view libquadmath/math/x2y2m1q.c @ 158:494b0b89df80 default tip

author Shinji KONO <>
date Mon, 25 May 2020 18:13:55 +0900
parents 1830386684a0
line wrap: on
line source

/* Compute x^2 + y^2 - 1, without large cancellation error.
   Copyright (C) 2012-2018 Free Software Foundation, Inc.
   This file is part of the GNU C Library.

   The GNU C Library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
   License as published by the Free Software Foundation; either
   version 2.1 of the License, or (at your option) any later version.

   The GNU C Library is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public
   License along with the GNU C Library; if not, see
   <>.  */

#include "quadmath-imp.h"

/* Calculate X + Y exactly and store the result in *HI + *LO.  It is
   given that |X| >= |Y| and the values are small enough that no
   overflow occurs.  */

static inline void
add_split (__float128 *hi, __float128 *lo, __float128 x, __float128 y)
  /* Apply Dekker's algorithm.  */
  *hi = x + y;
  *lo = (x - *hi) + y;

/* Compare absolute values of floating-point values pointed to by P
   and Q for qsort.  */

static int
compare (const void *p, const void *q)
  __float128 pld = fabsq (*(const __float128 *) p);
  __float128 qld = fabsq (*(const __float128 *) q);
  if (pld < qld)
    return -1;
  else if (pld == qld)
    return 0;
    return 1;

/* Return X^2 + Y^2 - 1, computed without large cancellation error.
   It is given that 1 > X >= Y >= epsilon / 2, and that X^2 + Y^2 >=
   0.5.  */

__quadmath_x2y2m1q (__float128 x, __float128 y)
  __float128 vals[5];
  mul_splitq (&vals[1], &vals[0], x, x);
  mul_splitq (&vals[3], &vals[2], y, y);
  vals[4] = -1;
  qsort (vals, 5, sizeof (__float128), compare);
  /* Add up the values so that each element of VALS has absolute value
     at most equal to the last set bit of the next nonzero
     element.  */
  for (size_t i = 0; i <= 3; i++)
      add_split (&vals[i + 1], &vals[i], vals[i + 1], vals[i]);
      qsort (vals + i + 1, 4 - i, sizeof (__float128), compare);
  /* Now any error from this addition will be small.  */
  return vals[4] + vals[3] + vals[2] + vals[1] + vals[0];