Home | History | Annotate | Line # | Download | only in noieee_src
n_j0.c revision 1.2
      1 /*	$NetBSD: n_j0.c,v 1.2 1997/10/20 14:12:44 ragge Exp $	*/
      2 /*-
      3  * Copyright (c) 1992, 1993
      4  *	The Regents of the University of California.  All rights reserved.
      5  *
      6  * Redistribution and use in source and binary forms, with or without
      7  * modification, are permitted provided that the following conditions
      8  * are met:
      9  * 1. Redistributions of source code must retain the above copyright
     10  *    notice, this list of conditions and the following disclaimer.
     11  * 2. Redistributions in binary form must reproduce the above copyright
     12  *    notice, this list of conditions and the following disclaimer in the
     13  *    documentation and/or other materials provided with the distribution.
     14  * 3. All advertising materials mentioning features or use of this software
     15  *    must display the following acknowledgement:
     16  *	This product includes software developed by the University of
     17  *	California, Berkeley and its contributors.
     18  * 4. Neither the name of the University nor the names of its contributors
     19  *    may be used to endorse or promote products derived from this software
     20  *    without specific prior written permission.
     21  *
     22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     32  * SUCH DAMAGE.
     33  */
     34 
     35 #ifndef lint
     36 #if 0
     37 static char sccsid[] = "@(#)j0.c	8.2 (Berkeley) 11/30/93";
     38 #endif
     39 #endif /* not lint */
     40 
     41 /*
     42  * 16 December 1992
     43  * Minor modifications by Peter McIlroy to adapt non-IEEE architecture.
     44  */
     45 
     46 /*
     47  * ====================================================
     48  * Copyright (C) 1992 by Sun Microsystems, Inc.
     49  *
     50  * Developed at SunPro, a Sun Microsystems, Inc. business.
     51  * Permission to use, copy, modify, and distribute this
     52  * software is freely granted, provided that this notice
     53  * is preserved.
     54  * ====================================================
     55  *
     56  * ******************* WARNING ********************
     57  * This is an alpha version of SunPro's FDLIBM (Freely
     58  * Distributable Math Library) for IEEE double precision
     59  * arithmetic. FDLIBM is a basic math library written
     60  * in C that runs on machines that conform to IEEE
     61  * Standard 754/854. This alpha version is distributed
     62  * for testing purpose. Those who use this software
     63  * should report any bugs to
     64  *
     65  *		fdlibm-comments (at) sunpro.eng.sun.com
     66  *
     67  * -- K.C. Ng, Oct 12, 1992
     68  * ************************************************
     69  */
     70 
     71 /* double j0(double x), y0(double x)
     72  * Bessel function of the first and second kinds of order zero.
     73  * Method -- j0(x):
     74  *	1. For tiny x, we use j0(x) = 1 - x^2/4 + x^4/64 - ...
     75  *	2. Reduce x to |x| since j0(x)=j0(-x),  and
     76  *	   for x in (0,2)
     77  *		j0(x) = 1-z/4+ z^2*R0/S0,  where z = x*x;
     78  *	   (precision:  |j0-1+z/4-z^2R0/S0 |<2**-63.67 )
     79  *	   for x in (2,inf)
     80  * 		j0(x) = sqrt(2/(pi*x))*(p0(x)*cos(x0)-q0(x)*sin(x0))
     81  * 	   where x0 = x-pi/4. It is better to compute sin(x0),cos(x0)
     82  *	   as follow:
     83  *		cos(x0) = cos(x)cos(pi/4)+sin(x)sin(pi/4)
     84  *			= 1/sqrt(2) * (cos(x) + sin(x))
     85  *		sin(x0) = sin(x)cos(pi/4)-cos(x)sin(pi/4)
     86  *			= 1/sqrt(2) * (sin(x) - cos(x))
     87  * 	   (To avoid cancellation, use
     88  *		sin(x) +- cos(x) = -cos(2x)/(sin(x) -+ cos(x))
     89  * 	    to compute the worse one.)
     90  *
     91  *	3 Special cases
     92  *		j0(nan)= nan
     93  *		j0(0) = 1
     94  *		j0(inf) = 0
     95  *
     96  * Method -- y0(x):
     97  *	1. For x<2.
     98  *	   Since
     99  *		y0(x) = 2/pi*(j0(x)*(ln(x/2)+Euler) + x^2/4 - ...)
    100  *	   therefore y0(x)-2/pi*j0(x)*ln(x) is an even function.
