1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
|
/* Ergo, version 3.8, a program for linear scaling electronic structure
* calculations.
* Copyright (C) 2019 Elias Rudberg, Emanuel H. Rubensson, Pawel Salek,
* and Anastasia Kruchinina.
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*
* Primary academic reference:
* Ergo: An open-source program for linear-scaling electronic structure
* calculations,
* Elias Rudberg, Emanuel H. Rubensson, Pawel Salek, and Anastasia
* Kruchinina,
* SoftwareX 7, 107 (2018),
* <http://dx.doi.org/10.1016/j.softx.2018.03.005>
*
* For further information about Ergo, see <http://www.ergoscf.org>.
*/
/* This file belongs to the template_lapack part of the Ergo source
* code. The source files in the template_lapack directory are modified
* versions of files originally distributed as CLAPACK, see the
* Copyright/license notice in the file template_lapack/COPYING.
*/
#ifndef TEMPLATE_LAPACK_STERF_HEADER
#define TEMPLATE_LAPACK_STERF_HEADER
#include "template_lapack_common.h"
template<class Treal>
int template_lapack_sterf(const integer *n, Treal *d__, Treal *e,
integer *info)
{
/* -- LAPACK routine (version 3.0) --
Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
Courant Institute, Argonne National Lab, and Rice University
June 30, 1999
Purpose
=======
DSTERF computes all eigenvalues of a symmetric tridiagonal matrix
using the Pal-Walker-Kahan variant of the QL or QR algorithm.
Arguments
=========
N (input) INTEGER
The order of the matrix. N >= 0.
D (input/output) DOUBLE PRECISION array, dimension (N)
On entry, the n diagonal elements of the tridiagonal matrix.
On exit, if INFO = 0, the eigenvalues in ascending order.
E (input/output) DOUBLE PRECISION array, dimension (N-1)
On entry, the (n-1) subdiagonal elements of the tridiagonal
matrix.
On exit, E has been destroyed.
INFO (output) INTEGER
= 0: successful exit
< 0: if INFO = -i, the i-th argument had an illegal value
> 0: the algorithm failed to find all of the eigenvalues in
a total of 30*N iterations; if INFO = i, then i
elements of E have not converged to zero.
=====================================================================
Test the input parameters.
Parameter adjustments */
/* Table of constant values */
integer c__0 = 0;
integer c__1 = 1;
Treal c_b32 = 1.;
/* System generated locals */
integer i__1;
Treal d__1, d__2, d__3;
/* Local variables */
Treal oldc;
integer lend, jtot;
Treal c__;
integer i__, l, m;
Treal p, gamma, r__, s, alpha, sigma, anorm;
integer l1;
Treal bb;
integer iscale;
Treal oldgam, safmin;
Treal safmax;
integer lendsv;
Treal ssfmin;
integer nmaxit;
Treal ssfmax, rt1, rt2, eps, rte;
integer lsv;
Treal eps2;
--e;
--d__;
/* Function Body */
*info = 0;
/* Quick return if possible */
if (*n < 0) {
*info = -1;
i__1 = -(*info);
template_blas_erbla("STERF ", &i__1);
return 0;
}
if (*n <= 1) {
return 0;
}
/* Determine the unit roundoff for this environment. */
eps = template_lapack_lamch("E", (Treal)0);
/* Computing 2nd power */
d__1 = eps;
eps2 = d__1 * d__1;
safmin = template_lapack_lamch("S", (Treal)0);
safmax = 1. / safmin;
ssfmax = template_blas_sqrt(safmax) / 3.;
ssfmin = template_blas_sqrt(safmin) / eps2;
/* Compute the eigenvalues of the tridiagonal matrix. */
nmaxit = *n * 30;
sigma = 0.;
jtot = 0;
/* Determine where the matrix splits and choose QL or QR iteration
for each block, according to whether top or bottom diagonal
element is smaller. */
l1 = 1;
L10:
if (l1 > *n) {
goto L170;
}
if (l1 > 1) {
e[l1 - 1] = 0.;
}
i__1 = *n - 1;
for (m = l1; m <= i__1; ++m) {
if ((d__3 = e[m], absMACRO(d__3)) <= template_blas_sqrt((d__1 = d__[m], absMACRO(d__1))) *
template_blas_sqrt((d__2 = d__[m + 1], absMACRO(d__2))) * eps) {
e[m] = 0.;
goto L30;
}
/* L20: */
}
m = *n;
L30:
l = l1;
lsv = l;
lend = m;
lendsv = lend;
l1 = m + 1;
if (lend == l) {
goto L10;
}
/* Scale submatrix in rows and columns L to LEND */
i__1 = lend - l + 1;
anorm = template_lapack_lanst("I", &i__1, &d__[l], &e[l]);
iscale = 0;
if (anorm > ssfmax) {
iscale = 1;
i__1 = lend - l + 1;
template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmax, &i__1, &c__1, &d__[l], n,
info);
i__1 = lend - l;
