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 /*
13  * __ieee754_jn(n, x), __ieee754_yn(n, x)
14  * floating point Bessel's function of the 1st and 2nd kind
15  * of order n
16  *
17  * Special cases:
18  *	y0(0)=y1(0)=yn(n,0) = -inf with division by zero signal;
19  *	y0(-ve)=y1(-ve)=yn(n,-ve) are NaN with invalid signal.
20  * Note 2. About jn(n,x), yn(n,x)
21  *	For n=0, j0(x) is called,
22  *	for n=1, j1(x) is called,
23  *	for n<x, forward recursion us used starting
24  *	from values of j0(x) and j1(x).
25  *	for n>x, a continued fraction approximation to
26  *	j(n,x)/j(n-1,x) is evaluated and then backward
27  *	recursion is used starting from a supposed value
28  *	for j(n,x). The resulting value of j(0,x) is
29  *	compared with the actual value to correct the
30  *	supposed value of j(n,x).
31  *
32  *	yn(n,x) is similar in all respects, except
33  *	that forward recursion is used for all
34  *	values of n>1.
35  *
36  */
37 
38 #include "math.h"
39 #include "math_private.h"
40 
41 static const double
42 invsqrtpi=  5.64189583547756279280e-01, /* 0x3FE20DD7, 0x50429B6D */
43 two   =  2.00000000000000000000e+00, /* 0x40000000, 0x00000000 */
44 one   =  1.00000000000000000000e+00; /* 0x3FF00000, 0x00000000 */
45 
46 static const double zero  =  0.00000000000000000000e+00;
47 
__ieee754_jn(int n,double x)48 double __ieee754_jn(int n, double x)
49 {
50 	int32_t i,hx,ix,lx, sgn;
51 	double a, b, temp=0, di;
52 	double z, w;
53 
54     /* J(-n,x) = (-1)^n * J(n, x), J(n, -x) = (-1)^n * J(n, x)
55      * Thus, J(-n,x) = J(n,-x)
56      */
57 	EXTRACT_WORDS(hx,lx,x);
58 	ix = 0x7fffffff&hx;
59     /* if J(n,NaN) is NaN */
60 	if((ix|((u_int32_t)(lx|-lx))>>31)>0x7ff00000) return x+x;
61 	if(n<0){
62 		n = -n;
63 		x = -x;
64 		hx ^= 0x80000000;
65 	}
66 	if(n==0) return(__ieee754_j0(x));
67 	if(n==1) return(__ieee754_j1(x));
68 	sgn = (n&1)&(hx>>31);	/* even n -- 0, odd n -- sign(x) */
69 	x = fabs(x);
70 	if((ix|lx)==0||ix>=0x7ff00000) 	/* if x is 0 or inf */
71 	    b = zero;
72 	else if((double)n<=x) {
73 		/* Safe to use J(n+1,x)=2n/x *J(n,x)-J(n-1,x) */
74 	    if(ix>=0x52D00000) { /* x > 2**302 */
75     /* (x >> n**2)
76      *	    Jn(x) = cos(x-(2n+1)*pi/4)*sqrt(2/x*pi)
77      *	    Yn(x) = sin(x-(2n+1)*pi/4)*sqrt(2/x*pi)
78      *	    Let s=sin(x), c=cos(x),
79      *		xn=x-(2n+1)*pi/4, sqt2 = sqrt(2),then
80      *
81      *		   n	sin(xn)*sqt2	cos(xn)*sqt2
82      *		----------------------------------
83      *		   0	 s-c		 c+s
84      *		   1	-s-c 		-c+s
85      *		   2	-s+c		-c-s
86      *		   3	 s+c		 c-s
87      */
88 		switch(n&3) {
89 		    case 0: temp =  cos(x)+sin(x); break;
90 		    case 1: temp = -cos(x)+sin(x); break;
91 		    case 2: temp = -cos(x)-sin(x); break;
92 		    case 3: temp =  cos(x)-sin(x); break;
93 		}
94 		b = invsqrtpi*temp/sqrt(x);
95 	    } else {
96 	        a = __ieee754_j0(x);
97 	        b = __ieee754_j1(x);
98 	        for(i=1;i<n;i++){
99 		    temp = b;
100 		    b = b*((double)(i+i)/x) - a; /* avoid underflow */
101 		    a = temp;
102 	        }
103 	    }
104 	} else {
105 	    if(ix<0x3e100000) {	/* x < 2**-29 */
106     /* x is tiny, return the first Taylor expansion of J(n,x)
107      * J(n,x) = 1/n!*(x/2)^n  - ...
108      */
109 		if(n>33)	/* underflow */
110 		    b = zero;
111 		else {
112 		    temp = x*0.5; b = temp;
113 		    for (a=one,i=2;i<=n;i++) {
114 			a *= (double)i;		/* a = n! */
115 			b *= temp;		/* b = (x/2)^n */
116 		    }
117 		    b = b/a;
118 		}
119 	    } else {
120 		/* use backward recurrence */
121 		/* 			x      x^2      x^2
122 		 *  J(n,x)/J(n-1,x) =  ----   ------   ------   .....
123 		 *			2n  - 2(n+1) - 2(n+2)
124 		 *
125 		 * 			1      1        1
126 		 *  (for large x)   =  ----  ------   ------   .....
