(root)/
gcc-13.2.0/
libquadmath/
math/
casinhq_kernel.c
       1  /* Return arc hyperbolic sine for a complex float type, with the
       2     imaginary part of the result possibly adjusted for use in
       3     computing other functions.
       4     Copyright (C) 1997-2018 Free Software Foundation, Inc.
       5     This file is part of the GNU C Library.
       6  
       7     The GNU C Library is free software; you can redistribute it and/or
       8     modify it under the terms of the GNU Lesser General Public
       9     License as published by the Free Software Foundation; either
      10     version 2.1 of the License, or (at your option) any later version.
      11  
      12     The GNU C Library is distributed in the hope that it will be useful,
      13     but WITHOUT ANY WARRANTY; without even the implied warranty of
      14     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
      15     Lesser General Public License for more details.
      16  
      17     You should have received a copy of the GNU Lesser General Public
      18     License along with the GNU C Library; if not, see
      19     <http://www.gnu.org/licenses/>.  */
      20  
      21  #include "quadmath-imp.h"
      22  
      23  /* Return the complex inverse hyperbolic sine of finite nonzero Z,
      24     with the imaginary part of the result subtracted from pi/2 if ADJ
      25     is nonzero.  */
      26  
      27  __complex128
      28  __quadmath_kernel_casinhq (__complex128 x, int adj)
      29  {
      30    __complex128 res;
      31    __float128 rx, ix;
      32    __complex128 y;
      33  
      34    /* Avoid cancellation by reducing to the first quadrant.  */
      35    rx = fabsq (__real__ x);
      36    ix = fabsq (__imag__ x);
      37  
      38    if (rx >= 1 / FLT128_EPSILON || ix >= 1 / FLT128_EPSILON)
      39      {
      40        /* For large x in the first quadrant, x + csqrt (1 + x * x)
      41  	 is sufficiently close to 2 * x to make no significant
      42  	 difference to the result; avoid possible overflow from
      43  	 the squaring and addition.  */
      44        __real__ y = rx;
      45        __imag__ y = ix;
      46  
      47        if (adj)
      48  	{
      49  	  __float128 t = __real__ y;
      50  	  __real__ y = copysignq (__imag__ y, __imag__ x);
      51  	  __imag__ y = t;
      52  	}
      53  
      54        res = clogq (y);
      55        __real__ res += (__float128) M_LN2q;
      56      }
      57    else if (rx >= 0.5Q && ix < FLT128_EPSILON / 8)
      58      {
      59        __float128 s = hypotq (1, rx);
      60  
      61        __real__ res = logq (rx + s);
      62        if (adj)
      63  	__imag__ res = atan2q (s, __imag__ x);
      64        else
      65  	__imag__ res = atan2q (ix, s);
      66      }
      67    else if (rx < FLT128_EPSILON / 8 && ix >= 1.5Q)
      68      {
      69        __float128 s = sqrtq ((ix + 1) * (ix - 1));
      70  
      71        __real__ res = logq (ix + s);
      72        if (adj)
      73  	__imag__ res = atan2q (rx, copysignq (s, __imag__ x));
      74        else
      75  	__imag__ res = atan2q (s, rx);
      76      }
      77    else if (ix > 1 && ix < 1.5Q && rx < 0.5Q)
      78      {
      79        if (rx < FLT128_EPSILON * FLT128_EPSILON)
      80  	{
      81  	  __float128 ix2m1 = (ix + 1) * (ix - 1);
      82  	  __float128 s = sqrtq (ix2m1);
      83  
      84  	  __real__ res = log1pq (2 * (ix2m1 + ix * s)) / 2;
      85  	  if (adj)
      86  	    __imag__ res = atan2q (rx, copysignq (s, __imag__ x));
      87  	  else
      88  	    __imag__ res = atan2q (s, rx);
      89  	}
      90        else
      91  	{
      92  	  __float128 ix2m1 = (ix + 1) * (ix - 1);
      93  	  __float128 rx2 = rx * rx;
      94  	  __float128 f = rx2 * (2 + rx2 + 2 * ix * ix);
