/root/doris/contrib/openblas/lapack-netlib/SRC/slasd4.c
Line | Count | Source |
1 | | #include <math.h> |
2 | | #include <stdlib.h> |
3 | | #include <string.h> |
4 | | #include <stdio.h> |
5 | | #include <complex.h> |
6 | | #ifdef complex |
7 | | #undef complex |
8 | | #endif |
9 | | #ifdef I |
10 | | #undef I |
11 | | #endif |
12 | | |
13 | | #if defined(_WIN64) |
14 | | typedef long long BLASLONG; |
15 | | typedef unsigned long long BLASULONG; |
16 | | #else |
17 | | typedef long BLASLONG; |
18 | | typedef unsigned long BLASULONG; |
19 | | #endif |
20 | | |
21 | | #ifdef LAPACK_ILP64 |
22 | | typedef BLASLONG blasint; |
23 | | #if defined(_WIN64) |
24 | | #define blasabs(x) llabs(x) |
25 | | #else |
26 | | #define blasabs(x) labs(x) |
27 | | #endif |
28 | | #else |
29 | | typedef int blasint; |
30 | | #define blasabs(x) abs(x) |
31 | | #endif |
32 | | |
33 | | typedef blasint integer; |
34 | | |
35 | | typedef unsigned int uinteger; |
36 | | typedef char *address; |
37 | | typedef short int shortint; |
38 | | typedef float real; |
39 | | typedef double doublereal; |
40 | | typedef struct { real r, i; } complex; |
41 | | typedef struct { doublereal r, i; } doublecomplex; |
42 | | #ifdef _MSC_VER |
43 | | static inline _Fcomplex Cf(complex *z) {_Fcomplex zz={z->r , z->i}; return zz;} |
44 | | static inline _Dcomplex Cd(doublecomplex *z) {_Dcomplex zz={z->r , z->i};return zz;} |
45 | | static inline _Fcomplex * _pCf(complex *z) {return (_Fcomplex*)z;} |
46 | | static inline _Dcomplex * _pCd(doublecomplex *z) {return (_Dcomplex*)z;} |
47 | | #else |
48 | 0 | static inline _Complex float Cf(complex *z) {return z->r + z->i*_Complex_I;} |
49 | 0 | static inline _Complex double Cd(doublecomplex *z) {return z->r + z->i*_Complex_I;} |
50 | 0 | static inline _Complex float * _pCf(complex *z) {return (_Complex float*)z;} |
51 | 0 | static inline _Complex double * _pCd(doublecomplex *z) {return (_Complex double*)z;} |
52 | | #endif |
53 | | #define pCf(z) (*_pCf(z)) |
54 | | #define pCd(z) (*_pCd(z)) |
55 | | typedef blasint logical; |
56 | | |
57 | | typedef char logical1; |
58 | | typedef char integer1; |
59 | | |
60 | 0 | #define TRUE_ (1) |
61 | 0 | #define FALSE_ (0) |
62 | | |
63 | | /* Extern is for use with -E */ |
64 | | #ifndef Extern |
65 | | #define Extern extern |
66 | | #endif |
67 | | |
68 | | /* I/O stuff */ |
69 | | |
70 | | typedef int flag; |
71 | | typedef int ftnlen; |
72 | | typedef int ftnint; |
73 | | |
74 | | /*external read, write*/ |
75 | | typedef struct |
76 | | { flag cierr; |
77 | | ftnint ciunit; |
78 | | flag ciend; |
79 | | char *cifmt; |
80 | | ftnint cirec; |
81 | | } cilist; |
82 | | |
83 | | /*internal read, write*/ |
84 | | typedef struct |
85 | | { flag icierr; |
86 | | char *iciunit; |
87 | | flag iciend; |
88 | | char *icifmt; |
89 | | ftnint icirlen; |
90 | | ftnint icirnum; |
91 | | } icilist; |
92 | | |
93 | | /*open*/ |
94 | | typedef struct |
95 | | { flag oerr; |
96 | | ftnint ounit; |
97 | | char *ofnm; |
98 | | ftnlen ofnmlen; |
99 | | char *osta; |
100 | | char *oacc; |
101 | | char *ofm; |
102 | | ftnint orl; |
103 | | char *oblnk; |
104 | | } olist; |
105 | | |
106 | | /*close*/ |
107 | | typedef struct |
108 | | { flag cerr; |
109 | | ftnint cunit; |
110 | | char *csta; |
111 | | } cllist; |
112 | | |
113 | | /*rewind, backspace, endfile*/ |
114 | | typedef struct |
115 | | { flag aerr; |
116 | | ftnint aunit; |
117 | | } alist; |
118 | | |
119 | | /* inquire */ |
120 | | typedef struct |
121 | | { flag inerr; |
122 | | ftnint inunit; |
123 | | char *infile; |
124 | | ftnlen infilen; |
125 | | ftnint *inex; /*parameters in standard's order*/ |
126 | | ftnint *inopen; |
127 | | ftnint *innum; |
128 | | ftnint *innamed; |
129 | | char *inname; |
130 | | ftnlen innamlen; |
131 | | char *inacc; |
132 | | ftnlen inacclen; |
133 | | char *inseq; |
134 | | ftnlen inseqlen; |
135 | | char *indir; |
136 | | ftnlen indirlen; |
137 | | char *infmt; |
138 | | ftnlen infmtlen; |
139 | | char *inform; |
140 | | ftnint informlen; |
141 | | char *inunf; |
142 | | ftnlen inunflen; |
143 | | ftnint *inrecl; |
144 | | ftnint *innrec; |
145 | | char *inblank; |
146 | | ftnlen inblanklen; |
147 | | } inlist; |
148 | | |
149 | | #define VOID void |
150 | | |
151 | | union Multitype { /* for multiple entry points */ |
152 | | integer1 g; |
153 | | shortint h; |
154 | | integer i; |
155 | | /* longint j; */ |
156 | | real r; |
157 | | doublereal d; |
158 | | complex c; |
159 | | doublecomplex z; |
160 | | }; |
161 | | |
162 | | typedef union Multitype Multitype; |
163 | | |
164 | | struct Vardesc { /* for Namelist */ |
165 | | char *name; |
166 | | char *addr; |
167 | | ftnlen *dims; |
168 | | int type; |
169 | | }; |
170 | | typedef struct Vardesc Vardesc; |
171 | | |
172 | | struct Namelist { |
173 | | char *name; |
174 | | Vardesc **vars; |
175 | | int nvars; |
176 | | }; |
177 | | typedef struct Namelist Namelist; |
178 | | |
179 | 0 | #define abs(x) ((x) >= 0 ? (x) : -(x)) |
180 | | #define dabs(x) (fabs(x)) |
181 | 0 | #define f2cmin(a,b) ((a) <= (b) ? (a) : (b)) |
182 | 0 | #define f2cmax(a,b) ((a) >= (b) ? (a) : (b)) |
183 | | #define dmin(a,b) (f2cmin(a,b)) |
184 | | #define dmax(a,b) (f2cmax(a,b)) |
185 | | #define bit_test(a,b) ((a) >> (b) & 1) |
186 | | #define bit_clear(a,b) ((a) & ~((uinteger)1 << (b))) |
187 | | #define bit_set(a,b) ((a) | ((uinteger)1 << (b))) |
188 | | |
189 | | #define abort_() { sig_die("Fortran abort routine called", 1); } |
190 | | #define c_abs(z) (cabsf(Cf(z))) |
191 | | #define c_cos(R,Z) { pCf(R)=ccos(Cf(Z)); } |
192 | | #ifdef _MSC_VER |
193 | | #define c_div(c, a, b) {Cf(c)._Val[0] = (Cf(a)._Val[0]/Cf(b)._Val[0]); Cf(c)._Val[1]=(Cf(a)._Val[1]/Cf(b)._Val[1]);} |
194 | | #define z_div(c, a, b) {Cd(c)._Val[0] = (Cd(a)._Val[0]/Cd(b)._Val[0]); Cd(c)._Val[1]=(Cd(a)._Val[1]/df(b)._Val[1]);} |
195 | | #else |
196 | | #define c_div(c, a, b) {pCf(c) = Cf(a)/Cf(b);} |
197 | | #define z_div(c, a, b) {pCd(c) = Cd(a)/Cd(b);} |
198 | | #endif |
199 | | #define c_exp(R, Z) {pCf(R) = cexpf(Cf(Z));} |
200 | | #define c_log(R, Z) {pCf(R) = clogf(Cf(Z));} |
201 | | #define c_sin(R, Z) {pCf(R) = csinf(Cf(Z));} |
202 | | //#define c_sqrt(R, Z) {*(R) = csqrtf(Cf(Z));} |
203 | | #define c_sqrt(R, Z) {pCf(R) = csqrtf(Cf(Z));} |
204 | | #define d_abs(x) (fabs(*(x))) |
205 | | #define d_acos(x) (acos(*(x))) |
206 | | #define d_asin(x) (asin(*(x))) |
207 | | #define d_atan(x) (atan(*(x))) |
208 | | #define d_atn2(x, y) (atan2(*(x),*(y))) |
209 | | #define d_cnjg(R, Z) { pCd(R) = conj(Cd(Z)); } |
210 | | #define r_cnjg(R, Z) { pCf(R) = conjf(Cf(Z)); } |
211 | | #define d_cos(x) (cos(*(x))) |
212 | | #define d_cosh(x) (cosh(*(x))) |
213 | | #define d_dim(__a, __b) ( *(__a) > *(__b) ? *(__a) - *(__b) : 0.0 ) |
214 | | #define d_exp(x) (exp(*(x))) |
215 | | #define d_imag(z) (cimag(Cd(z))) |
216 | | #define r_imag(z) (cimagf(Cf(z))) |
217 | | #define d_int(__x) (*(__x)>0 ? floor(*(__x)) : -floor(- *(__x))) |
218 | | #define r_int(__x) (*(__x)>0 ? floor(*(__x)) : -floor(- *(__x))) |
219 | | #define d_lg10(x) ( 0.43429448190325182765 * log(*(x)) ) |
220 | | #define r_lg10(x) ( 0.43429448190325182765 * log(*(x)) ) |
221 | | #define d_log(x) (log(*(x))) |
