Annotation of rpl/lapack/lapack/dptsvx.f, revision 1.1.1.1
1.1 bertrand 1: SUBROUTINE DPTSVX( FACT, N, NRHS, D, E, DF, EF, B, LDB, X, LDX,
2: $ RCOND, FERR, BERR, WORK, INFO )
3: *
4: * -- LAPACK routine (version 3.2) --
5: * -- LAPACK is a software package provided by Univ. of Tennessee, --
6: * -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
7: * November 2006
8: *
9: * .. Scalar Arguments ..
10: CHARACTER FACT
11: INTEGER INFO, LDB, LDX, N, NRHS
12: DOUBLE PRECISION RCOND
13: * ..
14: * .. Array Arguments ..
15: DOUBLE PRECISION B( LDB, * ), BERR( * ), D( * ), DF( * ),
16: $ E( * ), EF( * ), FERR( * ), WORK( * ),
17: $ X( LDX, * )
18: * ..
19: *
20: * Purpose
21: * =======
22: *
23: * DPTSVX uses the factorization A = L*D*L**T to compute the solution
24: * to a real system of linear equations A*X = B, where A is an N-by-N
25: * symmetric positive definite tridiagonal matrix and X and B are
26: * N-by-NRHS matrices.
27: *
28: * Error bounds on the solution and a condition estimate are also
29: * provided.
30: *
31: * Description
32: * ===========
33: *
34: * The following steps are performed:
35: *
36: * 1. If FACT = 'N', the matrix A is factored as A = L*D*L**T, where L
37: * is a unit lower bidiagonal matrix and D is diagonal. The
38: * factorization can also be regarded as having the form
39: * A = U**T*D*U.
40: *
41: * 2. If the leading i-by-i principal minor is not positive definite,
42: * then the routine returns with INFO = i. Otherwise, the factored
43: * form of A is used to estimate the condition number of the matrix
44: * A. If the reciprocal of the condition number is less than machine
45: * precision, INFO = N+1 is returned as a warning, but the routine
46: * still goes on to solve for X and compute error bounds as
47: * described below.
48: *
49: * 3. The system of equations is solved for X using the factored form
50: * of A.
51: *
52: * 4. Iterative refinement is applied to improve the computed solution
53: * matrix and calculate error bounds and backward error estimates
54: * for it.
55: *
56: * Arguments
57: * =========
58: *
59: * FACT (input) CHARACTER*1
60: * Specifies whether or not the factored form of A has been
61: * supplied on entry.
62: * = 'F': On entry, DF and EF contain the factored form of A.
63: * D, E, DF, and EF will not be modified.
64: * = 'N': The matrix A will be copied to DF and EF and
65: * factored.
66: *
67: * N (input) INTEGER
68: * The order of the matrix A. N >= 0.
69: *
70: * NRHS (input) INTEGER
71: * The number of right hand sides, i.e., the number of columns
72: * of the matrices B and X. NRHS >= 0.
73: *
74: * D (input) DOUBLE PRECISION array, dimension (N)
75: * The n diagonal elements of the tridiagonal matrix A.
76: *
77: * E (input) DOUBLE PRECISION array, dimension (N-1)
78: * The (n-1) subdiagonal elements of the tridiagonal matrix A.
79: *
80: * DF (input or output) DOUBLE PRECISION array, dimension (N)
81: * If FACT = 'F', then DF is an input argument and on entry
82: * contains the n diagonal elements of the diagonal matrix D
83: * from the L*D*L**T factorization of A.
84: * If FACT = 'N', then DF is an output argument and on exit
85: * contains the n diagonal elements of the diagonal matrix D
86: * from the L*D*L**T factorization of A.
87: *
88: * EF (input or output) DOUBLE PRECISION array, dimension (N-1)
89: * If FACT = 'F', then EF is an input argument and on entry
90: * contains the (n-1) subdiagonal elements of the unit
91: * bidiagonal factor L from the L*D*L**T factorization of A.
92: * If FACT = 'N', then EF is an output argument and on exit
93: * contains the (n-1) subdiagonal elements of the unit
94: * bidiagonal factor L from the L*D*L**T factorization of A.
95: *
96: * B (input) DOUBLE PRECISION array, dimension (LDB,NRHS)
97: * The N-by-NRHS right hand side matrix B.
98: *
99: * LDB (input) INTEGER
100: * The leading dimension of the array B. LDB >= max(1,N).
101: *
102: * X (output) DOUBLE PRECISION array, dimension (LDX,NRHS)
103: * If INFO = 0 of INFO = N+1, the N-by-NRHS solution matrix X.
104: *
105: * LDX (input) INTEGER
106: * The leading dimension of the array X. LDX >= max(1,N).
107: *
108: * RCOND (output) DOUBLE PRECISION
109: * The reciprocal condition number of the matrix A. If RCOND
110: * is less than the machine precision (in particular, if
111: * RCOND = 0), the matrix is singular to working precision.
