00001 /* sppt05.f -- translated by f2c (version 20061008). 00002 You must link the resulting object file with libf2c: 00003 on Microsoft Windows system, link with libf2c.lib; 00004 on Linux or Unix systems, link with .../path/to/libf2c.a -lm 00005 or, if you install libf2c.a in a standard place, with -lf2c -lm 00006 -- in that order, at the end of the command line, as in 00007 cc *.o -lf2c -lm 00008 Source for libf2c is in /netlib/f2c/libf2c.zip, e.g., 00009 00010 http://www.netlib.org/f2c/libf2c.zip 00011 */ 00012 00013 #include "f2c.h" 00014 #include "blaswrap.h" 00015 00016 /* Table of constant values */ 00017 00018 static integer c__1 = 1; 00019 00020 /* Subroutine */ int sppt05_(char *uplo, integer *n, integer *nrhs, real *ap, 00021 real *b, integer *ldb, real *x, integer *ldx, real *xact, integer * 00022 ldxact, real *ferr, real *berr, real *reslts) 00023 { 00024 /* System generated locals */ 00025 integer b_dim1, b_offset, x_dim1, x_offset, xact_dim1, xact_offset, i__1, 00026 i__2, i__3; 00027 real r__1, r__2, r__3; 00028 00029 /* Local variables */ 00030 integer i__, j, k, jc; 00031 real eps, tmp, diff, axbi; 00032 integer imax; 00033 real unfl, ovfl; 00034 extern logical lsame_(char *, char *); 00035 logical upper; 00036 real xnorm; 00037 extern doublereal slamch_(char *); 00038 real errbnd; 00039 extern integer isamax_(integer *, real *, integer *); 00040 00041 00042 /* -- LAPACK test routine (version 3.1) -- */ 00043 /* Univ. of Tennessee, Univ. of California Berkeley and NAG Ltd.. */ 00044 /* November 2006 */ 00045 00046 /* .. Scalar Arguments .. */ 00047 /* .. */ 00048 /* .. Array Arguments .. */ 00049 /* .. */ 00050 00051 /* Purpose */ 00052 /* ======= */ 00053 00054 /* SPPT05 tests the error bounds from iterative refinement for the */ 00055 /* computed solution to a system of equations A*X = B, where A is a */ 00056 /* symmetric matrix in packed storage format. */ 00057 00058 /* RESLTS(1) = test of the error bound */ 00059 /* = norm(X - XACT) / ( norm(X) * FERR ) */ 00060 00061 /* A large value is returned if this ratio is not less than one. */ 00062 00063 /* RESLTS(2) = residual from the iterative refinement routine */ 00064 /* = the maximum of BERR / ( (n+1)*EPS + (*) ), where */ 00065 /* (*) = (n+1)*UNFL / (min_i (abs(A)*abs(X) +abs(b))_i ) */ 00066 00067 /* Arguments */ 00068 /* ========= */ 00069 00070 /* UPLO (input) CHARACTER*1 */ 00071 /* Specifies whether the upper or lower triangular part of the */ 00072 /* symmetric matrix A is stored. */ 00073 /* = 'U': Upper triangular */ 00074 /* = 'L': Lower triangular */ 00075 00076 /* N (input) INTEGER */ 00077 /* The number of rows of the matrices X, B, and XACT, and the */ 00078 /* order of the matrix A. N >= 0. */ 00079 00080 /* NRHS (input) INTEGER */ 00081 /* The number of columns of the matrices X, B, and XACT. */ 00082 /* NRHS >= 0. */ 00083 