Subversion Repositories Kolibri OS

Rev

Go to most recent revision | Blame | Last modification | View Log | Download | RSS feed

  1.  
  2. /* @(#)e_exp.c 5.1 93/09/24 */
  3. /*
  4.  * ====================================================
  5.  * Copyright (C) 1993 by Sun Microsystems, Inc. All rights reserved.
  6.  *
  7.  * Developed at SunPro, a Sun Microsystems, Inc. business.
  8.  * Permission to use, copy, modify, and distribute this
  9.  * software is freely granted, provided that this notice
  10.  * is preserved.
  11.  * ====================================================
  12.  */
  13.  
  14. /* __ieee754_exp(x)
  15.  * Returns the exponential of x.
  16.  *
  17.  * Method
  18.  *   1. Argument reduction:
  19.  *      Reduce x to an r so that |r| <= 0.5*ln2 ~ 0.34658.
  20.  *      Given x, find r and integer k such that
  21.  *
  22.  *               x = k*ln2 + r,  |r| <= 0.5*ln2.  
  23.  *
  24.  *      Here r will be represented as r = hi-lo for better
  25.  *      accuracy.
  26.  *
  27.  *   2. Approximation of exp(r) by a special rational function on
  28.  *      the interval [0,0.34658]:
  29.  *      Write
  30.  *          R(r**2) = r*(exp(r)+1)/(exp(r)-1) = 2 + r*r/6 - r**4/360 + ...
  31.  *      We use a special Reme algorithm on [0,0.34658] to generate
  32.  *      a polynomial of degree 5 to approximate R. The maximum error
  33.  *      of this polynomial approximation is bounded by 2**-59. In
  34.  *      other words,
  35.  *          R(z) ~ 2.0 + P1*z + P2*z**2 + P3*z**3 + P4*z**4 + P5*z**5
  36.  *      (where z=r*r, and the values of P1 to P5 are listed below)
  37.  *      and
  38.  *          |                  5          |     -59
  39.  *          | 2.0+P1*z+...+P5*z   -  R(z) | <= 2
  40.  *          |                             |
  41.  *      The computation of exp(r) thus becomes
  42.  *                             2*r
  43.  *              exp(r) = 1 + -------
  44.  *                            R - r
  45.  *                                 r*R1(r)     
  46.  *                     = 1 + r + ----------- (for better accuracy)
  47.  *                                2 - R1(r)
  48.  *      where
  49.  *                               2       4             10
  50.  *              R1(r) = r - (P1*r  + P2*r  + ... + P5*r   ).
  51.  *     
  52.  *   3. Scale back to obtain exp(x):
  53.  *      From step 1, we have
  54.  *         exp(x) = 2^k * exp(r)
  55.  *
  56.  * Special cases:
  57.  *      exp(INF) is INF, exp(NaN) is NaN;
  58.  *      exp(-INF) is 0, and
  59.  *      for finite argument, only exp(0)=1 is exact.
  60.  *
  61.  * Accuracy:
  62.  *      according to an error analysis, the error is always less than
  63.  *      1 ulp (unit in the last place).
  64.  *
  65.  * Misc. info.
  66.  *      For IEEE double
  67.  *          if x >  7.09782712893383973096e+02 then exp(x) overflow
  68.  *          if x < -7.45133219101941108420e+02 then exp(x) underflow
  69.  *
  70.  * Constants:
  71.  * The hexadecimal values are the intended ones for the following
  72.  * constants. The decimal values may be used, provided that the
  73.  * compiler will convert from decimal to binary accurately enough
  74.  * to produce the hexadecimal values shown.
  75.  */
  76.  
  77. #include "fdlibm.h"
  78.  
  79. #ifndef _DOUBLE_IS_32BITS
  80.  
  81. #ifdef __STDC__
  82. static const double
  83. #else
  84. static double
  85. #endif
  86. one     = 1.0,
  87. halF[2] = {0.5,-0.5,},
  88. huge    = 1.0e+300,
  89. twom1000= 9.33263618503218878990e-302,     /* 2**-1000=0x01700000,0*/