template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmax, &i__1, &c__1, &e[l], n,
info);
} else if (anorm < ssfmin) {
iscale = 2;
i__1 = lend - l + 1;
template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmin, &i__1, &c__1, &d__[l], n,
info);
i__1 = lend - l;
template_lapack_lascl("G", &c__0, &c__0, &anorm, &ssfmin, &i__1, &c__1, &e[l], n,
info);
}
i__1 = lend - 1;
for (i__ = l; i__ <= i__1; ++i__) {
/* Computing 2nd power */
d__1 = e[i__];
e[i__] = d__1 * d__1;
/* L40: */
}
/* Choose between QL and QR iteration */
if ((d__1 = d__[lend], absMACRO(d__1)) < (d__2 = d__[l], absMACRO(d__2))) {
lend = lsv;
l = lendsv;
}
if (lend >= l) {
/* QL Iteration
Look for small subdiagonal element. */
L50:
if (l != lend) {
i__1 = lend - 1;
for (m = l; m <= i__1; ++m) {
if ((d__2 = e[m], absMACRO(d__2)) <= eps2 * (d__1 = d__[m] * d__[m
+ 1], absMACRO(d__1))) {
goto L70;
}
/* L60: */
}
}
m = lend;
L70:
if (m < lend) {
e[m] = 0.;
}
p = d__[l];
if (m == l) {
goto L90;
}
/* If remaining matrix is 2 by 2, use DLAE2 to compute its
eigenvalues. */
if (m == l + 1) {
rte = template_blas_sqrt(e[l]);
template_lapack_lae2(&d__[l], &rte, &d__[l + 1], &rt1, &rt2);
d__[l] = rt1;
d__[l + 1] = rt2;
e[l] = 0.;
l += 2;
if (l <= lend) {
goto L50;
}
goto L150;
}
if (jtot == nmaxit) {
goto L150;
}
++jtot;
/* Form shift. */
rte = template_blas_sqrt(e[l]);
sigma = (d__[l + 1] - p) / (rte * 2.);
r__ = template_lapack_lapy2(&sigma, &c_b32);
sigma = p - rte / (sigma + template_lapack_d_sign(&r__, &sigma));
c__ = 1.;
s = 0.;
gamma = d__[m] - sigma;
p = gamma * gamma;
/* Inner loop */
i__1 = l;
for (i__ = m - 1; i__ >= i__1; --i__) {
bb = e[i__];
r__ = p + bb;
if (i__ != m - 1) {
e[i__ + 1] = s * r__;
}
oldc = c__;
c__ = p / r__;
s = bb / r__;
oldgam = gamma;
alpha = d__[i__];
gamma = c__ * (alpha - sigma) - s * oldgam;
d__[i__ + 1] = oldgam + (alpha - gamma);
if (c__ != 0.) {
p = gamma * gamma / c__;
} else {
p = oldc * bb;
}
/* L80: */
}
e[l] = s * p;
d__[l] = sigma + gamma;
goto L50;
/* Eigenvalue found. */
L90:
d__[l] = p;
++l;
if (l <= lend) {
goto L50;
}
goto L150;
} else {
/* QR Iteration
Look for small superdiagonal element. */
L100:
i__1 = lend + 1;
for (m = l; m >= i__1; --m) {
if ((d__2 = e[m - 1], absMACRO(d__2)) <= eps2 * (d__1 = d__[m] * d__[m
- 1], absMACRO(d__1))) {
goto L120;
}
/* L110: */
}
m = lend;
L120:
if (m > lend) {
e[m - 1] = 0.;
}
p = d__[l];
if (m == l) {
goto L140;
}
/* If remaining matrix is 2 by 2, use DLAE2 to compute its
eigenvalues. */
if (m == l - 1) {
rte = template_blas_sqrt(e[l - 1]);
template_lapack_lae2(&d__[l], &rte, &d__[l - 1], &rt1, &rt2);
d__[l] = rt1;
d__[l - 1] = rt2;
e[l - 1] = 0.;
l += -2;
if (l >= lend) {
goto L100;
}
goto L150;
}
if (jtot == nmaxit) {
goto L150;
}
++jtot;
/* Form shift. */
rte = template_blas_sqrt(e[l - 1]);
sigma = (d__[l - 1] - p) / (rte * 2.);
r__ = template_lapack_lapy2(&sigma, &c_b32);
sigma = p - rte / (sigma + template_lapack_d_sign(&r__, &sigma));
c__ = 1.;
s = 0.;
gamma = d__[m] - sigma;
p = gamma * gamma;
/* Inner loop */
i__1 = l - 1;
for (i__ = m; i__ <= i__1; ++i__) {
bb = e[i__];
r__ = p + bb;
if (i__ != m) {
e[i__ - 1] = s * r__;
}
oldc = c__;
c__ = p / r__;
s = bb / r__;
oldgam = gamma;
alpha = d__[i__ + 1];
gamma = c__ * (alpha - sigma) - s * oldgam;
d__[i__] = oldgam + (alpha - gamma);
if (c__ != 0.) {
p = gamma * gamma / c__;
} else {
p = oldc * bb;
}
/* L130: */
}
e[l - 1] = s * p;
d__[l] = sigma + gamma;
goto L100;
/* Eigenvalue found. */
L140:
d__[l] = p;
--l;
if (l >= lend) {
goto L100;
}
goto L150;
}
/* Undo scaling if necessary */
L150:
if (iscale == 1) {
i__1 = lendsv - lsv + 1;
template_lapack_lascl("G", &c__0, &c__0, &ssfmax, &anorm, &i__1, &c__1, &d__[lsv],
n, info);
}
if (iscale == 2) {
i__1 = lendsv - lsv + 1;
template_lapack_lascl("G", &c__0, &c__0, &ssfmin, &anorm, &i__1, &c__1, &d__[lsv],
n, info);
}
/* Check for no convergence to an eigenvalue after a total
of N*MAXIT iterations. */
if (jtot < nmaxit) {
goto L10;
}
i__1 = *n - 1;
for (i__ = 1; i__ <= i__1; ++i__) {
if (e[i__] != 0.) {
++(*info);
}
/* L160: */
}
goto L180;
/* Sort eigenvalues in increasing order. */
L170:
template_lapack_lasrt("I", n, &d__[1], info);
L180:
return 0;
/* End of DSTERF */
} /* dsterf_ */
#endif
|