1 /*
2  * ====================================================
3  * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
4  *
5  * Developed at SunPro, a Sun Microsystems, Inc. business.
6  * Permission to use, copy, modify, and distribute this
7  * software is freely granted, provided that this notice
8  * is preserved.
9  * ====================================================
10  */
11 
12 /* Long double expansions are
13   Copyright (C) 2001 Stephen L. Moshier <moshier@na-net.ornl.gov>
14   and are incorporated herein by permission of the author.  The author
15   reserves the right to distribute this material elsewhere under different
16   copying permissions.  These modifications are distributed here under
17   the following terms:
18 
19     This library is free software; you can redistribute it and/or
20     modify it under the terms of the GNU Lesser General Public
21     License as published by the Free Software Foundation; either
22     version 2.1 of the License, or (at your option) any later version.
23 
24     This library is distributed in the hope that it will be useful,
25     but WITHOUT ANY WARRANTY; without even the implied warranty of
26     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
27     Lesser General Public License for more details.
28 
29     You should have received a copy of the GNU Lesser General Public
30     License along with this library; if not, see
31     <https://www.gnu.org/licenses/>.  */
32 
33 /* __ieee754_j0(x), __ieee754_y0(x)
34  * Bessel function of the first and second kinds of order zero.
35  * Method -- j0(x):
36  *	1. For tiny x, we use j0(x) = 1 - x^2/4 + x^4/64 - ...
37  *	2. Reduce x to |x| since j0(x)=j0(-x),  and
38  *	   for x in (0,2)
39  *		j0(x) = 1 - z/4 + z^2*R0/S0,  where z = x*x;
40  *	   for x in (2,inf)
41  *		j0(x) = sqrt(2/(pi*x))*(p0(x)*cos(x0)-q0(x)*sin(x0))
42  *	   where x0 = x-pi/4. It is better to compute sin(x0),cos(x0)
43  *	   as follow:
44  *		cos(x0) = cos(x)cos(pi/4)+sin(x)sin(pi/4)
45  *			= 1/sqrt(2) * (cos(x) + sin(x))
46  *		sin(x0) = sin(x)cos(pi/4)-cos(x)sin(pi/4)
47  *			= 1/sqrt(2) * (sin(x) - cos(x))
48  *	   (To avoid cancellation, use
49  *		sin(x) +- cos(x) = -cos(2x)/(sin(x) -+ cos(x))
50  *	    to compute the worse one.)
51  *
52  *	3 Special cases
53  *		j0(nan)= nan
54  *		j0(0) = 1
55  *		j0(inf) = 0
56  *
57  * Method -- y0(x):
58  *	1. For x<2.
59  *	   Since
60  *		y0(x) = 2/pi*(j0(x)*(ln(x/2)+Euler) + x^2/4 - ...)
61  *	   therefore y0(x)-2/pi*j0(x)*ln(x) is an even function.
62  *	   We use the following function to approximate y0,
63  *		y0(x) = U(z)/V(z) + (2/pi)*(j0(x)*ln(x)), z= x^2
64  *
65  *	   Note: For tiny x, U/V = u0 and j0(x)~1, hence
66  *		y0(tiny) = u0 + (2/pi)*ln(tiny), (choose tiny<2**-27)
67  *	2. For x>=2.
68  *		y0(x) = sqrt(2/(pi*x))*(p0(x)*cos(x0)+q0(x)*sin(x0))
69  *	   where x0 = x-pi/4. It is better to compute sin(x0),cos(x0)
70  *	   by the method mentioned above.
71  *	3. Special cases: y0(0)=-inf, y0(x<0)=NaN, y0(inf)=0.