    101  *	   We use the following function to approximate y0,
    102  *		y0(x) = U(z)/V(z) + (2/pi)*(j0(x)*ln(x)), z= x^2
    103  *	   where
    104  *		U(z) = u0 + u1*z + ... + u6*z^6
    105  *		V(z) = 1  + v1*z + ... + v4*z^4
    106  *	   with absolute approximation error bounded by 2**-72.
    107  *	   Note: For tiny x, U/V = u0 and j0(x)~1, hence
    108  *		y0(tiny) = u0 + (2/pi)*ln(tiny), (choose tiny<2**-27)
    109  *	2. For x>=2.
    110  * 		y0(x) = sqrt(2/(pi*x))*(p0(x)*cos(x0)+q0(x)*sin(x0))
    111  * 	   where x0 = x-pi/4. It is better to compute sin(x0),cos(x0)
    112  *	   by the method mentioned above.
    113  *	3. Special cases: y0(0)=-inf, y0(x<0)=NaN, y0(inf)=0.
    114  */
    115 
    116 #include "mathimpl.h"
    117 #include <float.h>
    118 #include <errno.h>
    119 
    120 #if defined(vax) || defined(tahoe)
    121 #define _IEEE	0
    122 #else
    123 #define _IEEE	1
    124 #define infnan(x) (0.0)
    125 #endif
    126 
    127 static double pzero __P((double)), qzero __P((double));
    128 
    129 static double
    130 huge 	= 1e300,
    131 zero    = 0.0,
    132 one	= 1.0,
    133 invsqrtpi= 5.641895835477562869480794515607725858441e-0001,
    134 tpi	= 0.636619772367581343075535053490057448,
    135  		/* R0/S0 on [0, 2.00] */
    136 r02 =   1.562499999999999408594634421055018003102e-0002,
    137 r03 =  -1.899792942388547334476601771991800712355e-0004,
    138 r04 =   1.829540495327006565964161150603950916854e-0006,
    139 r05 =  -4.618326885321032060803075217804816988758e-0009,
    140 s01 =   1.561910294648900170180789369288114642057e-0002,
    141 s02 =   1.169267846633374484918570613449245536323e-0004,
    142 s03 =   5.135465502073181376284426245689510134134e-0007,
    143 s04 =   1.166140033337900097836930825478674320464e-0009;
    144 
    145 double
    146 j0(x)
    147 	double x;
    148 {
    149 	double z, s,c,ss,cc,r,u,v;
    150 
    151 	if (!finite(x))
    152 		if (_IEEE) return one/(x*x);
    153 		else return (0);
    154 	x = fabs(x);
    155 	if (x >= 2.0) {	/* |x| >= 2.0 */
    156 		s = sin(x);
    157 		c = cos(x);
    158 		ss = s-c;
    159 		cc = s+c;
    160 		if (x < .5 * DBL_MAX) {  /* make sure x+x not overflow */
    161 		    z = -cos(x+x);
    162 		    if ((s*c)<zero) cc = z/ss;
    163 		    else 	    ss = z/cc;
    164 		}
    165 	/*
    166 	 * j0(x) = 1/sqrt(pi) * (P(0,x)*cc - Q(0,x)*ss) / sqrt(x)
    167 	 * y0(x) = 1/sqrt(pi) * (P(0,x)*ss + Q(0,x)*cc) / sqrt(x)
    168 	 */
    169 		if (_IEEE && x> 6.80564733841876927e+38) /* 2^129 */
    170 			z = (invsqrtpi*cc)/sqrt(x);
    171 		else {
    172 		    u = pzero(x); v = qzero(x);
    173 		    z = invsqrtpi*(u*cc-v*ss)/sqrt(x);
    174 		}
    175 		return z;
    176 	}
    177 	if (x < 1.220703125e-004) {		   /* |x| < 2**-13 */
    178 	    if (huge+x > one) {			   /* raise inexact if x != 0 */
    179 	        if (x < 7.450580596923828125e-009) /* |x|<2**-27 */
    180 			return one;
    181 	        else return (one - 0.25*x*x);
    182 	    }
    183 	}
    184 	z = x*x;
    185 	r =  z*(r02+z*(r03+z*(r04+z*r05)));
    186 	s =  one+z*(s01+z*(s02+z*(s03+z*s04)));
    187 	if (x < one) {			/* |x| < 1.00 */
    188 	    return (one + z*(-0.25+(r/s)));
    189 	} else {
    190 	    u = 0.5*x;