222 | | #define d_mod(x, y) (fmod(*(x), *(y))) |
223 | | #define u_nint(__x) ((__x)>=0 ? floor((__x) + .5) : -floor(.5 - (__x))) |
224 | | #define d_nint(x) u_nint(*(x)) |
225 | | #define u_sign(__a,__b) ((__b) >= 0 ? ((__a) >= 0 ? (__a) : -(__a)) : -((__a) >= 0 ? (__a) : -(__a))) |
226 | | #define d_sign(a,b) u_sign(*(a),*(b)) |
227 | | #define r_sign(a,b) u_sign(*(a),*(b)) |
228 | | #define d_sin(x) (sin(*(x))) |
229 | | #define d_sinh(x) (sinh(*(x))) |
230 | | #define d_sqrt(x) (sqrt(*(x))) |
231 | | #define d_tan(x) (tan(*(x))) |
232 | | #define d_tanh(x) (tanh(*(x))) |
233 | | #define i_abs(x) abs(*(x)) |
234 | | #define i_dnnt(x) ((integer)u_nint(*(x))) |
235 | | #define i_len(s, n) (n) |
236 | | #define i_nint(x) ((integer)u_nint(*(x))) |
237 | | #define i_sign(a,b) ((integer)u_sign((integer)*(a),(integer)*(b))) |
238 | | #define pow_dd(ap, bp) ( pow(*(ap), *(bp))) |
239 | | #define pow_si(B,E) spow_ui(*(B),*(E)) |
240 | | #define pow_ri(B,E) spow_ui(*(B),*(E)) |
241 | | #define pow_di(B,E) dpow_ui(*(B),*(E)) |
242 | | #define pow_zi(p, a, b) {pCd(p) = zpow_ui(Cd(a), *(b));} |
243 | | #define pow_ci(p, a, b) {pCf(p) = cpow_ui(Cf(a), *(b));} |
244 | | #define pow_zz(R,A,B) {pCd(R) = cpow(Cd(A),*(B));} |
245 | | #define s_cat(lpp, rpp, rnp, np, llp) { ftnlen i, nc, ll; char *f__rp, *lp; ll = (llp); lp = (lpp); for(i=0; i < (int)*(np); ++i) { nc = ll; if((rnp)[i] < nc) nc = (rnp)[i]; ll -= nc; f__rp = (rpp)[i]; while(--nc >= 0) *lp++ = *(f__rp)++; } while(--ll >= 0) *lp++ = ' '; } |
246 | | #define s_cmp(a,b,c,d) ((integer)strncmp((a),(b),f2cmin((c),(d)))) |
247 | | #define s_copy(A,B,C,D) { int __i,__m; for (__i=0, __m=f2cmin((C),(D)); __i<__m && (B)[__i] != 0; ++__i) (A)[__i] = (B)[__i]; } |
248 | | #define sig_die(s, kill) { exit(1); } |
249 | | #define s_stop(s, n) {exit(0);} |
250 | | static char junk[] = "\n@(#)LIBF77 VERSION 19990503\n"; |
251 | | #define z_abs(z) (cabs(Cd(z))) |
252 | | #define z_exp(R, Z) {pCd(R) = cexp(Cd(Z));} |
253 | | #define z_sqrt(R, Z) {pCd(R) = csqrt(Cd(Z));} |
254 | | #define myexit_() break; |
255 | | #define mycycle() continue; |
256 | | #define myceiling(w) {ceil(w)} |
257 | | #define myhuge(w) {HUGE_VAL} |
258 | | //#define mymaxloc_(w,s,e,n) {if (sizeof(*(w)) == sizeof(double)) dmaxloc_((w),*(s),*(e),n); else dmaxloc_((w),*(s),*(e),n);} |
259 | | #define mymaxloc(w,s,e,n) {dmaxloc_(w,*(s),*(e),n)} |
260 | | |
261 | | /* procedure parameter types for -A and -C++ */ |
262 | | |
263 | | |
264 | | #ifdef __cplusplus |
265 | | typedef logical (*L_fp)(...); |
266 | | #else |
267 | | typedef logical (*L_fp)(); |
268 | | #endif |
269 | | |
270 | 0 | static float spow_ui(float x, integer n) { |
271 | 0 | float pow=1.0; unsigned long int u; |
272 | 0 | if(n != 0) { |
273 | 0 | if(n < 0) n = -n, x = 1/x; |
274 | 0 | for(u = n; ; ) { |
275 | 0 | if(u & 01) pow *= x; |
276 | 0 | if(u >>= 1) x *= x; |
277 | 0 | else break; |
278 | 0 | } |
279 | 0 | } |
280 | 0 | return pow; |
281 | 0 | } |
282 | 0 | static double dpow_ui(double x, integer n) { |
283 | 0 | double pow=1.0; unsigned long int u; |
284 | 0 | if(n != 0) { |
285 | 0 | if(n < 0) n = -n, x = 1/x; |
286 | 0 | for(u = n; ; ) { |
287 | 0 | if(u & 01) pow *= x; |
288 | 0 | if(u >>= 1) x *= x; |
289 | 0 | else break; |
290 | 0 | } |
291 | 0 | } |
292 | 0 | return pow; |
293 | 0 | } |
294 | | #ifdef _MSC_VER |
295 | | static _Fcomplex cpow_ui(complex x, integer n) { |
296 | | complex pow={1.0,0.0}; unsigned long int u; |
297 | | if(n != 0) { |
298 | | if(n < 0) n = -n, x.r = 1/x.r, x.i=1/x.i; |
299 | | for(u = n; ; ) { |
300 | | if(u & 01) pow.r *= x.r, pow.i *= x.i; |
301 | | if(u >>= 1) x.r *= x.r, x.i *= x.i; |
302 | | else break; |
303 | | } |
304 | | } |
305 | | _Fcomplex p={pow.r, pow.i}; |
306 | | return p; |
307 | | } |
308 | | #else |
309 | 0 | static _Complex float cpow_ui(_Complex float x, integer n) { |
310 | 0 | _Complex float pow=1.0; unsigned long int u; |
311 | 0 | if(n != 0) { |
312 | 0 | if(n < 0) n = -n, x = 1/x; |
313 | 0 | for(u = n; ; ) { |
314 | 0 | if(u & 01) pow *= x; |
315 | 0 | if(u >>= 1) x *= x; |
316 | 0 | else break; |
317 | 0 | } |
318 | 0 | } |
319 | 0 | return pow; |
320 | 0 | } |
321 | | #endif |
322 | | #ifdef _MSC_VER |
323 | | static _Dcomplex zpow_ui(_Dcomplex x, integer n) { |
324 | | _Dcomplex pow={1.0,0.0}; unsigned long int u; |
325 | | if(n != 0) { |
326 | | if(n < 0) n = -n, x._Val[0] = 1/x._Val[0], x._Val[1] =1/x._Val[1]; |
327 | | for(u = n; ; ) { |
328 | | if(u & 01) pow._Val[0] *= x._Val[0], pow._Val[1] *= x._Val[1]; |
329 | | if(u >>= 1) x._Val[0] *= x._Val[0], x._Val[1] *= x._Val[1]; |
330 | | else break; |
331 | | } |
332 | | } |
333 | | _Dcomplex p = {pow._Val[0], pow._Val[1]}; |
334 | | return p; |
335 | | } |
336 | | #else |
337 | 0 | static _Complex double zpow_ui(_Complex double x, integer n) { |
338 | 0 | _Complex double pow=1.0; unsigned long int u; |
339 | 0 | if(n != 0) { |
340 | 0 | if(n < 0) n = -n, x = 1/x; |
341 | 0 | for(u = n; ; ) { |
342 | 0 | if(u & 01) pow *= x; |
343 | 0 | if(u >>= 1) x *= x; |
344 | 0 | else break; |
345 | 0 | } |
346 | 0 | } |
347 | 0 | return pow; |
348 | 0 | } |
349 | | #endif |
350 | 0 | static integer pow_ii(integer x, integer n) { |
351 | 0 | integer pow; unsigned long int u; |
352 | 0 | if (n <= 0) { |
353 | 0 | if (n == 0 || x == 1) pow = 1; |
354 | 0 | else if (x != -1) pow = x == 0 ? 1/x : 0; |
355 | 0 | else n = -n; |
356 | 0 | } |
357 | 0 | if ((n > 0) || !(n == 0 || x == 1 || x != -1)) { |
358 | 0 | u = n; |
359 | 0 | for(pow = 1; ; ) { |
360 | 0 | if(u & 01) pow *= x; |
361 | 0 | if(u >>= 1) x *= x; |
362 | 0 | else break; |
363 | 0 | } |
364 | 0 | } |
365 | 0 | return pow; |
366 | 0 | } |
367 | | static integer dmaxloc_(double *w, integer s, integer e, integer *n) |
368 | 0 | { |
369 | 0 | double m; integer i, mi; |
370 | 0 | for(m=w[s-1], mi=s, i=s+1; i<=e; i++) |
371 | 0 | if (w[i-1]>m) mi=i ,m=w[i-1]; |
372 | 0 | return mi-s+1; |
373 | 0 | } |
374 | | static integer smaxloc_(float *w, integer s, integer e, integer *n) |
375 | 0 | { |
376 | 0 | float m; integer i, mi; |
377 | 0 | for(m=w[s-1], mi=s, i=s+1; i<=e; i++) |
378 | 0 | if (w[i-1]>m) mi=i ,m=w[i-1]; |
379 | 0 | return mi-s+1; |
380 | 0 | } |
381 | 0 | static inline void cdotc_(complex *z, integer *n_, complex *x, integer *incx_, complex *y, integer *incy_) { |
382 | 0 | integer n = *n_, incx = *incx_, incy = *incy_, i; |
383 | 0 | #ifdef _MSC_VER |
384 | 0 | _Fcomplex zdotc = {0.0, 0.0}; |
385 | 0 | if (incx == 1 && incy == 1) { |
386 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
387 | 0 | zdotc._Val[0] += conjf(Cf(&x[i]))._Val[0] * Cf(&y[i])._Val[0]; |
388 | 0 | zdotc._Val[1] += conjf(Cf(&x[i]))._Val[1] * Cf(&y[i])._Val[1]; |
389 | 0 | } |
390 | 0 | } else { |
391 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
392 | 0 | zdotc._Val[0] += conjf(Cf(&x[i*incx]))._Val[0] * Cf(&y[i*incy])._Val[0]; |
393 | 0 | zdotc._Val[1] += conjf(Cf(&x[i*incx]))._Val[1] * Cf(&y[i*incy])._Val[1]; |
394 | 0 | } |
395 | 0 | } |
396 | 0 | pCf(z) = zdotc; |
397 | 0 | } |
398 | 0 | #else |
399 | 0 | _Complex float zdotc = 0.0; |
400 | 0 | if (incx == 1 && incy == 1) { |
401 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