72  */
73 
74 #include <math.h>
75 #include <math-barriers.h>
76 #include <math_private.h>
77 #include <libm-alias-finite.h>
78 
79 static long double pzero (long double), qzero (long double);
80 
81 static const long double
82   huge = 1e4930L,
83   one = 1.0L,
84   invsqrtpi = 5.6418958354775628694807945156077258584405e-1L,
85   tpi = 6.3661977236758134307553505349005744813784e-1L,
86 
87   /* J0(x) = 1 - x^2 / 4 + x^4 R0(x^2) / S0(x^2)
88      0 <= x <= 2
89      peak relative error 1.41e-22 */
90   R[5] = {
91   4.287176872744686992880841716723478740566E7L,
92   -6.652058897474241627570911531740907185772E5L,
93   7.011848381719789863458364584613651091175E3L,
94   -3.168040850193372408702135490809516253693E1L,
95   6.030778552661102450545394348845599300939E-2L,
96 },
97 
98  S[4] = {
99    2.743793198556599677955266341699130654342E9L,
100    3.364330079384816249840086842058954076201E7L,
101    1.924119649412510777584684927494642526573E5L,
102    6.239282256012734914211715620088714856494E2L,
103    /*   1.000000000000000000000000000000000000000E0L,*/
104 };
105 
106 static const long double zero = 0.0;
107 
108 long double
__ieee754_j0l(long double x)109 __ieee754_j0l (long double x)
110 {
111   long double z, s, c, ss, cc, r, u, v;
112   int32_t ix;
113   uint32_t se;
114 
115   GET_LDOUBLE_EXP (se, x);
116   ix = se & 0x7fff;
117   if (__glibc_unlikely (ix >= 0x7fff))
118     return one / (x * x);
119   x = fabsl (x);
120   if (ix >= 0x4000)		/* |x| >= 2.0 */
121     {
122       __sincosl (x, &s, &c);
123       ss = s - c;
124       cc = s + c;
125       if (ix < 0x7ffe)
126 	{			/* make sure x+x not overflow */
127 	  z = -__cosl (x + x);
128 	  if ((s * c) < zero)
129 	    cc = z / ss;
130 	  else
131 	    ss = z / cc;
132 	}
133       /*
134        * j0(x) = 1/sqrt(pi) * (P(0,x)*cc - Q(0,x)*ss) / sqrt(x)
135        * y0(x) = 1/sqrt(pi) * (P(0,x)*ss + Q(0,x)*cc) / sqrt(x)
136        */
137       if (__glibc_unlikely (ix > 0x408e))      	/* 2^143 */
138 	z = (invsqrtpi * cc) / sqrtl (x);
139       else
140 	{
141 	  u = pzero (x);
142 	  v = qzero (x);
143 	  z = invsqrtpi * (u * cc - v * ss) / sqrtl (x);
144 	}
145       return z;
146     }
147   if (__glibc_unlikely (ix < 0x3fef))       /* |x| < 2**-16 */
148     {
149       /* raise inexact if x != 0 */
150       math_force_eval (huge + x);
151       if (ix < 0x3fde) /* |x| < 2^-33 */
152 	return one;
153       else
154 	return one - 0.25 * x * x;
155     }
156   z = x * x;
157   r = z * (R[0] + z * (R[1] + z * (R[2] + z * (R[3] + z * R[4]))));
158   s = S[0] + z * (S[1] + z * (S[2] + z * (S[3] + z)));
159   if (ix < 0x3fff)
160     {				/* |x| < 1.00 */
161       return (one - 0.25 * z + z * (r / s));
162     }
163   else
164     {
165       u = 0.5 * x;
166       return ((one + u) * (one - u) + z * (r / s));
167     }
168 }
169 libm_alias_finite (__ieee754_j0l, __j0l)
170 
171 
172 /* y0(x) = 2/pi ln(x) J0(x) + U(x^2)/V(x^2)
173    0 < x <= 2
174    peak relative error 1.7e-21 */
175 static const long double
176 U[6] = {
177   -1.054912306975785573710813351985351350861E10L,