    191 	    return ((one+u)*(one-u)+z*(r/s));
    192 	}
    193 }
    194 
    195 static double
    196 u00 =  -7.380429510868722527422411862872999615628e-0002,
    197 u01 =   1.766664525091811069896442906220827182707e-0001,
    198 u02 =  -1.381856719455968955440002438182885835344e-0002,
    199 u03 =   3.474534320936836562092566861515617053954e-0004,
    200 u04 =  -3.814070537243641752631729276103284491172e-0006,
    201 u05 =   1.955901370350229170025509706510038090009e-0008,
    202 u06 =  -3.982051941321034108350630097330144576337e-0011,
    203 v01 =   1.273048348341237002944554656529224780561e-0002,
    204 v02 =   7.600686273503532807462101309675806839635e-0005,
    205 v03 =   2.591508518404578033173189144579208685163e-0007,
    206 v04 =   4.411103113326754838596529339004302243157e-0010;
    207 
    208 double
    209 y0(x)
    210 	double x;
    211 {
    212 	double z, s, c, ss, cc, u, v;
    213     /* Y0(NaN) is NaN, y0(-inf) is Nan, y0(inf) is 0  */
    214 	if (!finite(x))
    215 		if (_IEEE)
    216 			return (one/(x+x*x));
    217 		else
    218 			return (0);
    219         if (x == 0)
    220 		if (_IEEE)	return (-one/zero);
    221 		else		return(infnan(-ERANGE));
    222         if (x<0)
    223 		if (_IEEE)	return (zero/zero);
    224 		else		return (infnan(EDOM));
    225         if (x >= 2.00) {	/* |x| >= 2.0 */
    226         /* y0(x) = sqrt(2/(pi*x))*(p0(x)*sin(x0)+q0(x)*cos(x0))
    227          * where x0 = x-pi/4
    228          *      Better formula:
    229          *              cos(x0) = cos(x)cos(pi/4)+sin(x)sin(pi/4)
    230          *                      =  1/sqrt(2) * (sin(x) + cos(x))
    231          *              sin(x0) = sin(x)cos(3pi/4)-cos(x)sin(3pi/4)
    232          *                      =  1/sqrt(2) * (sin(x) - cos(x))
    233          * To avoid cancellation, use
    234          *              sin(x) +- cos(x) = -cos(2x)/(sin(x) -+ cos(x))
    235          * to compute the worse one.
    236          */
    237                 s = sin(x);
    238                 c = cos(x);
    239                 ss = s-c;
    240                 cc = s+c;
    241 	/*
    242 	 * j0(x) = 1/sqrt(pi) * (P(0,x)*cc - Q(0,x)*ss) / sqrt(x)
    243 	 * y0(x) = 1/sqrt(pi) * (P(0,x)*ss + Q(0,x)*cc) / sqrt(x)
    244 	 */
    245                 if (x < .5 * DBL_MAX) {  /* make sure x+x not overflow */
    246                     z = -cos(x+x);
    247                     if ((s*c)<zero) cc = z/ss;
    248                     else            ss = z/cc;
    249                 }
    250                 if (_IEEE && x > 6.80564733841876927e+38) /* > 2^129 */
    251 			z = (invsqrtpi*ss)/sqrt(x);
    252                 else {
    253                     u = pzero(x); v = qzero(x);
    254                     z = invsqrtpi*(u*ss+v*cc)/sqrt(x);
    255                 }
    256                 return z;
    257 	}
    258 	if (x <= 7.450580596923828125e-009) {		/* x < 2**-27 */
    259 	    return (u00 + tpi*log(x));
    260 	}
    261 	z = x*x;
    262 	u = u00+z*(u01+z*(u02+z*(u03+z*(u04+z*(u05+z*u06)))));
    263 	v = one+z*(v01+z*(v02+z*(v03+z*v04)));
    264 	return (u/v + tpi*(j0(x)*log(x)));
    265 }
    266 
    267 /* The asymptotic expansions of pzero is
    268  *	1 - 9/128 s^2 + 11025/98304 s^4 - ...,	where s = 1/x.