402 | 0 | zdotc += conjf(Cf(&x[i])) * Cf(&y[i]); |
403 | 0 | } |
404 | 0 | } else { |
405 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
406 | 0 | zdotc += conjf(Cf(&x[i*incx])) * Cf(&y[i*incy]); |
407 | 0 | } |
408 | 0 | } |
409 | 0 | pCf(z) = zdotc; |
410 | 0 | } |
411 | | #endif |
412 | 0 | static inline void zdotc_(doublecomplex *z, integer *n_, doublecomplex *x, integer *incx_, doublecomplex *y, integer *incy_) { |
413 | 0 | integer n = *n_, incx = *incx_, incy = *incy_, i; |
414 | 0 | #ifdef _MSC_VER |
415 | 0 | _Dcomplex zdotc = {0.0, 0.0}; |
416 | 0 | if (incx == 1 && incy == 1) { |
417 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
418 | 0 | zdotc._Val[0] += conj(Cd(&x[i]))._Val[0] * Cd(&y[i])._Val[0]; |
419 | 0 | zdotc._Val[1] += conj(Cd(&x[i]))._Val[1] * Cd(&y[i])._Val[1]; |
420 | 0 | } |
421 | 0 | } else { |
422 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
423 | 0 | zdotc._Val[0] += conj(Cd(&x[i*incx]))._Val[0] * Cd(&y[i*incy])._Val[0]; |
424 | 0 | zdotc._Val[1] += conj(Cd(&x[i*incx]))._Val[1] * Cd(&y[i*incy])._Val[1]; |
425 | 0 | } |
426 | 0 | } |
427 | 0 | pCd(z) = zdotc; |
428 | 0 | } |
429 | 0 | #else |
430 | 0 | _Complex double zdotc = 0.0; |
431 | 0 | if (incx == 1 && incy == 1) { |
432 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
433 | 0 | zdotc += conj(Cd(&x[i])) * Cd(&y[i]); |
434 | 0 | } |
435 | 0 | } else { |
436 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
437 | 0 | zdotc += conj(Cd(&x[i*incx])) * Cd(&y[i*incy]); |
438 | 0 | } |
439 | 0 | } |
440 | 0 | pCd(z) = zdotc; |
441 | 0 | } |
442 | | #endif |
443 | 0 | static inline void cdotu_(complex *z, integer *n_, complex *x, integer *incx_, complex *y, integer *incy_) { |
444 | 0 | integer n = *n_, incx = *incx_, incy = *incy_, i; |
445 | 0 | #ifdef _MSC_VER |
446 | 0 | _Fcomplex zdotc = {0.0, 0.0}; |
447 | 0 | if (incx == 1 && incy == 1) { |
448 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
449 | 0 | zdotc._Val[0] += Cf(&x[i])._Val[0] * Cf(&y[i])._Val[0]; |
450 | 0 | zdotc._Val[1] += Cf(&x[i])._Val[1] * Cf(&y[i])._Val[1]; |
451 | 0 | } |
452 | 0 | } else { |
453 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
454 | 0 | zdotc._Val[0] += Cf(&x[i*incx])._Val[0] * Cf(&y[i*incy])._Val[0]; |
455 | 0 | zdotc._Val[1] += Cf(&x[i*incx])._Val[1] * Cf(&y[i*incy])._Val[1]; |
456 | 0 | } |
457 | 0 | } |
458 | 0 | pCf(z) = zdotc; |
459 | 0 | } |
460 | 0 | #else |
461 | 0 | _Complex float zdotc = 0.0; |
462 | 0 | if (incx == 1 && incy == 1) { |
463 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
464 | 0 | zdotc += Cf(&x[i]) * Cf(&y[i]); |
465 | 0 | } |
466 | 0 | } else { |
467 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
468 | 0 | zdotc += Cf(&x[i*incx]) * Cf(&y[i*incy]); |
469 | 0 | } |
470 | 0 | } |
471 | 0 | pCf(z) = zdotc; |
472 | 0 | } |
473 | | #endif |
474 | 0 | static inline void zdotu_(doublecomplex *z, integer *n_, doublecomplex *x, integer *incx_, doublecomplex *y, integer *incy_) { |
475 | 0 | integer n = *n_, incx = *incx_, incy = *incy_, i; |
476 | 0 | #ifdef _MSC_VER |
477 | 0 | _Dcomplex zdotc = {0.0, 0.0}; |
478 | 0 | if (incx == 1 && incy == 1) { |
479 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
480 | 0 | zdotc._Val[0] += Cd(&x[i])._Val[0] * Cd(&y[i])._Val[0]; |
481 | 0 | zdotc._Val[1] += Cd(&x[i])._Val[1] * Cd(&y[i])._Val[1]; |
482 | 0 | } |
483 | 0 | } else { |
484 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
485 | 0 | zdotc._Val[0] += Cd(&x[i*incx])._Val[0] * Cd(&y[i*incy])._Val[0]; |
486 | 0 | zdotc._Val[1] += Cd(&x[i*incx])._Val[1] * Cd(&y[i*incy])._Val[1]; |
487 | 0 | } |
488 | 0 | } |
489 | 0 | pCd(z) = zdotc; |
490 | 0 | } |
491 | 0 | #else |
492 | 0 | _Complex double zdotc = 0.0; |
493 | 0 | if (incx == 1 && incy == 1) { |
494 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
495 | 0 | zdotc += Cd(&x[i]) * Cd(&y[i]); |
496 | 0 | } |
497 | 0 | } else { |
498 | 0 | for (i=0;i<n;i++) { /* zdotc = zdotc + dconjg(x(i))* y(i) */ |
499 | 0 | zdotc += Cd(&x[i*incx]) * Cd(&y[i*incy]); |
500 | 0 | } |
501 | 0 | } |
502 | 0 | pCd(z) = zdotc; |
503 | 0 | } |
504 | | #endif |
505 | | /* -- translated by f2c (version 20000121). |
506 | | You must link the resulting object file with the libraries: |
507 | | -lf2c -lm (in that order) |
508 | | */ |
509 | | |
510 | | |
511 | | |
512 | | |
513 | | /* > \brief \b SLASD4 computes the square root of the i-th updated eigenvalue of a positive symmetric rank-one |
514 | | modification to a positive diagonal matrix. Used by sbdsdc. */ |
515 | | |
516 | | /* =========== DOCUMENTATION =========== */ |
517 | | |
518 | | /* Online html documentation available at */ |
519 | | /* http://www.netlib.org/lapack/explore-html/ */ |
520 | | |
521 | | /* > \htmlonly */ |
522 | | /* > Download SLASD4 + dependencies */ |
523 | | /* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.tgz?format=tgz&filename=/lapack/lapack_routine/slasd4. |
524 | | f"> */ |
525 | | /* > [TGZ]</a> */ |
526 | | /* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.zip?format=zip&filename=/lapack/lapack_routine/slasd4. |
527 | | f"> */ |
528 | | /* > [ZIP]</a> */ |
529 | | /* > <a href="http://www.netlib.org/cgi-bin/netlibfiles.txt?format=txt&filename=/lapack/lapack_routine/slasd4. |
530 | | f"> */ |
531 | | /* > [TXT]</a> */ |
532 | | /* > \endhtmlonly */ |
533 | | |
534 | | /* Definition: */ |
535 | | /* =========== */ |
536 | | |
537 | | /* SUBROUTINE SLASD4( N, I, D, Z, DELTA, RHO, SIGMA, WORK, INFO ) */ |
538 | | |
539 | | /* INTEGER I, INFO, N */ |
540 | | /* REAL RHO, SIGMA */ |
541 | | /* REAL D( * ), DELTA( * ), WORK( * ), Z( * ) */ |
542 | | |
543 | | |
544 | | /* > \par Purpose: */ |
545 | | /* ============= */ |
546 | | /* > */ |
547 | | /* > \verbatim */ |
548 | | /* > */ |
549 | | /* > This subroutine computes the square root of the I-th updated */ |
550 | | /* > eigenvalue of a positive symmetric rank-one modification to */ |
551 | | /* > a positive diagonal matrix whose entries are given as the squares */ |
552 | | /* > of the corresponding entries in the array d, and that */ |
553 | | /* > */ |
554 | | /* > 0 <= D(i) < D(j) for i < j */ |
555 | | /* > */ |
556 | | /* > and that RHO > 0. This is arranged by the calling routine, and is */ |
557 | | /* > no loss in generality. The rank-one modified system is thus */ |
558 | | /* > */ |
559 | | /* > diag( D ) * diag( D ) + RHO * Z * Z_transpose. */ |
560 | | /* > */ |
561 | | /* > where we assume the Euclidean norm of Z is 1. */ |
562 | | /* > */ |
563 | | /* > The method consists of approximating the rational functions in the */ |
564 | | /* > secular equation by simpler interpolating rational functions. */ |
565 | | /* > \endverbatim */ |
566 | | |
567 | | /* Arguments: */ |
568 | | /* ========== */ |
569 | | |
570 | | /* > \param[in] N */ |
571 | | /* > \verbatim */ |
572 | | /* > N is INTEGER */ |
573 | | /* > The length of all arrays. */ |
574 | | /* > \endverbatim */ |
575 | | /* > */ |
576 | | /* > \param[in] I */ |
577 | | /* > \verbatim */ |
578 | | /* > I is INTEGER */ |
579 | | /* > The index of the eigenvalue to be computed. 1 <= I <= N. */ |
580 | | /* > \endverbatim */ |
581 | | /* > */ |
582 | | /* > \param[in] D */ |
