1 /* 2 * Copyright (c) 1985 Regents of the University of California. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms are permitted 6 * provided that the above copyright notice and this paragraph are 7 * duplicated in all such forms and that any documentation, 8 * advertising materials, and other materials related to such 9 * distribution and use acknowledge that the software was developed 10 * by the University of California, Berkeley. The name of the 11 * University may not be used to endorse or promote products derived 12 * from this software without specific prior written permission. 13 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR 14 * IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED 15 * WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR A PARTICULAR PURPOSE. 16 * 17 * All recipients should regard themselves as participants in an ongoing 18 * research project and hence should feel obligated to report their 19 * experiences (good or bad) with these elementary function codes, using 20 * the sendbug(8) program, to the authors. 21 */ 22 23 #ifndef lint 24 static char sccsid[] = "@(#)asincos.c 5.3 (Berkeley) 06/30/88"; 25 #endif /* not lint */ 26 27 /* ASIN(X) 28 * RETURNS ARC SINE OF X 29 * DOUBLE PRECISION (IEEE DOUBLE 53 bits, VAX D FORMAT 56 bits) 30 * CODED IN C BY K.C. NG, 4/16/85, REVISED ON 6/10/85. 31 * 32 * Required system supported functions: 33 * copysign(x,y) 34 * sqrt(x) 35 * 36 * Required kernel function: 37 * atan2(y,x) 38 * 39 * Method : 40 * asin(x) = atan2(x,sqrt(1-x*x)); for better accuracy, 1-x*x is 41 * computed as follows 42 * 1-x*x if x < 0.5, 43 * 2*(1-|x|)-(1-|x|)*(1-|x|) if x >= 0.5. 44 * 45 * Special cases: 46 * if x is NaN, return x itself; 47 * if |x|>1, return NaN. 48 * 49 * Accuracy: 50 * 1) If atan2() uses machine PI, then 51 * 52 * asin(x) returns (PI/pi) * (the exact arc sine of x) nearly rounded; 53 * and PI is the exact pi rounded to machine precision (see atan2 for 54 * details): 55 * 56 * in decimal: 57 * pi = 3.141592653589793 23846264338327 ..... 58 * 53 bits PI = 3.141592653589793 115997963 ..... , 59 * 56 bits PI = 3.141592653589793 227020265 ..... , 60 * 61 * in hexadecimal: 62 * pi = 3.243F6A8885A308D313198A2E.... 63 * 53 bits PI = 3.243F6A8885A30 = 2 * 1.921FB54442D18 error=.276ulps 64 * 56 bits PI = 3.243F6A8885A308 = 4 * .C90FDAA22168C2 error=.206ulps 65 * 66 * In a test run with more than 200,000 random arguments on a VAX, the 67 * maximum observed error in ulps (units in the last place) was 68 * 2.06 ulps. (comparing against (PI/pi)*(exact asin(x))); 69 * 70 * 2) If atan2() uses true pi, then 71 * 72 * asin(x) returns the exact asin(x) with error below about 2 ulps. 73 * 74 * In a test run with more than 1,024,000 random arguments on a VAX, the 75 * maximum observed error in ulps (units in the last place) was 76 * 1.99 ulps. 77 */ 78 79 double asin(x) 80 double x; 81 { 82 double s,t,copysign(),atan2(),sqrt(),one=1.0; 83 #if !defined(vax)&&!defined(tahoe) 84 if(x!=x) return(x); /* x is NaN */ 85 #endif /* !defined(vax)&&!defined(tahoe) */ 86 s=copysign(x,one); 87 if(s <= 0.5) 88 return(atan2(x,sqrt(one-x*x))); 89 else 90 { t=one-s; s=t+t; return(atan2(x,sqrt(s-t*t))); } 91 92 } 93 94 /* ACOS(X) 95 * RETURNS ARC COS OF X 96 * DOUBLE PRECISION (IEEE DOUBLE 53 bits, VAX D FORMAT 56 bits) 97 * CODED IN C BY K.C. NG, 4/16/85, REVISED ON 6/10/85. 98 * 99 * Required system supported functions: 100 * copysign(x,y) 101 * sqrt(x) 102 * 103 * Required kernel function: 104 * atan2(y,x) 105 * 106 * Method : 107 * ________ 108 * / 1 - x 109 * acos(x) = 2*atan2( / -------- , 1 ) . 110 * \/ 1 + x 111 * 112 * Special cases: 113 * if x is NaN, return x itself; 114 * if |x|>1, return NaN. 115 * 116 * Accuracy: 117 * 1) If atan2() uses machine PI, then 118 * 119 * acos(x) returns (PI/pi) * (the exact arc cosine of x) nearly rounded; 120 * and PI is the exact pi rounded to machine precision (see atan2 for 121 * details): 122 * 123 * in decimal: 124 * pi = 3.141592653589793 23846264338327 ..... 125 * 53 bits PI = 3.141592653589793 115997963 ..... , 126 * 56 bits PI = 3.141592653589793 227020265 ..... , 127 * 128 * in hexadecimal: 129 * pi = 3.243F6A8885A308D313198A2E.... 130 * 53 bits PI = 3.243F6A8885A30 = 2 * 1.921FB54442D18 error=.276ulps 131 * 56 bits PI = 3.243F6A8885A308 = 4 * .C90FDAA22168C2 error=.206ulps 132 * 133 * In a test run with more than 200,000 random arguments on a VAX, the 134 * maximum observed error in ulps (units in the last place) was 135 * 2.07 ulps. (comparing against (PI/pi)*(exact acos(x))); 136 * 137 * 2) If atan2() uses true pi, then 138 * 139 * acos(x) returns the exact acos(x) with error below about 2 ulps. 140 * 141 * In a test run with more than 1,024,000 random arguments on a VAX, the 142 * maximum observed error in ulps (units in the last place) was 143 * 2.15 ulps. 144 */ 145 146 double acos(x) 147 double x; 148 { 149 double t,copysign(),atan2(),sqrt(),one=1.0; 150 #if !defined(vax)&&!defined(tahoe) 151 if(x!=x) return(x); 152 #endif /* !defined(vax)&&!defined(tahoe) */ 153 if( x != -1.0) 154 t=atan2(sqrt((one-x)/(one+x)),one); 155 else 156 t=atan2(one,0.0); /* t = PI/2 */ 157 return(t+t); 158 } 159