    269  * For x >= 2, We approximate pzero by
    270  * 	pzero(x) = 1 + (R/S)
    271  * where  R = pr0 + pr1*s^2 + pr2*s^4 + ... + pr5*s^10
    272  * 	  S = 1 + ps0*s^2 + ... + ps4*s^10
    273  * and
    274  *	| pzero(x)-1-R/S | <= 2  ** ( -60.26)
    275  */
    276 static double pr8[6] = { /* for x in [inf, 8]=1/[0,0.125] */
    277    0.0,
    278   -7.031249999999003994151563066182798210142e-0002,
    279   -8.081670412753498508883963849859423939871e+0000,
    280   -2.570631056797048755890526455854482662510e+0002,
    281   -2.485216410094288379417154382189125598962e+0003,
    282   -5.253043804907295692946647153614119665649e+0003,
    283 };
    284 static double ps8[5] = {
    285    1.165343646196681758075176077627332052048e+0002,
    286    3.833744753641218451213253490882686307027e+0003,
    287    4.059785726484725470626341023967186966531e+0004,
    288    1.167529725643759169416844015694440325519e+0005,
    289    4.762772841467309430100106254805711722972e+0004,
    290 };
    291 
    292 static double pr5[6] = { /* for x in [8,4.5454]=1/[0.125,0.22001] */
    293   -1.141254646918944974922813501362824060117e-0011,
    294   -7.031249408735992804117367183001996028304e-0002,
    295   -4.159610644705877925119684455252125760478e+0000,
    296   -6.767476522651671942610538094335912346253e+0001,
    297   -3.312312996491729755731871867397057689078e+0002,
    298   -3.464333883656048910814187305901796723256e+0002,
    299 };
    300 static double ps5[5] = {
    301    6.075393826923003305967637195319271932944e+0001,
    302    1.051252305957045869801410979087427910437e+0003,
    303    5.978970943338558182743915287887408780344e+0003,
    304    9.625445143577745335793221135208591603029e+0003,
    305    2.406058159229391070820491174867406875471e+0003,
    306 };
    307 
    308 static double pr3[6] = {/* for x in [4.547,2.8571]=1/[0.2199,0.35001] */
    309   -2.547046017719519317420607587742992297519e-0009,
    310   -7.031196163814817199050629727406231152464e-0002,
    311   -2.409032215495295917537157371488126555072e+0000,
    312   -2.196597747348830936268718293366935843223e+0001,
    313   -5.807917047017375458527187341817239891940e+0001,
    314   -3.144794705948885090518775074177485744176e+0001,
    315 };
    316 static double ps3[5] = {
    317    3.585603380552097167919946472266854507059e+0001,
    318    3.615139830503038919981567245265266294189e+0002,
    319    1.193607837921115243628631691509851364715e+0003,
    320    1.127996798569074250675414186814529958010e+0003,
    321    1.735809308133357510239737333055228118910e+0002,
    322 };
    323 
    324 static double pr2[6] = {/* for x in [2.8570,2]=1/[0.3499,0.5] */
    325   -8.875343330325263874525704514800809730145e-0008,
    326   -7.030309954836247756556445443331044338352e-0002,
    327   -1.450738467809529910662233622603401167409e+0000,
    328   -7.635696138235277739186371273434739292491e+0000,
    329   -1.119316688603567398846655082201614524650e+0001,
    330   -3.233645793513353260006821113608134669030e+0000,
    331 };
    332 static double ps2[5] = {
    333    2.222029975320888079364901247548798910952e+0001,
    334    1.362067942182152109590340823043813120940e+0002,
    335    2.704702786580835044524562897256790293238e+0002,
    336    1.538753942083203315263554770476850028583e+0002,
    337    1.465761769482561965099880599279699314477e+0001,
    338 };
    339 
    340 static double pzero(x)
    341 	double x;
    342 {
    343 	double *p,*q,z,r,s;
    344 	if (x >= 8.00)			   {p = pr8; q= ps8;}
    345 	else if (x >= 4.54545211791992188) {p = pr5; q= ps5;}
    346 	else if (x >= 2.85714149475097656) {p = pr3; q= ps3;}
    347 	else /* if (x >= 2.00) */	   {p = pr2; q= ps2;}
    348 	z = one/(x*x);
    349 	r = p[0]+z*(p[1]+z*(p[2]+z*(p[3]+z*(p[4]+z*p[5]))));
    350 	s = one+z*(q[0]+z*(q[1]+z*(q[2]+z*(q[3]+z*q[4]))));
    351 	return one+ r/s;
    352 }
    353 
    354 
    355 /* For x >= 8, the asymptotic expansions of qzero is
    356  *	-1/8 s + 75/1024 s^3 - ..., where s = 1/x.