583 | | /* > \verbatim */ |
584 | | /* > D is REAL array, dimension ( N ) */ |
585 | | /* > The original eigenvalues. It is assumed that they are in */ |
586 | | /* > order, 0 <= D(I) < D(J) for I < J. */ |
587 | | /* > \endverbatim */ |
588 | | /* > */ |
589 | | /* > \param[in] Z */ |
590 | | /* > \verbatim */ |
591 | | /* > Z is REAL array, dimension ( N ) */ |
592 | | /* > The components of the updating vector. */ |
593 | | /* > \endverbatim */ |
594 | | /* > */ |
595 | | /* > \param[out] DELTA */ |
596 | | /* > \verbatim */ |
597 | | /* > DELTA is REAL array, dimension ( N ) */ |
598 | | /* > If N .ne. 1, DELTA contains (D(j) - sigma_I) in its j-th */ |
599 | | /* > component. If N = 1, then DELTA(1) = 1. The vector DELTA */ |
600 | | /* > contains the information necessary to construct the */ |
601 | | /* > (singular) eigenvectors. */ |
602 | | /* > \endverbatim */ |
603 | | /* > */ |
604 | | /* > \param[in] RHO */ |
605 | | /* > \verbatim */ |
606 | | /* > RHO is REAL */ |
607 | | /* > The scalar in the symmetric updating formula. */ |
608 | | /* > \endverbatim */ |
609 | | /* > */ |
610 | | /* > \param[out] SIGMA */ |
611 | | /* > \verbatim */ |
612 | | /* > SIGMA is REAL */ |
613 | | /* > The computed sigma_I, the I-th updated eigenvalue. */ |
614 | | /* > \endverbatim */ |
615 | | /* > */ |
616 | | /* > \param[out] WORK */ |
617 | | /* > \verbatim */ |
618 | | /* > WORK is REAL array, dimension ( N ) */ |
619 | | /* > If N .ne. 1, WORK contains (D(j) + sigma_I) in its j-th */ |
620 | | /* > component. If N = 1, then WORK( 1 ) = 1. */ |
621 | | /* > \endverbatim */ |
622 | | /* > */ |
623 | | /* > \param[out] INFO */ |
624 | | /* > \verbatim */ |
625 | | /* > INFO is INTEGER */ |
626 | | /* > = 0: successful exit */ |
627 | | /* > > 0: if INFO = 1, the updating process failed. */ |
628 | | /* > \endverbatim */ |
629 | | |
630 | | /* > \par Internal Parameters: */ |
631 | | /* ========================= */ |
632 | | /* > */ |
633 | | /* > \verbatim */ |
634 | | /* > Logical variable ORGATI (origin-at-i?) is used for distinguishing */ |
635 | | /* > whether D(i) or D(i+1) is treated as the origin. */ |
636 | | /* > */ |
637 | | /* > ORGATI = .true. origin at i */ |
638 | | /* > ORGATI = .false. origin at i+1 */ |
639 | | /* > */ |
640 | | /* > Logical variable SWTCH3 (switch-for-3-poles?) is for noting */ |
641 | | /* > if we are working with THREE poles! */ |
642 | | /* > */ |
643 | | /* > MAXIT is the maximum number of iterations allowed for each */ |
644 | | /* > eigenvalue. */ |
645 | | /* > \endverbatim */ |
646 | | |
647 | | /* Authors: */ |
648 | | /* ======== */ |
649 | | |
650 | | /* > \author Univ. of Tennessee */ |
651 | | /* > \author Univ. of California Berkeley */ |
652 | | /* > \author Univ. of Colorado Denver */ |
653 | | /* > \author NAG Ltd. */ |
654 | | |
655 | | /* > \date December 2016 */ |
656 | | |
657 | | /* > \ingroup OTHERauxiliary */ |
658 | | |
659 | | /* > \par Contributors: */ |
660 | | /* ================== */ |
661 | | /* > */ |
662 | | /* > Ren-Cang Li, Computer Science Division, University of California */ |
663 | | /* > at Berkeley, USA */ |
664 | | /* > */ |
665 | | /* ===================================================================== */ |
666 | | /* Subroutine */ void slasd4_(integer *n, integer *i__, real *d__, real *z__, |
667 | | real *delta, real *rho, real *sigma, real *work, integer *info) |
668 | 0 | { |
669 | | /* System generated locals */ |
670 | 0 | integer i__1; |
671 | 0 | real r__1; |
672 | | |
673 | | /* Local variables */ |
674 | 0 | real dphi, sglb, dpsi, sgub; |
675 | 0 | integer iter; |
676 | 0 | real temp, prew, temp1, temp2, a, b, c__; |
677 | 0 | integer j; |
678 | 0 | real w, dtiim, delsq, dtiip; |
679 | 0 | integer niter; |
680 | 0 | real dtisq; |
681 | 0 | logical swtch; |
682 | 0 | real dtnsq; |
683 | 0 | extern /* Subroutine */ void slaed6_(integer *, logical *, real *, real *, |
684 | 0 | real *, real *, real *, integer *); |
685 | 0 | real delsq2; |
686 | 0 | extern /* Subroutine */ void slasd5_(integer *, real *, real *, real *, |
687 | 0 | real *, real *, real *); |
688 | 0 | real dd[3], dtnsq1; |
689 | 0 | logical swtch3; |
690 | 0 | integer ii; |
691 | 0 | real dw; |
692 | 0 | extern real slamch_(char *); |
693 | 0 | real zz[3]; |
694 | 0 | logical orgati; |
695 | 0 | real erretm, dtipsq, rhoinv; |
696 | 0 | integer ip1; |
697 | 0 | real sq2, eta, phi, eps, tau, psi; |
698 | 0 | logical geomavg; |
699 | 0 | integer iim1, iip1; |
700 | 0 | real tau2; |
701 | | |
702 | | |
703 | | /* -- LAPACK auxiliary routine (version 3.7.0) -- */ |
704 | | /* -- LAPACK is a software package provided by Univ. of Tennessee, -- */ |
705 | | /* -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..-- */ |
706 | | /* December 2016 */ |
707 | | |
708 | | |
709 | | /* ===================================================================== */ |
710 | | |
711 | | |
712 | | /* Since this routine is called in an inner loop, we do no argument */ |
713 | | /* checking. */ |
714 | | |
715 | | /* Quick return for N=1 and 2. */ |
716 | | |
717 | | /* Parameter adjustments */ |
718 | 0 | --work; |
719 | 0 | --delta; |
720 | 0 | --z__; |
721 | 0 | --d__; |
722 | | |
723 | | /* Function Body */ |
724 | 0 | *info = 0; |
725 | 0 | if (*n == 1) { |
726 | | |
727 | | /* Presumably, I=1 upon entry */ |
728 | |
|
729 | 0 | *sigma = sqrt(d__[1] * d__[1] + *rho * z__[1] * z__[1]); |
730 | 0 | delta[1] = 1.f; |
731 | 0 | work[1] = 1.f; |
732 | 0 | return; |
733 | 0 | } |
734 | 0 | if (*n == 2) { |
735 | 0 | slasd5_(i__, &d__[1], &z__[1], &delta[1], rho, sigma, &work[1]); |
736 | 0 | return; |
737 | 0 | } |
738 | | |
739 | | /* Compute machine epsilon */ |
740 | | |
741 | 0 | eps = slamch_("Epsilon"); |
742 | 0 | rhoinv = 1.f / *rho; |
743 | 0 | tau2 = 0.f; |
744 | | |
745 | | /* The case I = N */ |
746 | |
|
747 | 0 | if (*i__ == *n) { |
748 | | |
749 | | /* Initialize some basic variables */ |
750 | |
|
751 | 0 | ii = *n - 1; |
752 | 0 | niter = 1; |
753 | | |
754 | | /* Calculate initial guess */ |
755 | |
|
756 | 0 | temp = *rho / 2.f; |
757 | | |
758 | | /* If ||Z||_2 is not one, then TEMP should be set to */ |
759 | | /* RHO * ||Z||_2^2 / TWO */ |
760 | |
|
761 | 0 | temp1 = temp / (d__[*n] + sqrt(d__[*n] * d__[*n] + temp)); |
762 | 0 | i__1 = *n; |
763 | 0 | for (j = 1; j <= i__1; ++j) { |
764 | 0 | work[j] = d__[j] + d__[*n] + temp1; |
765 | 0 | delta[j] = d__[j] - d__[*n] - temp1; |
766 | | /* L10: */ |
767 | 0 | } |
768 | |
|
769 | 0 | psi = 0.f; |
770 | 0 | i__1 = *n - 2; |
771 | 0 | for (j = 1; j <= i__1; ++j) { |
772 | 0 | psi += z__[j] * z__[j] / (delta[j] * work[j]); |
773 | | /* L20: */ |
774 | 0 | } |
775 | |
|
776 | 0 | c__ = rhoinv + psi; |
777 | 0 | w = c__ + z__[ii] * z__[ii] / (delta[ii] * work[ii]) + z__[*n] * z__[* |
778 | 0 | n] / (delta[*n] * work[*n]); |
779 | |
|
780 | 0 | if (w <= 0.f) { |
781 | 0 | temp1 = sqrt(d__[*n] * d__[*n] + *rho); |
782 | 0 | temp = z__[*n - 1] * z__[*n - 1] / ((d__[*n - 1] + temp1) * (d__[* |
783 | 0 | n] - d__[*n - 1] + *rho / (d__[*n] + temp1))) + z__[*n] * |
784 | 0 | z__[*n] / *rho; |
785 | | |
786 | | /* The following TAU2 is to approximate */ |
787 | | /* SIGMA_n^2 - D( N )*D( N ) */ |