178   2.520192609749295139432773849576523636127E10L,
179   -1.856426071075602001239955451329519093395E9L,
180   4.079209129698891442683267466276785956784E7L,
181   -3.440684087134286610316661166492641011539E5L,
182   1.005524356159130626192144663414848383774E3L,
183 };
184 static const long double
185 V[5] = {
186   1.429337283720789610137291929228082613676E11L,
187   2.492593075325119157558811370165695013002E9L,
188   2.186077620785925464237324417623665138376E7L,
189   1.238407896366385175196515057064384929222E5L,
190   4.693924035211032457494368947123233101664E2L,
191   /*  1.000000000000000000000000000000000000000E0L */
192 };
193 
194 long double
__ieee754_y0l(long double x)195 __ieee754_y0l (long double x)
196 {
197   long double z, s, c, ss, cc, u, v;
198   int32_t ix;
199   uint32_t se, i0, i1;
200 
201   GET_LDOUBLE_WORDS (se, i0, i1, x);
202   ix = se & 0x7fff;
203   /* Y0(NaN) is NaN, y0(-inf) is Nan, y0(inf) is 0  */
204   if (__glibc_unlikely (se & 0x8000))
205     return zero / (zero * x);
206   if (__glibc_unlikely (ix >= 0x7fff))
207     return one / (x + x * x);
208   if (__glibc_unlikely ((i0 | i1) == 0))
209     return -HUGE_VALL + x;  /* -inf and overflow exception.  */
210   if (ix >= 0x4000)
211     {				/* |x| >= 2.0 */
212 
213       /* y0(x) = sqrt(2/(pi*x))*(p0(x)*sin(x0)+q0(x)*cos(x0))
214        * where x0 = x-pi/4
215        *      Better formula:
216        *              cos(x0) = cos(x)cos(pi/4)+sin(x)sin(pi/4)
217        *                      =  1/sqrt(2) * (sin(x) + cos(x))
218        *              sin(x0) = sin(x)cos(3pi/4)-cos(x)sin(3pi/4)
219        *                      =  1/sqrt(2) * (sin(x) - cos(x))
220        * To avoid cancellation, use
221        *              sin(x) +- cos(x) = -cos(2x)/(sin(x) -+ cos(x))
222        * to compute the worse one.
223        */
224       __sincosl (x, &s, &c);
225       ss = s - c;
226       cc = s + c;
227       /*
228        * j0(x) = 1/sqrt(pi) * (P(0,x)*cc - Q(0,x)*ss) / sqrt(x)
229        * y0(x) = 1/sqrt(pi) * (P(0,x)*ss + Q(0,x)*cc) / sqrt(x)
230        */
231       if (ix < 0x7ffe)
232 	{			/* make sure x+x not overflow */
233 	  z = -__cosl (x + x);
234 	  if ((s * c) < zero)
235 	    cc = z / ss;
236 	  else
237 	    ss = z / cc;
238 	}
239       if (__glibc_unlikely (ix > 0x408e))      	/* 2^143 */
240 	z = (invsqrtpi * ss) / sqrtl (x);
241       else
242 	{
243 	  u = pzero (x);
244 	  v = qzero (x);
245 	  z = invsqrtpi * (u * ss + v * cc) / sqrtl (x);
246 	}
247       return z;
248     }
249   if (__glibc_unlikely (ix <= 0x3fde))       /* x < 2^-33 */
250     {
251       z = -7.380429510868722527629822444004602747322E-2L
252 	+ tpi * __ieee754_logl (x);
253       return z;
254     }
255   z = x * x;
256   u = U[0] + z * (U[1] + z * (U[2] + z * (U[3] + z * (U[4] + z * U[5]))));
257   v = V[0] + z * (V[1] + z * (V[2] + z * (V[3] + z * (V[4] + z))));
258   return (u / v + tpi * (__ieee754_j0l (x) * __ieee754_logl (x)));
259 }
260 libm_alias_finite (__ieee754_y0l, __y0l)
261 
262 /* The asymptotic expansions of pzero is
263  *	1 - 9/128 s^2 + 11025/98304 s^4 - ...,	where s = 1/x.