00084 /* AP (input) REAL array, dimension (N*(N+1)/2) */ 00085 /* The upper or lower triangle of the symmetric matrix A, packed */ 00086 /* columnwise in a linear array. The j-th column of A is stored */ 00087 /* in the array AP as follows: */ 00088 /* if UPLO = 'U', AP(i + (j-1)*j/2) = A(i,j) for 1<=i<=j; */ 00089 /* if UPLO = 'L', AP(i + (j-1)*(2n-j)/2) = A(i,j) for j<=i<=n. */ 00090 00091 /* B (input) REAL array, dimension (LDB,NRHS) */ 00092 /* The right hand side vectors for the system of linear */ 00093 /* equations. */ 00094 00095 /* LDB (input) INTEGER */ 00096 /* The leading dimension of the array B. LDB >= max(1,N). */ 00097 00098 /* X (input) REAL array, dimension (LDX,NRHS) */ 00099 /* The computed solution vectors. Each vector is stored as a */ 00100 /* column of the matrix X. */ 00101 00102 /* LDX (input) INTEGER */ 00103 /* The leading dimension of the array X. LDX >= max(1,N). */ 00104 00105 /* XACT (input) REAL array, dimension (LDX,NRHS) */ 00106 /* The exact solution vectors. Each vector is stored as a */ 00107 /* column of the matrix XACT. */ 00108 00109 /* LDXACT (input) INTEGER */ 00110 /* The leading dimension of the array XACT. LDXACT >= max(1,N). */ 00111 00112 /* FERR (input) REAL array, dimension (NRHS) */ 00113 /* The estimated forward error bounds for each solution vector */ 00114 /* X. If XTRUE is the true solution, FERR bounds the magnitude */ 00115 /* of the largest entry in (X - XTRUE) divided by the magnitude */ 00116 /* of the largest entry in X. */ 00117 00118 /* BERR (input) REAL array, dimension (NRHS) */ 00119 /* The componentwise relative backward error of each solution */ 00120 /* vector (i.e., the smallest relative change in any entry of A */ 00121 /* or B that makes X an exact solution). */ 00122 00123 /* RESLTS (output) REAL array, dimension (2) */ 00124 /* The maximum over the NRHS solution vectors of the ratios: */ 00125 /* RESLTS(1) = norm(X - XACT) / ( norm(X) * FERR ) */ 00126 /* RESLTS(2) = BERR / ( (n+1)*EPS + (*) ) */ 00127 00128 /* ===================================================================== */ 00129 00130 /* .. Parameters .. */ 00131 /* .. */ 00132 /* .. Local Scalars .. */ 00133 /* .. */ 00134 /* .. External Functions .. */ 00135 /* .. */ 00136 /* .. Intrinsic Functions .. */ 00137 /* .. */ 00138 /* .. Executable Statements .. */ 00139 00140 /* Quick exit if N = 0 or NRHS = 0. */ 00141 00142 /* Parameter adjustments */ 00143 --ap; 00144 b_dim1 = *ldb; 00145 b_offset = 1 + b_dim1; 00146 b -= b_offset; 00147 x_dim1 = *ldx; 00148 x_offset = 1 + x_dim1; 00149 x -= x_offset; 00150 xact_dim1 = *ldxact; 00151 xact_offset = 1 + xact_dim1; 00152 xact -= xact_offset; 00153 --ferr; 00154 --berr; 00155 --reslts; 00156 00157 /* Function Body */ 00158 if (*n <= 0 || *nrhs <= 0) { 00159 reslts[1] = 0.f; 00160 reslts[2] = 0.f; 00161 return 0; 00162 } 00163 00164 eps = slamch_("Epsilon"); 00165 unfl = slamch_("Safe minimum"); 00166 ovfl = 1.f / unfl; 00167 upper = lsame_(uplo, "U"); 00168 00169 /* Test 1: Compute the maximum of */ 00170 /* norm(X - XACT) / ( norm(X) * FERR ) */ 00171 /* over all the vectors X and XACT using the infinity-norm. */ 00172 00173 errbnd = 0.f; 00174 i__1 = *nrhs; 00175 for (j = 1; j <= i__1; ++j) { 00176 imax = isamax_(n, &x[j * x_dim1 + 1], &c__1); 00177 /* Computing MAX */ 00178 r__2 = (r__1 = x[imax + j * x_dim1], dabs(r__1)); 00179 xnorm = dmax(r__2,unfl); 00180 diff = 0.f; 00181 i__2 = *n; 00182 for (i__ = 1; i__ <= i__2; ++i__) { 00183 /* Computing MAX */ 00184 r__2 = diff, r__3 = (r__1 = x[i__ + j * x_dim1] - xact[i__ + j * 00185 xact_dim1], dabs(r__1)); 00186 diff = dmax(r__2,r__3); 00187 /* L10: */ 00188 } 00189 00190 if (xnorm > 1.f) { 00191 goto L20; 00192 } else if (diff <= ovfl * xnorm) { 00193 goto L20; 00194 } else { 00195 errbnd = 1.f / eps; 00196 goto L30; 00197 } 00198 00199 L20: 00200 if (diff / xnorm <= ferr[j]) { 00201 /* Computing MAX */ 00202 r__1 = errbnd, r__2 = diff / xnorm / ferr[j]; 00203 errbnd = dmax(r__1,r__2); 00204 } else { 00205 errbnd = 1.f / eps; 00206 } 00207 L30: 00208 ; 00209 } 00210 reslts[1] = errbnd; 00211 00212 /* Test 2: Compute the maximum of BERR / ( (n+1)*EPS + (*) ), where */ 00213 /* (*) = (n+1)*UNFL / (min_i (abs(A)*abs(X) +abs(b))_i ) */ 00214 00215 i__1 = *nrhs; 00216 for (k = 1; k <= i__1; ++k) { 00217 i__2 = *n; 00218 for (i__ = 1; i__ <= i__2; ++i__) { 00219 tmp = (r__1 = b[i__ + k * b_dim1], dabs(r__1)); 00220 if (upper) { 00221 jc = (i__ - 1) * i__ / 2; 00222 i__3 = i__; 00223 for (j = 1; j <= i__3; ++j) { 00224 tmp += (r__1 = ap[jc + j], dabs(r__1)) * (r__2 = x[j + k * 00225 x_dim1], dabs(r__2)); 00226 /* L40: */ 00227 } 00228 jc += i__; 00229 i__3 = *n; 00230 for (j = i__ + 1; j <= i__3; ++j) { 00231 tmp += (r__1 = ap[jc], dabs(r__1)) * (r__2 = x[j + k * 00232 x_dim1], dabs(r__2)); 00233 jc += j; 00234 /* L50: */ 00235 } 00236 } else { 00237 jc = i__; 00238 i__3 = i__ - 1; 00239 for (j = 1; j <= i__3; ++j) { 00240 tmp += (r__1 = ap[jc], dabs(r__1)) * (r__2 = x[j + k * 00241 x_dim1], dabs(r__2)); 00242 jc = jc + *n - j; 00243 /* L60: */ 00244 } 00245 i__3 = *n; 00246 for (j = i__; j <= i__3; ++j) { 00247 tmp += (r__1 = ap[jc + j - i__], dabs(r__1)) * (r__2 = x[ 00248 j + k * x_dim1], dabs(r__2)); 00249 /* L70: */ 00250 } 00251 } 00252 if (i__ == 1) { 00253 axbi = tmp; 00254 } else { 00255 axbi = dmin(axbi,tmp); 00256 } 00257 /* L80: */ 00258 } 00259 /* Computing MAX */ 00260 r__1 = axbi, r__2 = (*n + 1) * unfl; 00261 tmp = berr[k] / ((*n + 1) * eps + (*n + 1) * unfl / dmax(r__1,r__2)); 00262 if (k == 1) { 00263 reslts[2] = tmp; 00264 } else { 00265 reslts[2] = dmax(reslts[2],tmp); 00266 } 00267 /* L90: */ 00268 } 00269 00270 return 0; 00271 00272 /* End of SPPT05 */ 00273 00274 } /* sppt05_ */