788 | |
|
789 | 0 | if (c__ <= temp) { |
790 | 0 | tau = *rho; |
791 | 0 | } else { |
792 | 0 | delsq = (d__[*n] - d__[*n - 1]) * (d__[*n] + d__[*n - 1]); |
793 | 0 | a = -c__ * delsq + z__[*n - 1] * z__[*n - 1] + z__[*n] * z__[* |
794 | 0 | n]; |
795 | 0 | b = z__[*n] * z__[*n] * delsq; |
796 | 0 | if (a < 0.f) { |
797 | 0 | tau2 = b * 2.f / (sqrt(a * a + b * 4.f * c__) - a); |
798 | 0 | } else { |
799 | 0 | tau2 = (a + sqrt(a * a + b * 4.f * c__)) / (c__ * 2.f); |
800 | 0 | } |
801 | 0 | tau = tau2 / (d__[*n] + sqrt(d__[*n] * d__[*n] + tau2)); |
802 | 0 | } |
803 | | |
804 | | /* It can be proved that */ |
805 | | /* D(N)^2+RHO/2 <= SIGMA_n^2 < D(N)^2+TAU2 <= D(N)^2+RHO */ |
806 | |
|
807 | 0 | } else { |
808 | 0 | delsq = (d__[*n] - d__[*n - 1]) * (d__[*n] + d__[*n - 1]); |
809 | 0 | a = -c__ * delsq + z__[*n - 1] * z__[*n - 1] + z__[*n] * z__[*n]; |
810 | 0 | b = z__[*n] * z__[*n] * delsq; |
811 | | |
812 | | /* The following TAU2 is to approximate */ |
813 | | /* SIGMA_n^2 - D( N )*D( N ) */ |
814 | |
|
815 | 0 | if (a < 0.f) { |
816 | 0 | tau2 = b * 2.f / (sqrt(a * a + b * 4.f * c__) - a); |
817 | 0 | } else { |
818 | 0 | tau2 = (a + sqrt(a * a + b * 4.f * c__)) / (c__ * 2.f); |
819 | 0 | } |
820 | 0 | tau = tau2 / (d__[*n] + sqrt(d__[*n] * d__[*n] + tau2)); |
821 | | |
822 | | /* It can be proved that */ |
823 | | /* D(N)^2 < D(N)^2+TAU2 < SIGMA(N)^2 < D(N)^2+RHO/2 */ |
824 | |
|
825 | 0 | } |
826 | | |
827 | | /* The following TAU is to approximate SIGMA_n - D( N ) */ |
828 | | |
829 | | /* TAU = TAU2 / ( D( N )+SQRT( D( N )*D( N )+TAU2 ) ) */ |
830 | |
|
831 | 0 | *sigma = d__[*n] + tau; |
832 | 0 | i__1 = *n; |
833 | 0 | for (j = 1; j <= i__1; ++j) { |
834 | 0 | delta[j] = d__[j] - d__[*n] - tau; |
835 | 0 | work[j] = d__[j] + d__[*n] + tau; |
836 | | /* L30: */ |
837 | 0 | } |
838 | | |
839 | | /* Evaluate PSI and the derivative DPSI */ |
840 | |
|
841 | 0 | dpsi = 0.f; |
842 | 0 | psi = 0.f; |
843 | 0 | erretm = 0.f; |
844 | 0 | i__1 = ii; |
845 | 0 | for (j = 1; j <= i__1; ++j) { |
846 | 0 | temp = z__[j] / (delta[j] * work[j]); |
847 | 0 | psi += z__[j] * temp; |
848 | 0 | dpsi += temp * temp; |
849 | 0 | erretm += psi; |
850 | | /* L40: */ |
851 | 0 | } |
852 | 0 | erretm = abs(erretm); |
853 | | |
854 | | /* Evaluate PHI and the derivative DPHI */ |
855 | |
|
856 | 0 | temp = z__[*n] / (delta[*n] * work[*n]); |
857 | 0 | phi = z__[*n] * temp; |
858 | 0 | dphi = temp * temp; |
859 | 0 | erretm = (-phi - psi) * 8.f + erretm - phi + rhoinv; |
860 | | /* $ + ABS( TAU2 )*( DPSI+DPHI ) */ |
861 | |
|
862 | 0 | w = rhoinv + phi + psi; |
863 | | |
864 | | /* Test for convergence */ |
865 | |
|
866 | 0 | if (abs(w) <= eps * erretm) { |
867 | 0 | goto L240; |
868 | 0 | } |
869 | | |
870 | | /* Calculate the new step */ |
871 | | |
872 | 0 | ++niter; |
873 | 0 | dtnsq1 = work[*n - 1] * delta[*n - 1]; |
874 | 0 | dtnsq = work[*n] * delta[*n]; |
875 | 0 | c__ = w - dtnsq1 * dpsi - dtnsq * dphi; |
876 | 0 | a = (dtnsq + dtnsq1) * w - dtnsq * dtnsq1 * (dpsi + dphi); |
877 | 0 | b = dtnsq * dtnsq1 * w; |
878 | 0 | if (c__ < 0.f) { |
879 | 0 | c__ = abs(c__); |
880 | 0 | } |
881 | 0 | if (c__ == 0.f) { |
882 | 0 | eta = *rho - *sigma * *sigma; |
883 | 0 | } else if (a >= 0.f) { |
884 | 0 | eta = (a + sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1)))) / ( |
885 | 0 | c__ * 2.f); |
886 | 0 | } else { |
887 | 0 | eta = b * 2.f / (a - sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1) |
888 | 0 | ))); |
889 | 0 | } |
890 | | |
891 | | /* Note, eta should be positive if w is negative, and */ |
892 | | /* eta should be negative otherwise. However, */ |
893 | | /* if for some reason caused by roundoff, eta*w > 0, */ |
894 | | /* we simply use one Newton step instead. This way */ |
895 | | /* will guarantee eta*w < 0. */ |
896 | |
|
897 | 0 | if (w * eta > 0.f) { |
898 | 0 | eta = -w / (dpsi + dphi); |
899 | 0 | } |
900 | 0 | temp = eta - dtnsq; |
901 | 0 | if (temp > *rho) { |
902 | 0 | eta = *rho + dtnsq; |
903 | 0 | } |
904 | |
|
905 | 0 | eta /= *sigma + sqrt(eta + *sigma * *sigma); |
906 | 0 | tau += eta; |
907 | 0 | *sigma += eta; |
908 | |
|
909 | 0 | i__1 = *n; |
910 | 0 | for (j = 1; j <= i__1; ++j) { |
911 | 0 | delta[j] -= eta; |
912 | 0 | work[j] += eta; |
913 | | /* L50: */ |
914 | 0 | } |
915 | | |
916 | | /* Evaluate PSI and the derivative DPSI */ |
917 | |
|
918 | 0 | dpsi = 0.f; |
919 | 0 | psi = 0.f; |
920 | 0 | erretm = 0.f; |
921 | 0 | i__1 = ii; |
922 | 0 | for (j = 1; j <= i__1; ++j) { |
923 | 0 | temp = z__[j] / (work[j] * delta[j]); |
924 | 0 | psi += z__[j] * temp; |
925 | 0 | dpsi += temp * temp; |
926 | 0 | erretm += psi; |
927 | | /* L60: */ |
928 | 0 | } |
929 | 0 | erretm = abs(erretm); |
930 | | |
931 | | /* Evaluate PHI and the derivative DPHI */ |
932 | |
|
933 | 0 | tau2 = work[*n] * delta[*n]; |
934 | 0 | temp = z__[*n] / tau2; |
935 | 0 | phi = z__[*n] * temp; |
936 | 0 | dphi = temp * temp; |
937 | 0 | erretm = (-phi - psi) * 8.f + erretm - phi + rhoinv; |
938 | | /* $ + ABS( TAU2 )*( DPSI+DPHI ) */ |
939 | |
|
940 | 0 | w = rhoinv + phi + psi; |
941 | | |
942 | | /* Main loop to update the values of the array DELTA */ |
943 | |
|
944 | 0 | iter = niter + 1; |
945 | |
|
946 | 0 | for (niter = iter; niter <= 400; ++niter) { |
947 | | |
948 | | /* Test for convergence */ |
949 | |
|
950 | 0 | if (abs(w) <= eps * erretm) { |
951 | 0 | goto L240; |
952 | 0 | } |
953 | | |
954 | | /* Calculate the new step */ |
955 | | |
956 | 0 | dtnsq1 = work[*n - 1] * delta[*n - 1]; |
957 | 0 | dtnsq = work[*n] * delta[*n]; |
958 | 0 | c__ = w - dtnsq1 * dpsi - dtnsq * dphi; |
959 | 0 | a = (dtnsq + dtnsq1) * w - dtnsq1 * dtnsq * (dpsi + dphi); |
960 | 0 | b = dtnsq1 * dtnsq * w; |
961 | 0 | if (a >= 0.f) { |
962 | 0 | eta = (a + sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1)))) / |
963 | 0 | (c__ * 2.f); |
964 | 0 | } else { |
965 | 0 | eta = b * 2.f / (a - sqrt((r__1 = a * a - b * 4.f * c__, abs( |
966 | 0 | r__1)))); |
967 | 0 | } |
968 | | |
969 | | /* Note, eta should be positive if w is negative, and */ |
970 | | /* eta should be negative otherwise. However, */ |
971 | | /* if for some reason caused by roundoff, eta*w > 0, */ |
972 | | /* we simply use one Newton step instead. This way */ |
973 | | /* will guarantee eta*w < 0. */ |
974 | |
|
975 | 0 | if (w * eta > 0.f) { |
976 | 0 | eta = -w / (dpsi + dphi); |
977 | 0 | } |
978 | 0 | temp = eta - dtnsq; |
979 | 0 | if (temp <= 0.f) { |
980 | 0 | eta /= 2.f; |
981 | 0 | } |
982 | |
|
983 | 0 | eta /= *sigma + sqrt(eta + *sigma * *sigma); |
984 | 0 | tau += eta; |
985 | 0 | *sigma += eta; |
986 | |
|
987 | 0 | i__1 = *n; |
988 | 0 | for (j = 1; j <= i__1; ++j) { |
989 | 0 | delta[j] -= eta; |
990 | 0 | work[j] += eta; |
991 | | /* L70: */ |
992 | 0 | } |
993 | | |
994 | | /* Evaluate PSI and the derivative DPSI */ |
995 | |
|
996 | 0 | dpsi = 0.f; |
997 | 0 | psi = 0.f; |
998 | 0 | erretm = 0.f; |
999 | 0 | i__1 = ii; |
1000 | 0 | for (j = 1; j <= i__1; ++j) { |
1001 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1002 | 0 | psi += z__[j] * temp; |
1003 | 0 | dpsi += temp * temp; |
1004 | 0 | erretm += psi; |
1005 | | /* L80: */ |
1006 | 0 | } |