264  * For x >= 2, We approximate pzero by
265  *	pzero(x) = 1 + s^2 R(s^2) / S(s^2)
266  */
267 static const long double pR8[7] = {
268   /* 8 <= x <= inf
269      Peak relative error 4.62 */
270   -4.094398895124198016684337960227780260127E-9L,
271   -8.929643669432412640061946338524096893089E-7L,
272   -6.281267456906136703868258380673108109256E-5L,
273   -1.736902783620362966354814353559382399665E-3L,
274   -1.831506216290984960532230842266070146847E-2L,
275   -5.827178869301452892963280214772398135283E-2L,
276   -2.087563267939546435460286895807046616992E-2L,
277 };
278 static const long double pS8[6] = {
279   5.823145095287749230197031108839653988393E-8L,
280   1.279281986035060320477759999428992730280E-5L,
281   9.132668954726626677174825517150228961304E-4L,
282   2.606019379433060585351880541545146252534E-2L,
283   2.956262215119520464228467583516287175244E-1L,
284   1.149498145388256448535563278632697465675E0L,
285   /* 1.000000000000000000000000000000000000000E0L, */
286 };
287 
288 static const long double pR5[7] = {
289   /* 4.54541015625 <= x <= 8
290      Peak relative error 6.51E-22 */
291   -2.041226787870240954326915847282179737987E-7L,
292   -2.255373879859413325570636768224534428156E-5L,
293   -7.957485746440825353553537274569102059990E-4L,
294   -1.093205102486816696940149222095559439425E-2L,
295   -5.657957849316537477657603125260701114646E-2L,
296   -8.641175552716402616180994954177818461588E-2L,
297   -1.354654710097134007437166939230619726157E-2L,
298 };
299 static const long double pS5[6] = {
300   2.903078099681108697057258628212823545290E-6L,
301   3.253948449946735405975737677123673867321E-4L,
302   1.181269751723085006534147920481582279979E-2L,
303   1.719212057790143888884745200257619469363E-1L,
304   1.006306498779212467670654535430694221924E0L,
305   2.069568808688074324555596301126375951502E0L,
306   /* 1.000000000000000000000000000000000000000E0L, */
307 };
308 
309 static const long double pR3[7] = {
310   /* 2.85711669921875 <= x <= 4.54541015625
311      peak relative error 5.25e-21 */
312   -5.755732156848468345557663552240816066802E-6L,
313   -3.703675625855715998827966962258113034767E-4L,
314   -7.390893350679637611641350096842846433236E-3L,
315   -5.571922144490038765024591058478043873253E-2L,
316   -1.531290690378157869291151002472627396088E-1L,
317   -1.193350853469302941921647487062620011042E-1L,
318   -8.567802507331578894302991505331963782905E-3L,
319 };
320 static const long double pS3[6] = {
321   8.185931139070086158103309281525036712419E-5L,
322   5.398016943778891093520574483111255476787E-3L,
323   1.130589193590489566669164765853409621081E-1L,
324   9.358652328786413274673192987670237145071E-1L,
325   3.091711512598349056276917907005098085273E0L,
326   3.594602474737921977972586821673124231111E0L,
327   /* 1.000000000000000000000000000000000000000E0L, */
328 };
329 
330 static const long double pR2[7] = {
331   /* 2 <= x <= 2.85711669921875
332      peak relative error 2.64e-21 */
333   -1.219525235804532014243621104365384992623E-4L,
334   -4.838597135805578919601088680065298763049E-3L,
335   -5.732223181683569266223306197751407418301E-2L,
336   -2.472947430526425064982909699406646503758E-1L,
337   -3.753373645974077960207588073975976327695E-1L,
338   -1.556241316844728872406672349347137975495E-1L,