127 		 *			2n   2(n+1)   2(n+2)
128 		 *			-- - ------ - ------ -
129 		 *			 x     x         x
130 		 *
131 		 * Let w = 2n/x and h=2/x, then the above quotient
132 		 * is equal to the continued fraction:
133 		 *		    1
134 		 *	= -----------------------
135 		 *		       1
136 		 *	   w - -----------------
137 		 *			  1
138 		 * 	        w+h - ---------
139 		 *		       w+2h - ...
140 		 *
141 		 * To determine how many terms needed, let
142 		 * Q(0) = w, Q(1) = w(w+h) - 1,
143 		 * Q(k) = (w+k*h)*Q(k-1) - Q(k-2),
144 		 * When Q(k) > 1e4	good for single
145 		 * When Q(k) > 1e9	good for double
146 		 * When Q(k) > 1e17	good for quadruple
147 		 */
148 	    /* determine k */
149 		double t,v;
150 		double q0,q1,h,tmp; int32_t k,m;
151 		w  = (n+n)/(double)x; h = 2.0/(double)x;
152 		q0 = w;  z = w+h; q1 = w*z - 1.0; k=1;
153 		while(q1<1.0e9) {
154 			k += 1; z += h;
155 			tmp = z*q1 - q0;
156 			q0 = q1;
157 			q1 = tmp;
158 		}
159 		m = n+n;
160 		for(t=zero, i = 2*(n+k); i>=m; i -= 2) t = one/(i/x-t);
161 		a = t;
162 		b = one;
163 		/*  estimate log((2/x)^n*n!) = n*log(2/x)+n*ln(n)
164 		 *  Hence, if n*(log(2n/x)) > ...
165 		 *  single 8.8722839355e+01
166 		 *  double 7.09782712893383973096e+02
167 		 *  long double 1.1356523406294143949491931077970765006170e+04
168 		 *  then recurrent value may overflow and the result is
169 		 *  likely underflow to zero
170 		 */
171 		tmp = n;
172 		v = two/x;
173 		tmp = tmp*__ieee754_log(fabs(v*tmp));
174 		if(tmp<7.09782712893383973096e+02) {
175 	    	    for(i=n-1,di=(double)(i+i);i>0;i--){
176 		        temp = b;
177 			b *= di;
178 			b  = b/x - a;
179 		        a = temp;
180 			di -= two;
181 	     	    }
182 		} else {
183 	    	    for(i=n-1,di=(double)(i+i);i>0;i--){
184 		        temp = b;
185 			b *= di;
186 			b  = b/x - a;
187 		        a = temp;
188 			di -= two;
189 		    /* scale b to avoid spurious overflow */
190 			if(b>1e100) {
191 			    a /= b;
192 			    t /= b;
193 			    b  = one;
194 			}
195 	     	    }
196 		}
197 	    	b = (t*__ieee754_j0(x)/b);
198 	    }
199 	}
200 	if(sgn==1) return -b; else return b;
201 }
202 
__ieee754_yn(int n,double x)203 double __ieee754_yn(int n, double x)
204 {
205 	int32_t i,hx,ix,lx;
206 	int32_t sign;
207 	double a, b, temp=0;
208 
209 	EXTRACT_WORDS(hx,lx,x);
210 	ix = 0x7fffffff&hx;
211     /* if Y(n,NaN) is NaN */
212 	if((ix|((u_int32_t)(lx|-lx))>>31)>0x7ff00000) return x+x;
213 	if((ix|lx)==0) return -one/zero;
214 	if(hx<0) return zero/zero;
215 	sign = 1;
216 	if(n<0){
217 		n = -n;
218 		sign = 1 - ((n&1)<<1);
219 	}
220 	if(n==0) return(__ieee754_y0(x));
221 	if(n==1) return(sign*__ieee754_y1(x));
222 	if(ix==0x7ff00000) return zero;
223 	if(ix>=0x52D00000) { /* x > 2**302 */
224     /* (x >> n**2)
225      *	    Jn(x) = cos(x-(2n+1)*pi/4)*sqrt(2/x*pi)
226      *	    Yn(x) = sin(x-(2n+1)*pi/4)*sqrt(2/x*pi)
227      *	    Let s=sin(x), c=cos(x),
228      *		xn=x-(2n+1)*pi/4, sqt2 = sqrt(2),then
229      *
230      *		   n	sin(xn)*sqt2	cos(xn)*sqt2
231      *		----------------------------------
232      *		   0	 s-c		 c+s
233      *		   1	-s-c 		-c+s
234      *		   2	-s+c		-c-s
235      *		   3	 s+c		 c-s
236      */
237 		switch(n&3) {
238 		    case 0: temp =  sin(x)-cos(x); break;
239 		    case 1: temp = -sin(x)-cos(x); break;
240 		    case 2: temp = -sin(x)+cos(x); break;
241 		    case 3: temp =  sin(x)+cos(x); break;
242 		}
243 		b = invsqrtpi*temp/sqrt(x);
244 	} else {
245 	    u_int32_t high;
246 	    a = __ieee754_y0(x);
247 	    b = __ieee754_y1(x);
248 	/* quit if b is -inf */
249 	    GET_HIGH_WORD(high,b);
250 	    for(i=1;i<n&&high!=0xfff00000;i++){
251 		temp = b;
252 		b = ((double)(i+i)/x)*b - a;
253 		GET_HIGH_WORD(high,b);
254 		a = temp;
255 	    }
256 	}
257 	if(sign>0) return b; else return -b;
258 }
259