1007 | 0 | erretm = abs(erretm); |
1008 | | |
1009 | | /* Evaluate PHI and the derivative DPHI */ |
1010 | |
|
1011 | 0 | tau2 = work[*n] * delta[*n]; |
1012 | 0 | temp = z__[*n] / tau2; |
1013 | 0 | phi = z__[*n] * temp; |
1014 | 0 | dphi = temp * temp; |
1015 | 0 | erretm = (-phi - psi) * 8.f + erretm - phi + rhoinv; |
1016 | | /* $ + ABS( TAU2 )*( DPSI+DPHI ) */ |
1017 | |
|
1018 | 0 | w = rhoinv + phi + psi; |
1019 | | /* L90: */ |
1020 | 0 | } |
1021 | | |
1022 | | /* Return with INFO = 1, NITER = MAXIT and not converged */ |
1023 | | |
1024 | 0 | *info = 1; |
1025 | 0 | goto L240; |
1026 | | |
1027 | | /* End for the case I = N */ |
1028 | |
|
1029 | 0 | } else { |
1030 | | |
1031 | | /* The case for I < N */ |
1032 | |
|
1033 | 0 | niter = 1; |
1034 | 0 | ip1 = *i__ + 1; |
1035 | | |
1036 | | /* Calculate initial guess */ |
1037 | |
|
1038 | 0 | delsq = (d__[ip1] - d__[*i__]) * (d__[ip1] + d__[*i__]); |
1039 | 0 | delsq2 = delsq / 2.f; |
1040 | 0 | sq2 = sqrt((d__[*i__] * d__[*i__] + d__[ip1] * d__[ip1]) / 2.f); |
1041 | 0 | temp = delsq2 / (d__[*i__] + sq2); |
1042 | 0 | i__1 = *n; |
1043 | 0 | for (j = 1; j <= i__1; ++j) { |
1044 | 0 | work[j] = d__[j] + d__[*i__] + temp; |
1045 | 0 | delta[j] = d__[j] - d__[*i__] - temp; |
1046 | | /* L100: */ |
1047 | 0 | } |
1048 | |
|
1049 | 0 | psi = 0.f; |
1050 | 0 | i__1 = *i__ - 1; |
1051 | 0 | for (j = 1; j <= i__1; ++j) { |
1052 | 0 | psi += z__[j] * z__[j] / (work[j] * delta[j]); |
1053 | | /* L110: */ |
1054 | 0 | } |
1055 | |
|
1056 | 0 | phi = 0.f; |
1057 | 0 | i__1 = *i__ + 2; |
1058 | 0 | for (j = *n; j >= i__1; --j) { |
1059 | 0 | phi += z__[j] * z__[j] / (work[j] * delta[j]); |
1060 | | /* L120: */ |
1061 | 0 | } |
1062 | 0 | c__ = rhoinv + psi + phi; |
1063 | 0 | w = c__ + z__[*i__] * z__[*i__] / (work[*i__] * delta[*i__]) + z__[ |
1064 | 0 | ip1] * z__[ip1] / (work[ip1] * delta[ip1]); |
1065 | |
|
1066 | 0 | geomavg = FALSE_; |
1067 | 0 | if (w > 0.f) { |
1068 | | |
1069 | | /* d(i)^2 < the ith sigma^2 < (d(i)^2+d(i+1)^2)/2 */ |
1070 | | |
1071 | | /* We choose d(i) as origin. */ |
1072 | |
|
1073 | 0 | orgati = TRUE_; |
1074 | 0 | ii = *i__; |
1075 | 0 | sglb = 0.f; |
1076 | 0 | sgub = delsq2 / (d__[*i__] + sq2); |
1077 | 0 | a = c__ * delsq + z__[*i__] * z__[*i__] + z__[ip1] * z__[ip1]; |
1078 | 0 | b = z__[*i__] * z__[*i__] * delsq; |
1079 | 0 | if (a > 0.f) { |
1080 | 0 | tau2 = b * 2.f / (a + sqrt((r__1 = a * a - b * 4.f * c__, abs( |
1081 | 0 | r__1)))); |
1082 | 0 | } else { |
1083 | 0 | tau2 = (a - sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1)))) / |
1084 | 0 | (c__ * 2.f); |
1085 | 0 | } |
1086 | | |
1087 | | /* TAU2 now is an estimation of SIGMA^2 - D( I )^2. The */ |
1088 | | /* following, however, is the corresponding estimation of */ |
1089 | | /* SIGMA - D( I ). */ |
1090 | |
|
1091 | 0 | tau = tau2 / (d__[*i__] + sqrt(d__[*i__] * d__[*i__] + tau2)); |
1092 | 0 | temp = sqrt(eps); |
1093 | 0 | if (d__[*i__] <= temp * d__[ip1] && (r__1 = z__[*i__], abs(r__1)) |
1094 | 0 | <= temp && d__[*i__] > 0.f) { |
1095 | | /* Computing MIN */ |
1096 | 0 | r__1 = d__[*i__] * 10.f; |
1097 | 0 | tau = f2cmin(r__1,sgub); |
1098 | 0 | geomavg = TRUE_; |
1099 | 0 | } |
1100 | 0 | } else { |
1101 | | |
1102 | | /* (d(i)^2+d(i+1)^2)/2 <= the ith sigma^2 < d(i+1)^2/2 */ |
1103 | | |
1104 | | /* We choose d(i+1) as origin. */ |
1105 | |
|
1106 | 0 | orgati = FALSE_; |
1107 | 0 | ii = ip1; |
1108 | 0 | sglb = -delsq2 / (d__[ii] + sq2); |
1109 | 0 | sgub = 0.f; |
1110 | 0 | a = c__ * delsq - z__[*i__] * z__[*i__] - z__[ip1] * z__[ip1]; |
1111 | 0 | b = z__[ip1] * z__[ip1] * delsq; |
1112 | 0 | if (a < 0.f) { |
1113 | 0 | tau2 = b * 2.f / (a - sqrt((r__1 = a * a + b * 4.f * c__, abs( |
1114 | 0 | r__1)))); |
1115 | 0 | } else { |
1116 | 0 | tau2 = -(a + sqrt((r__1 = a * a + b * 4.f * c__, abs(r__1)))) |
1117 | 0 | / (c__ * 2.f); |
1118 | 0 | } |
1119 | | |
1120 | | /* TAU2 now is an estimation of SIGMA^2 - D( IP1 )^2. The */ |
1121 | | /* following, however, is the corresponding estimation of */ |
1122 | | /* SIGMA - D( IP1 ). */ |
1123 | |
|
1124 | 0 | tau = tau2 / (d__[ip1] + sqrt((r__1 = d__[ip1] * d__[ip1] + tau2, |
1125 | 0 | abs(r__1)))); |
1126 | 0 | } |
1127 | |
|
1128 | 0 | *sigma = d__[ii] + tau; |
1129 | 0 | i__1 = *n; |
1130 | 0 | for (j = 1; j <= i__1; ++j) { |
1131 | 0 | work[j] = d__[j] + d__[ii] + tau; |
1132 | 0 | delta[j] = d__[j] - d__[ii] - tau; |
1133 | | /* L130: */ |
1134 | 0 | } |
1135 | 0 | iim1 = ii - 1; |
1136 | 0 | iip1 = ii + 1; |
1137 | | |
1138 | | /* Evaluate PSI and the derivative DPSI */ |
1139 | |
|
1140 | 0 | dpsi = 0.f; |
1141 | 0 | psi = 0.f; |
1142 | 0 | erretm = 0.f; |
1143 | 0 | i__1 = iim1; |
1144 | 0 | for (j = 1; j <= i__1; ++j) { |
1145 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1146 | 0 | psi += z__[j] * temp; |
1147 | 0 | dpsi += temp * temp; |
1148 | 0 | erretm += psi; |
1149 | | /* L150: */ |
1150 | 0 | } |
1151 | 0 | erretm = abs(erretm); |
1152 | | |
1153 | | /* Evaluate PHI and the derivative DPHI */ |
1154 | |
|
1155 | 0 | dphi = 0.f; |
1156 | 0 | phi = 0.f; |
1157 | 0 | i__1 = iip1; |
1158 | 0 | for (j = *n; j >= i__1; --j) { |
1159 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1160 | 0 | phi += z__[j] * temp; |
1161 | 0 | dphi += temp * temp; |
1162 | 0 | erretm += phi; |
1163 | | /* L160: */ |
1164 | 0 | } |
1165 | |
|
1166 | 0 | w = rhoinv + phi + psi; |
1167 | | |
1168 | | /* W is the value of the secular function with */ |
1169 | | /* its ii-th element removed. */ |
1170 | |
|
1171 | 0 | swtch3 = FALSE_; |
1172 | 0 | if (orgati) { |
1173 | 0 | if (w < 0.f) { |
1174 | 0 | swtch3 = TRUE_; |
1175 | 0 | } |
1176 | 0 | } else { |
1177 | 0 | if (w > 0.f) { |
1178 | 0 | swtch3 = TRUE_; |
1179 | 0 | } |
1180 | 0 | } |
1181 | 0 | if (ii == 1 || ii == *n) { |
1182 | 0 | swtch3 = FALSE_; |
1183 | 0 | } |
1184 | |
|
1185 | 0 | temp = z__[ii] / (work[ii] * delta[ii]); |
1186 | 0 | dw = dpsi + dphi + temp * temp; |
1187 | 0 | temp = z__[ii] * temp; |
1188 | 0 | w += temp; |
1189 | 0 | erretm = (phi - psi) * 8.f + erretm + rhoinv * 2.f + abs(temp) * 3.f; |
1190 | | /* $ + ABS( TAU2 )*DW */ |
1191 | | |
1192 | | /* Test for convergence */ |
1193 | |
|
1194 | 0 | if (abs(w) <= eps * erretm) { |
1195 | 0 | goto L240; |
1196 | 0 | } |
1197 | | |
1198 | 0 | if (w <= 0.f) { |
1199 | 0 | sglb = f2cmax(sglb,tau); |
1200 | 0 | } else { |
1201 | 0 | sgub = f2cmin(sgub,tau); |
1202 | 0 | } |
1203 | | |
1204 | | /* Calculate the new step */ |
1205 | |
|
1206 | 0 | ++niter; |
1207 | 0 | if (! swtch3) { |
1208 | 0 | dtipsq = work[ip1] * delta[ip1]; |
1209 | 0 | dtisq = work[*i__] * delta[*i__]; |
1210 | 0 | if (orgati) { |
1211 | | /* Computing 2nd power */ |
1212 | 0 | r__1 = z__[*i__] / dtisq; |
1213 | 0 | c__ = w - dtipsq * dw + delsq * (r__1 * r__1); |
1214 | 0 | } else { |
1215 | | /* Computing 2nd power */ |
1216 | 0 | r__1 = z__[ip1] / dtipsq; |
1217 | 0 | c__ = w - dtisq * dw - delsq * (r__1 * r__1); |
1218 | 0 | } |
1219 | 0 | a = (dtipsq + dtisq) * w - dtipsq * dtisq * dw; |
1220 | 0 | b = dtipsq * dtisq * w; |
1221 | 0 | if (c__ == 0.f) { |
1222 | 0 | if (a == 0.f) { |
1223 | 0 | if (orgati) { |
1224 | 0 | a = z__[*i__] * z__[*i__] + dtipsq * dtipsq * (dpsi + |