  90. o_threshold=  7.09782712893383973096e+02,  /* 0x40862E42, 0xFEFA39EF */
  91. u_threshold= -7.45133219101941108420e+02,  /* 0xc0874910, 0xD52D3051 */
  92. ln2HI[2]   ={ 6.93147180369123816490e-01,  /* 0x3fe62e42, 0xfee00000 */
  93.              -6.93147180369123816490e-01,},/* 0xbfe62e42, 0xfee00000 */
  94. ln2LO[2]   ={ 1.90821492927058770002e-10,  /* 0x3dea39ef, 0x35793c76 */
  95.              -1.90821492927058770002e-10,},/* 0xbdea39ef, 0x35793c76 */
  96. invln2 =  1.44269504088896338700e+00, /* 0x3ff71547, 0x652b82fe */
  97. P1   =  1.66666666666666019037e-01, /* 0x3FC55555, 0x5555553E */
  98. P2   = -2.77777777770155933842e-03, /* 0xBF66C16C, 0x16BEBD93 */
  99. P3   =  6.61375632143793436117e-05, /* 0x3F11566A, 0xAF25DE2C */
  100. P4   = -1.65339022054652515390e-06, /* 0xBEBBBD41, 0xC5D26BF1 */
  101. P5   =  4.13813679705723846039e-08; /* 0x3E663769, 0x72BEA4D0 */
  102.  
  103.  
  104. #ifdef __STDC__
  105.         double __ieee754_exp(double x)  /* default IEEE double exp */
  106. #else
  107.         double __ieee754_exp(x) /* default IEEE double exp */
  108.         double x;
  109. #endif
  110. {
  111.         double y,hi,lo,c,t;
  112.         __int32_t k = 0,xsb;
  113.         __uint32_t hx;
  114.  
  115.         GET_HIGH_WORD(hx,x);
  116.         xsb = (hx>>31)&1;               /* sign bit of x */
  117.         hx &= 0x7fffffff;               /* high word of |x| */
  118.  
  119.     /* filter out non-finite argument */
  120.         if(hx >= 0x40862E42) {                  /* if |x|>=709.78... */
  121.             if(hx>=0x7ff00000) {
  122.                 __uint32_t lx;
  123.                 GET_LOW_WORD(lx,x);
  124.                 if(((hx&0xfffff)|lx)!=0)
  125.                      return x+x;                /* NaN */
  126.                 else return (xsb==0)? x:0.0;    /* exp(+-inf)={inf,0} */
  127.             }
  128.             if(x > o_threshold) return huge*huge; /* overflow */
  129.             if(x < u_threshold) return twom1000*twom1000; /* underflow */
  130.         }
  131.  
  132.     /* argument reduction */
  133.         if(hx > 0x3fd62e42) {           /* if  |x| > 0.5 ln2 */
  134.             if(hx < 0x3FF0A2B2) {       /* and |x| < 1.5 ln2 */
  135.                 hi = x-ln2HI[xsb]; lo=ln2LO[xsb]; k = 1-xsb-xsb;
  136.             } else {
  137.                 k  = invln2*x+halF[xsb];
  138.                 t  = k;
  139.                 hi = x - t*ln2HI[0];    /* t*ln2HI is exact here */
  140.                 lo = t*ln2LO[0];
  141.             }
  142.             x  = hi - lo;
  143.         }
  144.         else if(hx < 0x3e300000)  {     /* when |x|<2**-28 */
  145.             if(huge+x>one) return one+x;/* trigger inexact */
  146.         }
  147.  
  148.     /* x is now in primary range */
  149.         t  = x*x;
  150.         c  = x - t*(P1+t*(P2+t*(P3+t*(P4+t*P5))));
  151.         if(k==0)        return one-((x*c)/(c-2.0)-x);
  152.         else            y = one-((lo-(x*c)/(2.0-c))-hi);
  153.         if(k >= -1021) {
  154.             __uint32_t hy;
  155.             GET_HIGH_WORD(hy,y);
  156.             SET_HIGH_WORD(y,hy+(k<<20));        /* add k to y's exponent */
  157.             return y;
  158.         } else {
  159.             __uint32_t hy;
  160.             GET_HIGH_WORD(hy,y);
  161.             SET_HIGH_WORD(y,hy+((k+1000)<<20)); /* add k to y's exponent */
  162.             return y*twom1000;
  163.         }
  164. }
  165.  
  166. #endif /* defined(_DOUBLE_IS_32BITS) */
  167.