339   -5.355423239526452209595316733635519506958E-3L,
340 };
341 static const long double pS2[6] = {
342   1.734442793664291412489066256138894953823E-3L,
343   7.158111826468626405416300895617986926008E-2L,
344   9.153839713992138340197264669867993552641E-1L,
345   4.539209519433011393525841956702487797582E0L,
346   8.868932430625331650266067101752626253644E0L,
347   6.067161890196324146320763844772857713502E0L,
348   /* 1.000000000000000000000000000000000000000E0L, */
349 };
350 
351 static long double
pzero(long double x)352 pzero (long double x)
353 {
354   const long double *p, *q;
355   long double z, r, s;
356   int32_t ix;
357   uint32_t se, i0, i1;
358 
359   GET_LDOUBLE_WORDS (se, i0, i1, x);
360   ix = se & 0x7fff;
361   /* ix >= 0x4000 for all calls to this function.  */
362   if (ix >= 0x4002)
363     {
364       p = pR8;
365       q = pS8;
366     }				/* x >= 8 */
367   else
368     {
369       i1 = (ix << 16) | (i0 >> 16);
370       if (i1 >= 0x40019174)	/* x >= 4.54541015625 */
371 	{
372 	  p = pR5;
373 	  q = pS5;
374 	}
375       else if (i1 >= 0x4000b6db)	/* x >= 2.85711669921875 */
376 	{
377 	  p = pR3;
378 	  q = pS3;
379 	}
380       else	/* x >= 2 */
381 	{
382 	  p = pR2;
383 	  q = pS2;
384 	}
385     }
386   z = one / (x * x);
387   r =
388     p[0] + z * (p[1] +
389 		z * (p[2] + z * (p[3] + z * (p[4] + z * (p[5] + z * p[6])))));
390   s =
391     q[0] + z * (q[1] + z * (q[2] + z * (q[3] + z * (q[4] + z * (q[5] + z)))));
392   return (one + z * r / s);
393 }
394 
395 
396 /* For x >= 8, the asymptotic expansions of qzero is
397  *	-1/8 s + 75/1024 s^3 - ..., where s = 1/x.
398  * We approximate qzero by
399  *	qzero(x) = s*(-.125 + R(s^2) / S(s^2))
400  */
401 static const long double qR8[7] = {
402   /* 8 <= x <= inf
403      peak relative error 2.23e-21 */
404   3.001267180483191397885272640777189348008E-10L,
405   8.693186311430836495238494289942413810121E-8L,
406   8.496875536711266039522937037850596580686E-6L,
407   3.482702869915288984296602449543513958409E-4L,
408   6.036378380706107692863811938221290851352E-3L,
409   3.881970028476167836382607922840452192636E-2L,
410   6.132191514516237371140841765561219149638E-2L,
411 };
412 static const long double qS8[7] = {
413   4.097730123753051126914971174076227600212E-9L,
414   1.199615869122646109596153392152131139306E-6L,
415   1.196337580514532207793107149088168946451E-4L,
416   5.099074440112045094341500497767181211104E-3L,
417   9.577420799632372483249761659674764460583E-2L,
418   7.385243015344292267061953461563695918646E-1L,
419   1.917266424391428937962682301561699055943E0L,
420   /* 1.000000000000000000000000000000000000000E0L, */
421 };
422 
423 static const long double qR5[7] = {
424   /* 4.54541015625 <= x <= 8
425      peak relative error 1.03e-21 */
426   3.406256556438974327309660241748106352137E-8L,
427   4.855492710552705436943630087976121021980E-6L,
428   2.301011739663737780613356017352912281980E-4L,
429   4.500470249273129953870234803596619899226E-3L,
430   3.651376459725695502726921248173637054828E-2L,
431   1.071578819056574524416060138514508609805E-1L,
432   7.458950172851611673015774675225656063757E-2L,
433 };
434 static const long double qS5[7] = {
435   4.650675622764245276538207123618745150785E-7L,
436   6.773573292521412265840260065635377164455E-5L,
437   3.340711249876192721980146877577806687714E-3L,