1225 | 0 | dphi); |
1226 | 0 | } else { |
1227 | 0 | a = z__[ip1] * z__[ip1] + dtisq * dtisq * (dpsi + |
1228 | 0 | dphi); |
1229 | 0 | } |
1230 | 0 | } |
1231 | 0 | eta = b / a; |
1232 | 0 | } else if (a <= 0.f) { |
1233 | 0 | eta = (a - sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1)))) / |
1234 | 0 | (c__ * 2.f); |
1235 | 0 | } else { |
1236 | 0 | eta = b * 2.f / (a + sqrt((r__1 = a * a - b * 4.f * c__, abs( |
1237 | 0 | r__1)))); |
1238 | 0 | } |
1239 | 0 | } else { |
1240 | | |
1241 | | /* Interpolation using THREE most relevant poles */ |
1242 | |
|
1243 | 0 | dtiim = work[iim1] * delta[iim1]; |
1244 | 0 | dtiip = work[iip1] * delta[iip1]; |
1245 | 0 | temp = rhoinv + psi + phi; |
1246 | 0 | if (orgati) { |
1247 | 0 | temp1 = z__[iim1] / dtiim; |
1248 | 0 | temp1 *= temp1; |
1249 | 0 | c__ = temp - dtiip * (dpsi + dphi) - (d__[iim1] - d__[iip1]) * |
1250 | 0 | (d__[iim1] + d__[iip1]) * temp1; |
1251 | 0 | zz[0] = z__[iim1] * z__[iim1]; |
1252 | 0 | if (dpsi < temp1) { |
1253 | 0 | zz[2] = dtiip * dtiip * dphi; |
1254 | 0 | } else { |
1255 | 0 | zz[2] = dtiip * dtiip * (dpsi - temp1 + dphi); |
1256 | 0 | } |
1257 | 0 | } else { |
1258 | 0 | temp1 = z__[iip1] / dtiip; |
1259 | 0 | temp1 *= temp1; |
1260 | 0 | c__ = temp - dtiim * (dpsi + dphi) - (d__[iip1] - d__[iim1]) * |
1261 | 0 | (d__[iim1] + d__[iip1]) * temp1; |
1262 | 0 | if (dphi < temp1) { |
1263 | 0 | zz[0] = dtiim * dtiim * dpsi; |
1264 | 0 | } else { |
1265 | 0 | zz[0] = dtiim * dtiim * (dpsi + (dphi - temp1)); |
1266 | 0 | } |
1267 | 0 | zz[2] = z__[iip1] * z__[iip1]; |
1268 | 0 | } |
1269 | 0 | zz[1] = z__[ii] * z__[ii]; |
1270 | 0 | dd[0] = dtiim; |
1271 | 0 | dd[1] = delta[ii] * work[ii]; |
1272 | 0 | dd[2] = dtiip; |
1273 | 0 | slaed6_(&niter, &orgati, &c__, dd, zz, &w, &eta, info); |
1274 | |
|
1275 | 0 | if (*info != 0) { |
1276 | | |
1277 | | /* If INFO is not 0, i.e., SLAED6 failed, switch back */ |
1278 | | /* to 2 pole interpolation. */ |
1279 | |
|
1280 | 0 | swtch3 = FALSE_; |
1281 | 0 | *info = 0; |
1282 | 0 | dtipsq = work[ip1] * delta[ip1]; |
1283 | 0 | dtisq = work[*i__] * delta[*i__]; |
1284 | 0 | if (orgati) { |
1285 | | /* Computing 2nd power */ |
1286 | 0 | r__1 = z__[*i__] / dtisq; |
1287 | 0 | c__ = w - dtipsq * dw + delsq * (r__1 * r__1); |
1288 | 0 | } else { |
1289 | | /* Computing 2nd power */ |
1290 | 0 | r__1 = z__[ip1] / dtipsq; |
1291 | 0 | c__ = w - dtisq * dw - delsq * (r__1 * r__1); |
1292 | 0 | } |
1293 | 0 | a = (dtipsq + dtisq) * w - dtipsq * dtisq * dw; |
1294 | 0 | b = dtipsq * dtisq * w; |
1295 | 0 | if (c__ == 0.f) { |
1296 | 0 | if (a == 0.f) { |
1297 | 0 | if (orgati) { |
1298 | 0 | a = z__[*i__] * z__[*i__] + dtipsq * dtipsq * ( |
1299 | 0 | dpsi + dphi); |
1300 | 0 | } else { |
1301 | 0 | a = z__[ip1] * z__[ip1] + dtisq * dtisq * (dpsi + |
1302 | 0 | dphi); |
1303 | 0 | } |
1304 | 0 | } |
1305 | 0 | eta = b / a; |
1306 | 0 | } else if (a <= 0.f) { |
1307 | 0 | eta = (a - sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1))) |
1308 | 0 | ) / (c__ * 2.f); |
1309 | 0 | } else { |
1310 | 0 | eta = b * 2.f / (a + sqrt((r__1 = a * a - b * 4.f * c__, |
1311 | 0 | abs(r__1)))); |
1312 | 0 | } |
1313 | 0 | } |
1314 | 0 | } |
1315 | | |
1316 | | /* Note, eta should be positive if w is negative, and */ |
1317 | | /* eta should be negative otherwise. However, */ |
1318 | | /* if for some reason caused by roundoff, eta*w > 0, */ |
1319 | | /* we simply use one Newton step instead. This way */ |
1320 | | /* will guarantee eta*w < 0. */ |
1321 | |
|
1322 | 0 | if (w * eta >= 0.f) { |
1323 | 0 | eta = -w / dw; |
1324 | 0 | } |
1325 | |
|
1326 | 0 | eta /= *sigma + sqrt(*sigma * *sigma + eta); |
1327 | 0 | temp = tau + eta; |
1328 | 0 | if (temp > sgub || temp < sglb) { |
1329 | 0 | if (w < 0.f) { |
1330 | 0 | eta = (sgub - tau) / 2.f; |
1331 | 0 | } else { |
1332 | 0 | eta = (sglb - tau) / 2.f; |
1333 | 0 | } |
1334 | 0 | if (geomavg) { |
1335 | 0 | if (w < 0.f) { |
1336 | 0 | if (tau > 0.f) { |
1337 | 0 | eta = sqrt(sgub * tau) - tau; |
1338 | 0 | } |
1339 | 0 | } else { |
1340 | 0 | if (sglb > 0.f) { |
1341 | 0 | eta = sqrt(sglb * tau) - tau; |
1342 | 0 | } |
1343 | 0 | } |
1344 | 0 | } |
1345 | 0 | } |
1346 | |
|
1347 | 0 | prew = w; |
1348 | |
|
1349 | 0 | tau += eta; |
1350 | 0 | *sigma += eta; |
1351 | |
|
1352 | 0 | i__1 = *n; |
1353 | 0 | for (j = 1; j <= i__1; ++j) { |
1354 | 0 | work[j] += eta; |
1355 | 0 | delta[j] -= eta; |
1356 | | /* L170: */ |
1357 | 0 | } |
1358 | | |
1359 | | /* Evaluate PSI and the derivative DPSI */ |
1360 | |
|
1361 | 0 | dpsi = 0.f; |
1362 | 0 | psi = 0.f; |
1363 | 0 | erretm = 0.f; |
1364 | 0 | i__1 = iim1; |
1365 | 0 | for (j = 1; j <= i__1; ++j) { |
1366 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1367 | 0 | psi += z__[j] * temp; |
1368 | 0 | dpsi += temp * temp; |
1369 | 0 | erretm += psi; |
1370 | | /* L180: */ |
1371 | 0 | } |
1372 | 0 | erretm = abs(erretm); |
1373 | | |
1374 | | /* Evaluate PHI and the derivative DPHI */ |
1375 | |
|
1376 | 0 | dphi = 0.f; |
1377 | 0 | phi = 0.f; |
1378 | 0 | i__1 = iip1; |
1379 | 0 | for (j = *n; j >= i__1; --j) { |
1380 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1381 | 0 | phi += z__[j] * temp; |
1382 | 0 | dphi += temp * temp; |
1383 | 0 | erretm += phi; |
1384 | | /* L190: */ |
1385 | 0 | } |
1386 | |
|
1387 | 0 | tau2 = work[ii] * delta[ii]; |
1388 | 0 | temp = z__[ii] / tau2; |
1389 | 0 | dw = dpsi + dphi + temp * temp; |
1390 | 0 | temp = z__[ii] * temp; |
1391 | 0 | w = rhoinv + phi + psi + temp; |
1392 | 0 | erretm = (phi - psi) * 8.f + erretm + rhoinv * 2.f + abs(temp) * 3.f; |
1393 | | /* $ + ABS( TAU2 )*DW */ |
1394 | |
|
1395 | 0 | swtch = FALSE_; |
1396 | 0 | if (orgati) { |
1397 | 0 | if (-w > abs(prew) / 10.f) { |
1398 | 0 | swtch = TRUE_; |
1399 | 0 | } |
1400 | 0 | } else { |
1401 | 0 | if (w > abs(prew) / 10.f) { |
1402 | 0 | swtch = TRUE_; |
1403 | 0 | } |
1404 | 0 | } |
1405 | | |
1406 | | /* Main loop to update the values of the array DELTA and WORK */ |
1407 | |
|
1408 | 0 | iter = niter + 1; |
1409 | |
|
1410 | 0 | for (niter = iter; niter <= 400; ++niter) { |
1411 | | |
1412 | | /* Test for convergence */ |
1413 | |
|
1414 | 0 | if (abs(w) <= eps * erretm) { |
1415 | | /* $ .OR. (SGUB-SGLB).LE.EIGHT*ABS(SGUB+SGLB) ) THEN */ |
1416 | 0 | goto L240; |
1417 | 0 | } |
1418 | | |
1419 | 0 | if (w <= 0.f) { |
1420 | 0 | sglb = f2cmax(sglb,tau); |
1421 | 0 | } else { |
1422 | 0 | sgub = f2cmin(sgub,tau); |
1423 | 0 | } |
1424 | | |
1425 | | /* Calculate the new step */ |
1426 | |
|
1427 | 0 | if (! swtch3) { |
1428 | 0 | dtipsq = work[ip1] * delta[ip1]; |
1429 | 0 | dtisq = work[*i__] * delta[*i__]; |
1430 | 0 | if (! swtch) { |
1431 | 0 | if (orgati) { |
1432 | | /* Computing 2nd power */ |
1433 | 0 | r__1 = z__[*i__] / dtisq; |
1434 | 0 | c__ = w - dtipsq * dw + delsq * (r__1 * r__1); |
1435 | 0 | } else { |
1436 | | /* Computing 2nd power */ |
1437 | 0 | r__1 = z__[ip1] / dtipsq; |
1438 | 0 | c__ = w - dtisq * dw - delsq * (r__1 * r__1); |
1439 | 0 | } |
1440 | 0 | } else { |
1441 | 0 | temp = z__[ii] / (work[ii] * delta[ii]); |
1442 | 0 | if (orgati) { |
1443 | 0 | dpsi += temp * temp; |
1444 | 0 | } else { |
1445 | 0 | dphi += temp * temp; |