112: * This condition is indicated by a return code of INFO > 0.
113: *
114: * FERR (output) DOUBLE PRECISION array, dimension (NRHS)
115: * The forward error bound for each solution vector
116: * X(j) (the j-th column of the solution matrix X).
117: * If XTRUE is the true solution corresponding to X(j), FERR(j)
118: * is an estimated upper bound for the magnitude of the largest
119: * element in (X(j) - XTRUE) divided by the magnitude of the
120: * largest element in X(j).
121: *
122: * BERR (output) DOUBLE PRECISION array, dimension (NRHS)
123: * The componentwise relative backward error of each solution
124: * vector X(j) (i.e., the smallest relative change in any
125: * element of A or B that makes X(j) an exact solution).
126: *
127: * WORK (workspace) DOUBLE PRECISION array, dimension (2*N)
128: *
129: * INFO (output) INTEGER
130: * = 0: successful exit
131: * < 0: if INFO = -i, the i-th argument had an illegal value
132: * > 0: if INFO = i, and i is
133: * <= N: the leading minor of order i of A is
134: * not positive definite, so the factorization
135: * could not be completed, and the solution has not
136: * been computed. RCOND = 0 is returned.
137: * = N+1: U is nonsingular, but RCOND is less than machine
138: * precision, meaning that the matrix is singular
139: * to working precision. Nevertheless, the
140: * solution and error bounds are computed because
141: * there are a number of situations where the
142: * computed solution can be more accurate than the
143: * value of RCOND would suggest.
144: *
145: * =====================================================================
146: *
147: * .. Parameters ..
148: DOUBLE PRECISION ZERO
149: PARAMETER ( ZERO = 0.0D+0 )
150: * ..
151: * .. Local Scalars ..
152: LOGICAL NOFACT
153: DOUBLE PRECISION ANORM
154: * ..
155: * .. External Functions ..
156: LOGICAL LSAME
157: DOUBLE PRECISION DLAMCH, DLANST
158: EXTERNAL LSAME, DLAMCH, DLANST
159: * ..
160: * .. External Subroutines ..
161: EXTERNAL DCOPY, DLACPY, DPTCON, DPTRFS, DPTTRF, DPTTRS,
162: $ XERBLA
163: * ..
164: * .. Intrinsic Functions ..
165: INTRINSIC MAX
166: * ..
167: * .. Executable Statements ..
168: *
169: * Test the input parameters.
170: *
171: INFO = 0
172: NOFACT = LSAME( FACT, 'N' )
173: IF( .NOT.NOFACT .AND. .NOT.LSAME( FACT, 'F' ) ) THEN
174: INFO = -1
175: ELSE IF( N.LT.0 ) THEN
176: INFO = -2
177: ELSE IF( NRHS.LT.0 ) THEN
178: INFO = -3
179: ELSE IF( LDB.LT.MAX( 1, N ) ) THEN
180: INFO = -9
181: ELSE IF( LDX.LT.MAX( 1, N ) ) THEN
182: INFO = -11
183: END IF
184: IF( INFO.NE.0 ) THEN
185: CALL XERBLA( 'DPTSVX', -INFO )
186: RETURN
187: END IF
188: *
189: IF( NOFACT ) THEN
190: *
191: * Compute the L*D*L' (or U'*D*U) factorization of A.
192: *
193: CALL DCOPY( N, D, 1, DF, 1 )
194: IF( N.GT.1 )
195: $ CALL DCOPY( N-1, E, 1, EF, 1 )
196: CALL DPTTRF( N, DF, EF, INFO )
197: *
198: * Return if INFO is non-zero.
199: *
200: IF( INFO.GT.0 )THEN
201: RCOND = ZERO
202: RETURN
203: END IF
204: END IF
205: *
206: * Compute the norm of the matrix A.
207: *
208: ANORM = DLANST( '1', N, D, E )
209: *
210: * Compute the reciprocal of the condition number of A.
211: *
212: CALL DPTCON( N, DF, EF, ANORM, RCOND, WORK, INFO )
213: *
214: * Compute the solution vectors X.
215: *
216: CALL DLACPY( 'Full', N, NRHS, B, LDB, X, LDX )
217: CALL DPTTRS( N, NRHS, DF, EF, X, LDX, INFO )
218: *
219: * Use iterative refinement to improve the computed solutions and
220: * compute error bounds and backward error estimates for them.
221: *
222: CALL DPTRFS( N, NRHS, D, E, DF, EF, B, LDB, X, LDX, FERR, BERR,
223: $ WORK, INFO )
224: *
225: * Set INFO = N+1 if the matrix is singular to working precision.
226: *
227: IF( RCOND.LT.DLAMCH( 'Epsilon' ) )
228: $ INFO = N + 1
229: *
230: RETURN
231: *
232: * End of DPTSVX
233: *
234: END
CVSweb interface <joel.bertrand@systella.fr>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to dptsvx.f CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [local] / rpl / lapack / lapack