U' ) .AND. .NOT.LSAME( UPLO, 'L' ) )
254: $ THEN
255: INFO = -2
256: ELSE IF( N.LT.0 ) THEN
257: INFO = -3
258: ELSE IF( NRHS.LT.0 ) THEN
259: INFO = -4
260: ELSE IF( LDA.LT.MAX( 1, N ) ) THEN
261: INFO = -6
262: ELSE IF( LDAF.LT.MAX( 1, N ) ) THEN
263: INFO = -8
264: ELSE IF( LSAME( FACT, 'F' ) .AND. .NOT.
265: $ ( RCEQU .OR. LSAME( EQUED, 'N' ) ) ) THEN
266: INFO = -9
267: ELSE
268: IF( RCEQU ) THEN
269: SMIN = BIGNUM
270: SMAX = ZERO
271: DO 10 J = 1, N
272: SMIN = MIN( SMIN, S( J ) )
273: SMAX = MAX( SMAX, S( J ) )
274: 10 CONTINUE
275: IF( SMIN.LE.ZERO ) THEN
276: INFO = -10
277: ELSE IF( N.GT.0 ) THEN
278: SCOND = MAX( SMIN, SMLNUM ) / MIN( SMAX, BIGNUM )
279: ELSE
280: SCOND = ONE
281: END IF
282: END IF
283: IF( INFO.EQ.0 ) THEN
284: IF( LDB.LT.MAX( 1, N ) ) THEN
285: INFO = -12
286: ELSE IF( LDX.LT.MAX( 1, N ) ) THEN
287: INFO = -14
288: END IF
289: END IF
290: END IF
291: *
292: IF( INFO.NE.0 ) THEN
293: CALL XERBLA( 'DPOSVX', -INFO )
294: RETURN
295: END IF
296: *
297: IF( EQUIL ) THEN
298: *
299: * Compute row and column scalings to equilibrate the matrix A.
300: *
301: CALL DPOEQU( N, A, LDA, S, SCOND, AMAX, INFEQU )
302: IF( INFEQU.EQ.0 ) THEN
303: *
304: * Equilibrate the matrix.
305: *
306: CALL DLAQSY( UPLO, N, A, LDA, S, SCOND, AMAX, EQUED )
307: RCEQU = LSAME( EQUED, 'Y' )
308: END IF
309: END IF
310: *
311: * Scale the right hand side.
312: *
313: IF( RCEQU ) THEN
314: DO 30 J = 1, NRHS
315: DO 20 I = 1, N
316: B( I, J ) = S( I )*B( I, J )
317: 20 CONTINUE
318: 30 CONTINUE
319: END IF
320: *
321: IF( NOFACT .OR. EQUIL ) THEN
322: *
323: * Compute the Cholesky factorization A = U'*U or A = L*L'.
324: *
325: CALL DLACPY( UPLO, N, N, A, LDA, AF, LDAF )
326: CALL DPOTRF( UPLO, N, AF, LDAF, INFO )
327: *
328: * Return if INFO is non-zero.
329: *
330: IF( INFO.GT.0 )THEN
331: RCOND = ZERO
332: RETURN
333: END IF
334: END IF
335: *
336: * Compute the norm of the matrix A.
337: *
338: ANORM = DLANSY( '1', UPLO, N, A, LDA, WORK )
339: *
340: * Compute the reciprocal of the condition number of A.
341: *
342: CALL DPOCON( UPLO, N, AF, LDAF, ANORM, RCOND, WORK, IWORK, INFO )
343: *
344: * Compute the solution matrix X.
345: *
346: CALL DLACPY( 'Full', N, NRHS, B, LDB, X, LDX )
347: CALL DPOTRS( UPLO, N, NRHS, AF, LDAF, X, LDX, INFO )
348: *
349: * Use iterative refinement to improve the computed solution and
350: * compute error bounds and backward error estimates for it.
351: *
352: CALL DPORFS( UPLO, N, NRHS, A, LDA, AF, LDAF, B, LDB, X, LDX,
353: $ FERR, BERR, WORK, IWORK, INFO )
354: *
355: * Transform the solution matrix X to a solution of the original
356: * system.
357: *
358: IF( RCEQU ) THEN
359: DO 50 J = 1, NRHS
360: DO 40 I = 1, N
361: X( I, J ) = S( I )*X( I, J )
362: 40 CONTINUE
363: 50 CONTINUE
364: DO 60 J = 1, NRHS
365: FERR( J ) = FERR( J ) / SCOND
366: 60 CONTINUE
367: END IF
368: *
369: * Set INFO = N+1 if the matrix is singular to working precision.
370: *
371: IF( RCOND.LT.DLAMCH( 'Epsilon' ) )
372: $ INFO = N + 1
373: *
374: RETURN
375: *
376: * End of DPOSVX
377: *
378: END
CVSweb interface <joel.bertrand@systella.fr>