slasq3.c 8.57 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12
/* slasq3.f -- translated by f2c (version 20061008).
   You must link the resulting object file with libf2c:
	on Microsoft Windows system, link with libf2c.lib;
	on Linux or Unix systems, link with .../path/to/libf2c.a -lm
	or, if you install libf2c.a in a standard place, with -lf2c -lm
	-- in that order, at the end of the command line, as in
		cc *.o -lf2c -lm
	Source for libf2c is in /netlib/f2c/libf2c.zip, e.g.,

		http://www.netlib.org/f2c/libf2c.zip
*/

13 14
#include "clapack.h"

15

16 17
/* Subroutine */ int slasq3_(integer *i0, integer *n0, real *z__, integer *pp, 
	 real *dmin__, real *sigma, real *desig, real *qmax, integer *nfail, 
18 19 20
	integer *iter, integer *ndiv, logical *ieee, integer *ttype, real *
	dmin1, real *dmin2, real *dn, real *dn1, real *dn2, real *g, real *
	tau)
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
{
    /* System generated locals */
    integer i__1;
    real r__1, r__2;

    /* Builtin functions */
    double sqrt(doublereal);

    /* Local variables */
    real s, t;
    integer j4, nn;
    real eps, tol;
    integer n0in, ipn4;
    real tol2, temp;
    extern /* Subroutine */ int slasq4_(integer *, integer *, real *, integer 
	    *, integer *, real *, real *, real *, real *, real *, real *, 
37 38 39 40
	    real *, integer *, real *), slasq5_(integer *, integer *, real *, 
	    integer *, real *, real *, real *, real *, real *, real *, real *, 
	     logical *), slasq6_(integer *, integer *, real *, integer *, 
	    real *, real *, real *, real *, real *, real *);
41
    extern doublereal slamch_(char *);
42 43
    extern logical sisnan_(real *);

44

45
/*  -- LAPACK routine (version 3.2)                                    -- */
46

47 48 49 50 51 52 53
/*  -- Contributed by Osni Marques of the Lawrence Berkeley National   -- */
/*  -- Laboratory and Beresford Parlett of the Univ. of California at  -- */
/*  -- Berkeley                                                        -- */
/*  -- November 2008                                                   -- */

/*  -- LAPACK is a software package provided by Univ. of Tennessee,    -- */
/*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- */
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

/*     .. Scalar Arguments .. */
/*     .. */
/*     .. Array Arguments .. */
/*     .. */

/*  Purpose */
/*  ======= */

/*  SLASQ3 checks for deflation, computes a shift (TAU) and calls dqds. */
/*  In case of failure it changes shifts, and tries again until output */
/*  is positive. */

/*  Arguments */
/*  ========= */

/*  I0     (input) INTEGER */
/*         First index. */

/*  N0     (input) INTEGER */
/*         Last index. */

/*  Z      (input) REAL array, dimension ( 4*N ) */
/*         Z holds the qd array. */

79
/*  PP     (input/output) INTEGER */
80
/*         PP=0 for ping, PP=1 for pong. */
81 82 83
/*         PP=2 indicates that flipping was applied to the Z array */
/*         and that the initial tests for deflation should not be */
/*         performed. */
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

/*  DMIN   (output) REAL */
/*         Minimum value of d. */

/*  SIGMA  (output) REAL */
/*         Sum of shifts used in current segment. */

/*  DESIG  (input/output) REAL */
/*         Lower order part of SIGMA */

/*  QMAX   (input) REAL */
/*         Maximum value of q. */

/*  NFAIL  (output) INTEGER */
/*         Number of times shift was too big. */

/*  ITER   (output) INTEGER */
/*         Number of iterations. */

/*  NDIV   (output) INTEGER */
/*         Number of divisions. */

/*  IEEE   (input) LOGICAL */
/*         Flag for IEEE or non IEEE arithmetic (passed to SLASQ5). */

109 110 111 112 113 114 115
/*  TTYPE  (input/output) INTEGER */
/*         Shift type. */

/*  DMIN1, DMIN2, DN, DN1, DN2, G, TAU (input/output) REAL */
/*         These are passed as arguments in order to save their values */
/*         between calls to SLASQ3. */