1446 | 0 | } |
1447 | 0 | c__ = w - dtisq * dpsi - dtipsq * dphi; |
1448 | 0 | } |
1449 | 0 | a = (dtipsq + dtisq) * w - dtipsq * dtisq * dw; |
1450 | 0 | b = dtipsq * dtisq * w; |
1451 | 0 | if (c__ == 0.f) { |
1452 | 0 | if (a == 0.f) { |
1453 | 0 | if (! swtch) { |
1454 | 0 | if (orgati) { |
1455 | 0 | a = z__[*i__] * z__[*i__] + dtipsq * dtipsq * |
1456 | 0 | (dpsi + dphi); |
1457 | 0 | } else { |
1458 | 0 | a = z__[ip1] * z__[ip1] + dtisq * dtisq * ( |
1459 | 0 | dpsi + dphi); |
1460 | 0 | } |
1461 | 0 | } else { |
1462 | 0 | a = dtisq * dtisq * dpsi + dtipsq * dtipsq * dphi; |
1463 | 0 | } |
1464 | 0 | } |
1465 | 0 | eta = b / a; |
1466 | 0 | } else if (a <= 0.f) { |
1467 | 0 | eta = (a - sqrt((r__1 = a * a - b * 4.f * c__, abs(r__1))) |
1468 | 0 | ) / (c__ * 2.f); |
1469 | 0 | } else { |
1470 | 0 | eta = b * 2.f / (a + sqrt((r__1 = a * a - b * 4.f * c__, |
1471 | 0 | abs(r__1)))); |
1472 | 0 | } |
1473 | 0 | } else { |
1474 | | |
1475 | | /* Interpolation using THREE most relevant poles */ |
1476 | |
|
1477 | 0 | dtiim = work[iim1] * delta[iim1]; |
1478 | 0 | dtiip = work[iip1] * delta[iip1]; |
1479 | 0 | temp = rhoinv + psi + phi; |
1480 | 0 | if (swtch) { |
1481 | 0 | c__ = temp - dtiim * dpsi - dtiip * dphi; |
1482 | 0 | zz[0] = dtiim * dtiim * dpsi; |
1483 | 0 | zz[2] = dtiip * dtiip * dphi; |
1484 | 0 | } else { |
1485 | 0 | if (orgati) { |
1486 | 0 | temp1 = z__[iim1] / dtiim; |
1487 | 0 | temp1 *= temp1; |
1488 | 0 | temp2 = (d__[iim1] - d__[iip1]) * (d__[iim1] + d__[ |
1489 | 0 | iip1]) * temp1; |
1490 | 0 | c__ = temp - dtiip * (dpsi + dphi) - temp2; |
1491 | 0 | zz[0] = z__[iim1] * z__[iim1]; |
1492 | 0 | if (dpsi < temp1) { |
1493 | 0 | zz[2] = dtiip * dtiip * dphi; |
1494 | 0 | } else { |
1495 | 0 | zz[2] = dtiip * dtiip * (dpsi - temp1 + dphi); |
1496 | 0 | } |
1497 | 0 | } else { |
1498 | 0 | temp1 = z__[iip1] / dtiip; |
1499 | 0 | temp1 *= temp1; |
1500 | 0 | temp2 = (d__[iip1] - d__[iim1]) * (d__[iim1] + d__[ |
1501 | 0 | iip1]) * temp1; |
1502 | 0 | c__ = temp - dtiim * (dpsi + dphi) - temp2; |
1503 | 0 | if (dphi < temp1) { |
1504 | 0 | zz[0] = dtiim * dtiim * dpsi; |
1505 | 0 | } else { |
1506 | 0 | zz[0] = dtiim * dtiim * (dpsi + (dphi - temp1)); |
1507 | 0 | } |
1508 | 0 | zz[2] = z__[iip1] * z__[iip1]; |
1509 | 0 | } |
1510 | 0 | } |
1511 | 0 | dd[0] = dtiim; |
1512 | 0 | dd[1] = delta[ii] * work[ii]; |
1513 | 0 | dd[2] = dtiip; |
1514 | 0 | slaed6_(&niter, &orgati, &c__, dd, zz, &w, &eta, info); |
1515 | |
|
1516 | 0 | if (*info != 0) { |
1517 | | |
1518 | | /* If INFO is not 0, i.e., SLAED6 failed, switch */ |
1519 | | /* back to two pole interpolation */ |
1520 | |
|
1521 | 0 | swtch3 = FALSE_; |
1522 | 0 | *info = 0; |
1523 | 0 | dtipsq = work[ip1] * delta[ip1]; |
1524 | 0 | dtisq = work[*i__] * delta[*i__]; |
1525 | 0 | if (! swtch) { |
1526 | 0 | if (orgati) { |
1527 | | /* Computing 2nd power */ |
1528 | 0 | r__1 = z__[*i__] / dtisq; |
1529 | 0 | c__ = w - dtipsq * dw + delsq * (r__1 * r__1); |
1530 | 0 | } else { |
1531 | | /* Computing 2nd power */ |
1532 | 0 | r__1 = z__[ip1] / dtipsq; |
1533 | 0 | c__ = w - dtisq * dw - delsq * (r__1 * r__1); |
1534 | 0 | } |
1535 | 0 | } else { |
1536 | 0 | temp = z__[ii] / (work[ii] * delta[ii]); |
1537 | 0 | if (orgati) { |
1538 | 0 | dpsi += temp * temp; |
1539 | 0 | } else { |
1540 | 0 | dphi += temp * temp; |
1541 | 0 | } |
1542 | 0 | c__ = w - dtisq * dpsi - dtipsq * dphi; |
1543 | 0 | } |
1544 | 0 | a = (dtipsq + dtisq) * w - dtipsq * dtisq * dw; |
1545 | 0 | b = dtipsq * dtisq * w; |
1546 | 0 | if (c__ == 0.f) { |
1547 | 0 | if (a == 0.f) { |
1548 | 0 | if (! swtch) { |
1549 | 0 | if (orgati) { |
1550 | 0 | a = z__[*i__] * z__[*i__] + dtipsq * |
1551 | 0 | dtipsq * (dpsi + dphi); |
1552 | 0 | } else { |
1553 | 0 | a = z__[ip1] * z__[ip1] + dtisq * dtisq * |
1554 | 0 | (dpsi + dphi); |
1555 | 0 | } |
1556 | 0 | } else { |
1557 | 0 | a = dtisq * dtisq * dpsi + dtipsq * dtipsq * |
1558 | 0 | dphi; |
1559 | 0 | } |
1560 | 0 | } |
1561 | 0 | eta = b / a; |
1562 | 0 | } else if (a <= 0.f) { |
1563 | 0 | eta = (a - sqrt((r__1 = a * a - b * 4.f * c__, abs( |
1564 | 0 | r__1)))) / (c__ * 2.f); |
1565 | 0 | } else { |
1566 | 0 | eta = b * 2.f / (a + sqrt((r__1 = a * a - b * 4.f * |
1567 | 0 | c__, abs(r__1)))); |
1568 | 0 | } |
1569 | 0 | } |
1570 | 0 | } |
1571 | | |
1572 | | /* Note, eta should be positive if w is negative, and */ |
1573 | | /* eta should be negative otherwise. However, */ |
1574 | | /* if for some reason caused by roundoff, eta*w > 0, */ |
1575 | | /* we simply use one Newton step instead. This way */ |
1576 | | /* will guarantee eta*w < 0. */ |
1577 | |
|
1578 | 0 | if (w * eta >= 0.f) { |
1579 | 0 | eta = -w / dw; |
1580 | 0 | } |
1581 | |
|
1582 | 0 | eta /= *sigma + sqrt(*sigma * *sigma + eta); |
1583 | 0 | temp = tau + eta; |
1584 | 0 | if (temp > sgub || temp < sglb) { |
1585 | 0 | if (w < 0.f) { |
1586 | 0 | eta = (sgub - tau) / 2.f; |
1587 | 0 | } else { |
1588 | 0 | eta = (sglb - tau) / 2.f; |
1589 | 0 | } |
1590 | 0 | if (geomavg) { |
1591 | 0 | if (w < 0.f) { |
1592 | 0 | if (tau > 0.f) { |
1593 | 0 | eta = sqrt(sgub * tau) - tau; |
1594 | 0 | } |
1595 | 0 | } else { |
1596 | 0 | if (sglb > 0.f) { |
1597 | 0 | eta = sqrt(sglb * tau) - tau; |
1598 | 0 | } |
1599 | 0 | } |
1600 | 0 | } |
1601 | 0 | } |
1602 | |
|
1603 | 0 | prew = w; |
1604 | |
|
1605 | 0 | tau += eta; |
1606 | 0 | *sigma += eta; |
1607 | |
|
1608 | 0 | i__1 = *n; |
1609 | 0 | for (j = 1; j <= i__1; ++j) { |
1610 | 0 | work[j] += eta; |
1611 | 0 | delta[j] -= eta; |
1612 | | /* L200: */ |
1613 | 0 | } |
1614 | | |
1615 | | /* Evaluate PSI and the derivative DPSI */ |
1616 | |
|
1617 | 0 | dpsi = 0.f; |
1618 | 0 | psi = 0.f; |
1619 | 0 | erretm = 0.f; |
1620 | 0 | i__1 = iim1; |
1621 | 0 | for (j = 1; j <= i__1; ++j) { |
1622 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1623 | 0 | psi += z__[j] * temp; |
1624 | 0 | dpsi += temp * temp; |
1625 | 0 | erretm += psi; |
1626 | | /* L210: */ |
1627 | 0 | } |
1628 | 0 | erretm = abs(erretm); |
1629 | | |
1630 | | /* Evaluate PHI and the derivative DPHI */ |
1631 | |
|
1632 | 0 | dphi = 0.f; |
1633 | 0 | phi = 0.f; |
1634 | 0 | i__1 = iip1; |
1635 | 0 | for (j = *n; j >= i__1; --j) { |
1636 | 0 | temp = z__[j] / (work[j] * delta[j]); |
1637 | 0 | phi += z__[j] * temp; |
1638 | 0 | dphi += temp * temp; |
1639 | 0 | erretm += phi; |
1640 | | /* L220: */ |
1641 | 0 | } |
1642 | |
|
1643 | 0 | tau2 = work[ii] * delta[ii]; |
1644 | 0 | temp = z__[ii] / tau2; |
1645 | 0 | dw = dpsi + dphi + temp * temp; |
1646 | 0 | temp = z__[ii] * temp; |
1647 | 0 | w = rhoinv + phi + psi + temp; |
1648 | 0 | erretm = (phi - psi) * 8.f + erretm + rhoinv * 2.f + abs(temp) * |
1649 | 0 | 3.f; |
1650 | | /* $ + ABS( TAU2 )*DW */ |
1651 | |
|
1652 | 0 | if (w * prew > 0.f && abs(w) > abs(prew) / 10.f) { |
1653 | 0 | swtch = ! swtch; |
1654 | 0 | } |
1655 | | |
1656 | | /* L230: */ |
1657 | 0 | } |
1658 | | |
1659 | | /* Return with INFO = 1, NITER = MAXIT and not converged */ |
1660 | | |
1661 | 0 | *info = 1; |
1662 | |
|
1663 | 0 | } |
1664 | | |
1665 | 0 | L240: |
1666 | 0 | return; |
1667 | | |
1668 | | /* End of SLASD4 */ |
1669 | |
|
1670 | 0 | } /* slasd4_ */ |
1671 | | |