438   7.036218046856839214741678375536970613501E-2L,
439   6.569599559163872573895171876511377891143E-1L,
440   2.557525022583599204591036677199171155186E0L,
441   3.457237396120935674982927714210361269133E0L,
442   /* 1.000000000000000000000000000000000000000E0L,*/
443 };
444 
445 static const long double qR3[7] = {
446   /* 2.85711669921875 <= x <= 4.54541015625
447      peak relative error 5.24e-21 */
448   1.749459596550816915639829017724249805242E-6L,
449   1.446252487543383683621692672078376929437E-4L,
450   3.842084087362410664036704812125005761859E-3L,
451   4.066369994699462547896426554180954233581E-2L,
452   1.721093619117980251295234795188992722447E-1L,
453   2.538595333972857367655146949093055405072E-1L,
454   8.560591367256769038905328596020118877936E-2L,
455 };
456 static const long double qS3[7] = {
457   2.388596091707517488372313710647510488042E-5L,
458   2.048679968058758616370095132104333998147E-3L,
459   5.824663198201417760864458765259945181513E-2L,
460   6.953906394693328750931617748038994763958E-1L,
461   3.638186936390881159685868764832961092476E0L,
462   7.900169524705757837298990558459547842607E0L,
463   5.992718532451026507552820701127504582907E0L,
464   /* 1.000000000000000000000000000000000000000E0L, */
465 };
466 
467 static const long double qR2[7] = {
468   /* 2 <= x <= 2.85711669921875
469      peak relative error 1.58e-21  */
470   6.306524405520048545426928892276696949540E-5L,
471   3.209606155709930950935893996591576624054E-3L,
472   5.027828775702022732912321378866797059604E-2L,
473   3.012705561838718956481911477587757845163E-1L,
474   6.960544893905752937420734884995688523815E-1L,
475   5.431871999743531634887107835372232030655E-1L,
476   9.447736151202905471899259026430157211949E-2L,
477 };
478 static const long double qS2[7] = {
479   8.610579901936193494609755345106129102676E-4L,
480   4.649054352710496997203474853066665869047E-2L,
481   8.104282924459837407218042945106320388339E-1L,
482   5.807730930825886427048038146088828206852E0L,
483   1.795310145936848873627710102199881642939E1L,
484   2.281313316875375733663657188888110605044E1L,
485   1.011242067883822301487154844458322200143E1L,
486   /* 1.000000000000000000000000000000000000000E0L, */
487 };
488 
489 static long double
qzero(long double x)490 qzero (long double x)
491 {
492   const long double *p, *q;
493   long double s, r, z;
494   int32_t ix;
495   uint32_t se, i0, i1;
496 
497   GET_LDOUBLE_WORDS (se, i0, i1, x);
498   ix = se & 0x7fff;
499   /* ix >= 0x4000 for all calls to this function.  */
500   if (ix >= 0x4002)		/* x >= 8 */
501     {
502       p = qR8;
503       q = qS8;
504     }
505   else
506     {
507       i1 = (ix << 16) | (i0 >> 16);
508       if (i1 >= 0x40019174)	/* x >= 4.54541015625 */
509 	{
510 	  p = qR5;
511 	  q = qS5;
512 	}
513       else if (i1 >= 0x4000b6db)	/* x >= 2.85711669921875 */
514 	{
515 	  p = qR3;
516 	  q = qS3;
517 	}
518       else	/* x >= 2 */
519 	{
520 	  p = qR2;
521 	  q = qS2;
522 	}
523     }
524   z = one / (x * x);
525   r =
526     p[0] + z * (p[1] +
527 		z * (p[2] + z * (p[3] + z * (p[4] + z * (p[5] + z * p[6])))));
528   s =
529     q[0] + z * (q[1] +
530 		z * (q[2] +
531 		     z * (q[3] + z * (q[4] + z * (q[5] + z * (q[6] + z))))));
532   return (-.125 + z * r / s) / x;
533 }
534