116 117 118 119 120 121 122 123 124 125 126 127
/*  ===================================================================== */

/*     .. Parameters .. */
/*     .. */
/*     .. Local Scalars .. */
/*     .. */
/*     .. External Subroutines .. */
/*     .. */
/*     .. External Function .. */
/*     .. */
/*     .. Intrinsic Functions .. */
/*     .. */
128 129
/*     .. Executable Statements .. */

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
    /* Parameter adjustments */
    --z__;

    /* Function Body */
    n0in = *n0;
    eps = slamch_("Precision");
    tol = eps * 100.f;
/* Computing 2nd power */
    r__1 = tol;
    tol2 = r__1 * r__1;

/*     Check for deflation. */

L10:

    if (*n0 < *i0) {
	return 0;
    }
    if (*n0 == *i0) {
	goto L20;
    }
    nn = (*n0 << 2) + *pp;
    if (*n0 == *i0 + 1) {
	goto L40;
    }

/*     Check whether E(N0-1) is negligible, 1 eigenvalue. */

    if (z__[nn - 5] > tol2 * (*sigma + z__[nn - 3]) && z__[nn - (*pp << 1) - 
	    4] > tol2 * z__[nn - 7]) {
	goto L30;
    }

L20:

    z__[(*n0 << 2) - 3] = z__[(*n0 << 2) + *pp - 3] + *sigma;
    --(*n0);
    goto L10;

/*     Check  whether E(N0-2) is negligible, 2 eigenvalues. */

L30:

    if (z__[nn - 9] > tol2 * *sigma && z__[nn - (*pp << 1) - 8] > tol2 * z__[
	    nn - 11]) {
	goto L50;
    }

L40:

    if (z__[nn - 3] > z__[nn - 7]) {
	s = z__[nn - 3];
	z__[nn - 3] = z__[nn - 7];
	z__[nn - 7] = s;
    }
    if (z__[nn - 5] > z__[nn - 3] * tol2) {
	t = (z__[nn - 7] - z__[nn - 3] + z__[nn - 5]) * .5f;
	s = z__[nn - 3] * (z__[nn - 5] / t);
	if (s <= t) {
	    s = z__[nn - 3] * (z__[nn - 5] / (t * (sqrt(s / t + 1.f) + 1.f)));
	} else {
	    s = z__[nn - 3] * (z__[nn - 5] / (t + sqrt(t) * sqrt(t + s)));
	}
	t = z__[nn - 7] + (s + z__[nn - 5]);
	z__[nn - 3] *= z__[nn - 7] / t;
	z__[nn - 7] = t;
    }
    z__[(*n0 << 2) - 7] = z__[nn - 7] + *sigma;
    z__[(*n0 << 2) - 3] = z__[nn - 3] + *sigma;
    *n0 += -2;
    goto L10;

L50:
203 204 205
    if (*pp == 2) {
	*pp = 0;
    }
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

/*     Reverse the qd-array, if warranted. */

    if (*dmin__ <= 0.f || *n0 < n0in) {
	if (z__[(*i0 << 2) + *pp - 3] * 1.5f < z__[(*n0 << 2) + *pp - 3]) {
	    ipn4 = *i0 + *n0 << 2;
	    i__1 = *i0 + *n0 - 1 << 1;
	    for (j4 = *i0 << 2; j4 <= i__1; j4 += 4) {
		temp = z__[j4 - 3];
		z__[j4 - 3] = z__[ipn4 - j4 - 3];
		z__[ipn4 - j4 - 3] = temp;
		temp = z__[j4 - 2];
		z__[j4 - 2] = z__[ipn4 - j4 - 2];
		z__[ipn4 - j4 - 2] = temp;
		temp = z__[j4 - 1];
		z__[j4 - 1] = z__[ipn4 - j4 - 5];
		z__[ipn4 - j4 - 5] = temp;
		temp = z__[j4];
		z__[j4] = z__[ipn4 - j4 - 4];
		z__[ipn4 - j4 - 4] = temp;
/* L60: */
	    }
	    if (*n0 - *i0 <= 4) {
		z__[(*n0 << 2) + *pp - 1] = z__[(*i0 << 2) + *pp - 1];
		z__[(*n0 << 2) - *pp] = z__[(*i0 << 2) - *pp];
	    }
/* Computing MIN */
233 234
	    r__1 = *dmin2, r__2 = z__[(*n0 << 2) + *pp - 1];
	    *dmin2 = dmin(r__1,r__2);
235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
/* Computing MIN */
	    r__1 = z__[(*n0 << 2) + *pp - 1], r__2 = z__[(*i0 << 2) + *pp - 1]
		    , r__1 = min(r__1,r__2), r__2 = z__[(*i0 << 2) + *pp + 3];
	    z__[(*n0 << 2) + *pp - 1] = dmin(r__1,r__2);
/* Computing MIN */
	    r__1 = z__[(*n0 << 2) - *pp], r__2 = z__[(*i0 << 2) - *pp], r__1 =
		     min(r__1,r__2), r__2 = z__[(*i0 << 2) - *pp + 4];
	    z__[(*n0 << 2) - *pp] = dmin(r__1,r__2);
/* Computing MAX */
	    r__1 = *qmax, r__2 = z__[(*i0 << 2) + *pp - 3], r__1 = max(r__1,
		    r__2), r__2 = z__[(*i0 << 2) + *pp + 1];
	    *qmax = dmax(r__1,r__2);
	    *dmin__ = -0.f;
	}
    }