    357  * We approximate pzero by
    358  * 	qzero(x) = s*(-1.25 + (R/S))
    359  * where  R = qr0 + qr1*s^2 + qr2*s^4 + ... + qr5*s^10
    360  * 	  S = 1 + qs0*s^2 + ... + qs5*s^12
    361  * and
    362  *	| qzero(x)/s +1.25-R/S | <= 2  ** ( -61.22)
    363  */
    364 static double qr8[6] = { /* for x in [inf, 8]=1/[0,0.125] */
    365    0.0,
    366    7.324218749999350414479738504551775297096e-0002,
    367    1.176820646822526933903301695932765232456e+0001,
    368    5.576733802564018422407734683549251364365e+0002,
    369    8.859197207564685717547076568608235802317e+0003,
    370    3.701462677768878501173055581933725704809e+0004,
    371 };
    372 static double qs8[6] = {
    373    1.637760268956898345680262381842235272369e+0002,
    374    8.098344946564498460163123708054674227492e+0003,
    375    1.425382914191204905277585267143216379136e+0005,
    376    8.033092571195144136565231198526081387047e+0005,
    377    8.405015798190605130722042369969184811488e+0005,
    378   -3.438992935378666373204500729736454421006e+0005,
    379 };
    380 
    381 static double qr5[6] = { /* for x in [8,4.5454]=1/[0.125,0.22001] */
    382    1.840859635945155400568380711372759921179e-0011,
    383    7.324217666126847411304688081129741939255e-0002,
    384    5.835635089620569401157245917610984757296e+0000,
    385    1.351115772864498375785526599119895942361e+0002,
    386    1.027243765961641042977177679021711341529e+0003,
    387    1.989977858646053872589042328678602481924e+0003,
    388 };
    389 static double qs5[6] = {
    390    8.277661022365377058749454444343415524509e+0001,
    391    2.077814164213929827140178285401017305309e+0003,
    392    1.884728877857180787101956800212453218179e+0004,
    393    5.675111228949473657576693406600265778689e+0004,
    394    3.597675384251145011342454247417399490174e+0004,
    395   -5.354342756019447546671440667961399442388e+0003,
    396 };
    397 
    398 static double qr3[6] = {/* for x in [4.547,2.8571]=1/[0.2199,0.35001] */
    399    4.377410140897386263955149197672576223054e-0009,
    400    7.324111800429115152536250525131924283018e-0002,
    401    3.344231375161707158666412987337679317358e+0000,
    402    4.262184407454126175974453269277100206290e+0001,
    403    1.708080913405656078640701512007621675724e+0002,
    404    1.667339486966511691019925923456050558293e+0002,
    405 };
    406 static double qs3[6] = {
    407    4.875887297245871932865584382810260676713e+0001,
    408    7.096892210566060535416958362640184894280e+0002,
    409    3.704148226201113687434290319905207398682e+0003,
    410    6.460425167525689088321109036469797462086e+0003,
    411    2.516333689203689683999196167394889715078e+0003,
    412   -1.492474518361563818275130131510339371048e+0002,
    413 };
    414 
    415 static double qr2[6] = {/* for x in [2.8570,2]=1/[0.3499,0.5] */
    416    1.504444448869832780257436041633206366087e-0007,
    417    7.322342659630792930894554535717104926902e-0002,
    418    1.998191740938159956838594407540292600331e+0000,
    419    1.449560293478857407645853071687125850962e+0001,
    420    3.166623175047815297062638132537957315395e+0001,
    421    1.625270757109292688799540258329430963726e+0001,
    422 };
    423 static double qs2[6] = {
    424    3.036558483552191922522729838478169383969e+0001,
    425    2.693481186080498724211751445725708524507e+0002,
    426    8.447837575953201460013136756723746023736e+0002,
    427    8.829358451124885811233995083187666981299e+0002,
    428    2.126663885117988324180482985363624996652e+0002,
    429   -5.310954938826669402431816125780738924463e+0000,
    430 };
    431 
    432 static double qzero(x)
    433 	double x;
    434 {
    435 	double *p,*q, s,r,z;
    436 	if (x >= 8.00)			   {p = qr8; q= qs8;}
    437 	else if (x >= 4.54545211791992188) {p = qr5; q= qs5;}
    438 	else if (x >= 2.85714149475097656) {p = qr3; q= qs3;}
    439 	else /* if (x >= 2.00) */	   {p = qr2; q= qs2;}
    440 	z = one/(x*x);
    441 	r = p[0]+z*(p[1]+z*(p[2]+z*(p[3]+z*(p[4]+z*p[5]))));
    442 	s = one+z*(q[0]+z*(q[1]+z*(q[2]+z*(q[3]+z*(q[4]+z*q[5])))));
    443 	return (-.125 + r/s)/x;
    444 }
    445