      95  	  __float128 d = sqrtq (ix2m1 * ix2m1 + f);
      96  	  __float128 dp = d + ix2m1;
      97  	  __float128 dm = f / dp;
      98  	  __float128 r1 = sqrtq ((dm + rx2) / 2);
      99  	  __float128 r2 = rx * ix / r1;
     100  
     101  	  __real__ res = log1pq (rx2 + dp + 2 * (rx * r1 + ix * r2)) / 2;
     102  	  if (adj)
     103  	    __imag__ res = atan2q (rx + r1, copysignq (ix + r2, __imag__ x));
     104  	  else
     105  	    __imag__ res = atan2q (ix + r2, rx + r1);
     106  	}
     107      }
     108    else if (ix == 1 && rx < 0.5Q)
     109      {
     110        if (rx < FLT128_EPSILON / 8)
     111  	{
     112  	  __real__ res = log1pq (2 * (rx + sqrtq (rx))) / 2;
     113  	  if (adj)
     114  	    __imag__ res = atan2q (sqrtq (rx), copysignq (1, __imag__ x));
     115  	  else
     116  	    __imag__ res = atan2q (1, sqrtq (rx));
     117  	}
     118        else
     119  	{
     120  	  __float128 d = rx * sqrtq (4 + rx * rx);
     121  	  __float128 s1 = sqrtq ((d + rx * rx) / 2);
     122  	  __float128 s2 = sqrtq ((d - rx * rx) / 2);
     123  
     124  	  __real__ res = log1pq (rx * rx + d + 2 * (rx * s1 + s2)) / 2;
     125  	  if (adj)
     126  	    __imag__ res = atan2q (rx + s1, copysignq (1 + s2, __imag__ x));
     127  	  else
     128  	    __imag__ res = atan2q (1 + s2, rx + s1);
     129  	}
     130      }
     131    else if (ix < 1 && rx < 0.5Q)
     132      {
     133        if (ix >= FLT128_EPSILON)
     134  	{
     135  	  if (rx < FLT128_EPSILON * FLT128_EPSILON)
     136  	    {
     137  	      __float128 onemix2 = (1 + ix) * (1 - ix);
     138  	      __float128 s = sqrtq (onemix2);
     139  
     140  	      __real__ res = log1pq (2 * rx / s) / 2;
     141  	      if (adj)
     142  		__imag__ res = atan2q (s, __imag__ x);
     143  	      else
     144  		__imag__ res = atan2q (ix, s);
     145  	    }
     146  	  else
     147  	    {
     148  	      __float128 onemix2 = (1 + ix) * (1 - ix);
     149  	      __float128 rx2 = rx * rx;
     150  	      __float128 f = rx2 * (2 + rx2 + 2 * ix * ix);
     151  	      __float128 d = sqrtq (onemix2 * onemix2 + f);
     152  	      __float128 dp = d + onemix2;
     153  	      __float128 dm = f / dp;
     154  	      __float128 r1 = sqrtq ((dp + rx2) / 2);
     155  	      __float128 r2 = rx * ix / r1;
     156  
     157  	      __real__ res = log1pq (rx2 + dm + 2 * (rx * r1 + ix * r2)) / 2;
     158  	      if (adj)
     159  		__imag__ res = atan2q (rx + r1, copysignq (ix + r2,
     160  							     __imag__ x));
     161  	      else
     162  		__imag__ res = atan2q (ix + r2, rx + r1);
     163  	    }
     164  	}
     165        else
     166  	{
     167  	  __float128 s = hypotq (1, rx);
     168  
     169  	  __real__ res = log1pq (2 * rx * (rx + s)) / 2;
     170  	  if (adj)
     171  	    __imag__ res = atan2q (s, __imag__ x);
     172  	  else
     173  	    __imag__ res = atan2q (ix, s);
     174  	}
     175        math_check_force_underflow_nonneg (__real__ res);
     176      }
     177    else
     178      {
     179        __real__ y = (rx - ix) * (rx + ix) + 1;
     180        __imag__ y = 2 * rx * ix;
     181  
     182        y = csqrtq (y);
     183  
     184        __real__ y += rx;
     185        __imag__ y += ix;
     186  
     187        if (adj)
     188  	{
     189  	  __float128 t = __real__ y;
     190  	  __real__ y = copysignq (__imag__ y, __imag__ x);
     191  	  __imag__ y = t;
     192  	}
     193  
     194        res = clogq (y);
     195      }
     196  
     197    /* Give results the correct sign for the original argument.  */
     198    __real__ res = copysignq (__real__ res, __real__ x);
     199    __imag__ res = copysignq (__imag__ res, (adj ? 1 : __imag__ x));
     200  
     201    return res;
     202  }