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   Long double expansions are
14   Copyright (C) 2001 Stephen L. Moshier <moshier@na-net.ornl.gov>
15   and are incorporated herein by permission of the author.  The author
16   reserves the right to distribute this material elsewhere under different
17   copying permissions.  These modifications are distributed here under
18   the following terms:
19 
20     This library is free software; you can redistribute it and/or
21     modify it under the terms of the GNU Lesser General Public
22     License as published by the Free Software Foundation; either
23     version 2.1 of the License, or (at your option) any later version.
24 
25     This library is distributed in the hope that it will be useful,
26     but WITHOUT ANY WARRANTY; without even the implied warranty of
27     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
28     Lesser General Public License for more details.
29 
30     You should have received a copy of the GNU Lesser General Public
31     License along with this library; if not, see
32     <http://www.gnu.org/licenses/>.  */
33 
34 /* __quadmath_kernel_tanq( x, y, k )
35  * kernel tan function on [-pi/4, pi/4], pi/4 ~ 0.7854
36  * Input x is assumed to be bounded by ~pi/4 in magnitude.
37  * Input y is the tail of x.
38  * Input k indicates whether tan (if k=1) or
39  * -1/tan (if k= -1) is returned.
40  *
41  * Algorithm
42  *	1. Since tan(-x) = -tan(x), we need only to consider positive x.
43  *	2. if x < 2^-57, return x with inexact if x!=0.
44  *	3. tan(x) is approximated by a rational form x + x^3 / 3 + x^5 R(x^2)
45  *          on [0,0.67433].
46  *
47  *	   Note: tan(x+y) = tan(x) + tan'(x)*y
48  *		          ~ tan(x) + (1+x*x)*y
49  *	   Therefore, for better accuracy in computing tan(x+y), let
50  *		r = x^3 * R(x^2)
51  *	   then
52  *		tan(x+y) = x + (x^3 / 3 + (x^2 *(r+y)+y))
53  *
54  *      4. For x in [0.67433,pi/4],  let y = pi/4 - x, then
55  *		tan(x) = tan(pi/4-y) = (1-tan(y))/(1+tan(y))
56  *		       = 1 - 2*(tan(y) - (tan(y)^2)/(1+tan(y)))
57  */
58 
59 #include "quadmath-imp.h"
60 
61 static const __float128
62   one = 1,
63   pio4hi = 7.8539816339744830961566084581987569936977E-1Q,
64   pio4lo = 2.1679525325309452561992610065108379921906E-35Q,
65 
66   /* tan x = x + x^3 / 3 + x^5 T(x^2)/U(x^2)
67      0 <= x <= 0.6743316650390625
68      Peak relative error 8.0e-36  */
69  TH =  3.333333333333333333333333333333333333333E-1Q,
70  T0 = -1.813014711743583437742363284336855889393E7Q,
71  T1 =  1.320767960008972224312740075083259247618E6Q,
72  T2 = -2.626775478255838182468651821863299023956E4Q,
73  T3 =  1.764573356488504935415411383687150199315E2Q,
74  T4 = -3.333267763822178690794678978979803526092E-1Q,
75 
76  U0 = -1.359761033807687578306772463253710042010E8Q,
77  U1 =  6.494370630656893175666729313065113194784E7Q,
78  U2 = -4.180787672237927475505536849168729386782E6Q,
79  U3 =  8.031643765106170040139966622980914621521E4Q,
80  U4 = -5.323131271912475695157127875560667378597E2Q;
81   /* 1.000000000000000000000000000000000000000E0 */
82 
83 
84 __float128
__quadmath_kernel_tanq(__float128 x,__float128 y,int iy)85 __quadmath_kernel_tanq (__float128 x, __float128 y, int iy)
86 {
87   __float128 z, r, v, w, s;
88   int32_t ix, sign;
89   ieee854_float128 u, u1;
90 
91   u.value = x;
92   ix = u.words32.w0 & 0x7fffffff;
93   if (ix < 0x3fc60000)		/* x < 2**-57 */
94     {
95       if ((int) x == 0)
96 	{			/* generate inexact */
97 	  if ((ix | u.words32.w1 | u.words32.w2 | u.words32.w3
98 	       | (iy + 1)) == 0)
99 	    return one / fabsq (x);
100 	  else if (iy == 1)
101 	    {
102 	      math_check_force_underflow (x);
103 	      return x;
104 	    }
105 	  else
106 	    return -one / x;
107 	}
108     }
109   if (ix >= 0x3ffe5942) /* |x| >= 0.6743316650390625 */
110     {
111       if ((u.words32.w0 & 0x80000000) != 0)
112 	{
113 	  x = -x;
114 	  y = -y;
115 	  sign = -1;
116 	}
117       else
118 	sign = 1;
119       z = pio4hi - x;
120       w = pio4lo - y;
121       x = z + w;
122       y = 0.0;
123     }
124   z = x * x;
125   r = T0 + z * (T1 + z * (T2 + z * (T3 + z * T4)));
126   v = U0 + z * (U1 + z * (U2 + z * (U3 + z * (U4 + z))));
127   r = r / v;
128 
129   s = z * x;
130   r = y + z * (s * r + y);
131   r += TH * s;
132   w = x + r;
133   if (ix >= 0x3ffe5942)
134     {
135       v = (__float128) iy;
136       w = (v - 2.0 * (x - (w * w / (w + v) - r)));
137       /* SIGN is set for arguments that reach this code, but not
138 	 otherwise, resulting in warnings that it may be used
139 	 uninitialized although in the cases where it is used it has
140 	 always been set.  */
141 
142 
143       if (sign < 0)
144 	w = -w;
145 
146       return w;
147     }
148   if (iy == 1)
149     return w;
150   else
151     {				/* if allow error up to 2 ulp,
152 				   simply return -1.0/(x+r) here */
153       /*  compute -1.0/(x+r) accurately */
154       u1.value = w;
155       u1.words32.w2 = 0;
156       u1.words32.w3 = 0;
157       v = r - (u1.value - x);		/* u1+v = r+x */
158       z = -1.0 / w;
159       u.value = z;
160       u.words32.w2 = 0;
161       u.words32.w3 = 0;
162       s = 1.0 + u.value * u1.value;
163       return u.value + z * (s + u.value * v);
164     }
165 }
166