251
/*     Choose a shift. */
252

253 254
    slasq4_(i0, n0, &z__[1], pp, &n0in, dmin__, dmin1, dmin2, dn, dn1, dn2, 
	    tau, ttype, g);
255

256
/*     Call dqds until DMIN > 0. */
257

258
L70:
259

260 261
    slasq5_(i0, n0, &z__[1], pp, tau, dmin__, dmin1, dmin2, dn, dn1, dn2, 
	    ieee);
262

263 264
    *ndiv += *n0 - *i0 + 2;
    ++(*iter);
265

266
/*     Check status. */
267

268
    if (*dmin__ >= 0.f && *dmin1 > 0.f) {
269

270
/*        Success. */
271

272
	goto L90;
273

274 275
    } else if (*dmin__ < 0.f && *dmin1 > 0.f && z__[(*n0 - 1 << 2) - *pp] < 
	    tol * (*sigma + *dn1) && dabs(*dn) < tol * *sigma) {
276

277
/*        Convergence hidden by negative DN. */
278

279 280 281 282
	z__[(*n0 - 1 << 2) - *pp + 2] = 0.f;
	*dmin__ = 0.f;
	goto L90;
    } else if (*dmin__ < 0.f) {
283

284
/*        TAU too big. Select new TAU and try again. */
285

286 287
	++(*nfail);
	if (*ttype < -22) {
288

289
/*           Failed twice. Play it safe. */
290

291 292
	    *tau = 0.f;
	} else if (*dmin1 > 0.f) {
293

294
/*           Late failure. Gives excellent shift. */
295

296 297 298
	    *tau = (*tau + *dmin__) * (1.f - eps * 2.f);
	    *ttype += -11;
	} else {
299

300
/*           Early failure. Divide by 4. */
301

302 303 304 305 306
	    *tau *= .25f;
	    *ttype += -12;
	}
	goto L70;
    } else if (sisnan_(dmin__)) {
307

308
/*        NaN. */
309

310
	if (*tau == 0.f) {
311 312
	    goto L80;
	} else {
313 314 315 316
	    *tau = 0.f;
	    goto L70;
	}
    } else {
317

318
/*        Possible underflow. Play it safe. */
319

320
	goto L80;
321 322 323 324
    }

/*     Risk of underflow. */

325 326
L80:
    slasq6_(i0, n0, &z__[1], pp, dmin__, dmin1, dmin2, dn, dn1, dn2);
327 328
    *ndiv += *n0 - *i0 + 2;
    ++(*iter);
329
    *tau = 0.f;
330

331 332 333
L90:
    if (*tau < *sigma) {
	*desig += *tau;
334 335 336
	t = *sigma + *desig;
	*desig -= t - *sigma;
    } else {
337 338
	t = *sigma + *tau;
	*desig = *sigma - (t - *tau) + *desig;
339 340 341 342 343 344 345 346
    }
    *sigma = t;

    return 0;

/*     End of SLASQ3 */

} /* slasq3_ */