summaryrefslogtreecommitdiff
path: root/src/atanu.c
blob: 57c9083eaf1c72712cb490fc9e9bb0aa3721c248 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
/* mpfr_atanu  -- atanu(x)  = atan(x)*u/(2*pi)
   mpfr_atanpi -- atanpi(x) = atan(x)/pi

Copyright 2021 Free Software Foundation, Inc.
Contributed by the AriC and Caramba projects, INRIA.

This file is part of the GNU MPFR Library.

The GNU MPFR 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 3 of the License, or (at your
option) any later version.

The GNU MPFR Library is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
License for more details.

You should have received a copy of the GNU Lesser General Public License
along with the GNU MPFR Library; see the file COPYING.LESSER.  If not, see
https://www.gnu.org/licenses/ or write to the Free Software Foundation, Inc.,
51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA. */

#define MPFR_NEED_LONGLONG_H
#include "mpfr-impl.h"

/* put in y the correctly rounded value of atan(x)*u/(2*pi) */
int
mpfr_atanu (mpfr_ptr y, mpfr_srcptr x, unsigned long u, mpfr_rnd_t rnd_mode)
{
  mpfr_t tmp, pi;
  mpfr_prec_t prec;
  mpfr_exp_t expx;
  int inex;
  MPFR_SAVE_EXPO_DECL (expo);
  MPFR_ZIV_DECL (loop);

  MPFR_LOG_FUNC
    (("x[%Pu]=%.*Rg u=%lu rnd=%d", mpfr_get_prec(x), mpfr_log_prec, x, u,
      rnd_mode),
     ("y[%Pu]=%.*Rg inexact=%d", mpfr_get_prec (y), mpfr_log_prec, y,
      inex));

  /* Singular cases */
  if (MPFR_UNLIKELY (MPFR_IS_SINGULAR (x)))
    {
      if (MPFR_IS_NAN (x))
        {
          MPFR_SET_NAN (y);
          MPFR_RET_NAN;
        }
      else if (MPFR_IS_INF (x))
        {
          /* atanu(+Inf,u) = u/4, atanu(-Inf,u) = -u/4 */
          if (MPFR_IS_POS (x))
            return mpfr_set_ui_2exp (y, u, -2, rnd_mode);
          else
            {
              inex = mpfr_set_ui_2exp (y, u, -2, MPFR_INVERT_RND (rnd_mode));
              MPFR_CHANGE_SIGN (y);
              return -inex;
            }
        }
      else /* necessarily x=0 */
        {
          MPFR_ASSERTD(MPFR_IS_ZERO(x));
          /* atan(0)=0 with same sign, even when u=0 to ensure
             atanu(-x,u) = -atanu(x,u) */
          MPFR_SET_ZERO (y);
          MPFR_SET_SAME_SIGN (y, x);
          MPFR_RET (0); /* exact result */
        }
    }

  if (u == 0) /* return 0 with sign of x, which is coherent with case x=0 */
    {
      MPFR_SET_ZERO (y);
      MPFR_SET_SAME_SIGN (y, x);
      MPFR_RET (0);
    }

  if (mpfr_cmpabs_ui (x, 1) == 0)
    {
      /* |x| = 1: atanu(1,u) = u/8, atanu(-1,u)=-u/8 */
      /* we can't use mpfr_set_si_2exp with -u since -u might not be
         representable as long */
      if (MPFR_SIGN(x) > 0)
        return mpfr_set_ui_2exp (y, u, -3, rnd_mode);
      else
        {
          inex = mpfr_set_ui_2exp (y, u, -3, MPFR_INVERT_RND(rnd_mode));
          MPFR_CHANGE_SIGN(y);
          return -inex;
        }
    }

  /* For x>=1, we have pi/2-1/x < atan(x) < pi/2, thus
     u/4-u/(2*pi*x) < atanu(x,u) < u/4, and the relative difference between
     atanu(x,u) and u/4 is less than 2/(pi*x) < 1/x <= 2^(1-EXP(x)).
     If the relative difference is <= 2^(-prec-2), then the difference
     between atanu(x,u) and u/4 is <= 1/4*ulp(u/4) <= 1/2*ulp(RN(u/4)).
     We also require x >= 2^64, which implies x > 2*u/pi, so that
     (u-1)/4 < u/4-u/(2*pi*x) < u/4. */
  expx = MPFR_GET_EXP(x);
  if (expx >= 65 && expx - 1 >= MPFR_PREC(y) + 2)
    {
      prec = (MPFR_PREC(y) <= 63) ? 65 : MPFR_PREC(y) + 2;
      /* now prec > 64 and prec > MPFR_PREC(y)+1 */
      mpfr_init2 (tmp, prec);
      /* since expx >= 65, we have emax >= 65, thus u is representable here,
         and we don't need to work in an extended exponent range */
      inex = mpfr_set_ui (tmp, u, MPFR_RNDN); /* exact since prec >= 64 */
      MPFR_ASSERTD(inex == 0);
      mpfr_nextbelow (tmp);
      /* Since prec >= 65, the last significant bit of tmp is 1, and since
         prec > PREC(y), tmp is not representable in the target precision,
         which ensures we will get a correct ternary value below. */
      MPFR_ASSERTD(mpfr_min_prec(tmp) > MPFR_PREC(y));
      if (MPFR_SIGN(x) < 0)
        MPFR_CHANGE_SIGN(tmp);
      /* since prec >= PREC(y)+2, the rounding of tmp is correct */
      inex = mpfr_div_2ui (y, tmp, 2, rnd_mode);
      mpfr_clear (tmp);
      return inex;
    }

  prec = MPFR_PREC (y);

  MPFR_SAVE_EXPO_MARK (expo);

  prec += MPFR_INT_CEIL_LOG2(prec) + 10;

  mpfr_init2 (tmp, prec);
  mpfr_init2 (pi, prec);

  MPFR_ZIV_INIT (loop, prec);
  for (;;)
    {
      /* In the error analysis below, each thetax denotes a variable such that
         |thetax| <= 2^(1-prec) */
      mpfr_atan (tmp, x, MPFR_RNDA);
      /* tmp = atan(x) * (1 + theta1), and tmp cannot be zero since we rounded
         away from zero, and the case x=0 was treated before */
      /* first multiply by u to avoid underflow issues */
      mpfr_mul_ui (tmp, tmp, u, MPFR_RNDA);
      /* tmp = atan(x)*u * (1 + theta2)^2, and |tmp| >= 0.5*2^emin */
      mpfr_const_pi (pi, MPFR_RNDZ); /* round toward zero since we we will
                                        divide by pi, to round tmp away */
      /* pi = Pi * (1 + theta3) */
      mpfr_div (tmp, tmp, pi, MPFR_RNDA);
      /* tmp = atan(x)*u/Pi * (1 + theta4)^4, with |tmp| > 0 */
      /* since we rounded away from 0, if we get 0.5*2^emin here, it means
         |atanu(x,u)| < 0.25*2^emin (pi is not exact) thus we have underflow */
      if (MPFR_EXP(tmp) == __gmpfr_emin)
        {
          /* mpfr_underflow rounds away for RNDN */
          mpfr_clear (tmp);
          mpfr_clear (pi);
          MPFR_SAVE_EXPO_FREE (expo);
          return mpfr_underflow (y,
                            (rnd_mode == MPFR_RNDN) ? MPFR_RNDZ : rnd_mode, 1);
        }
      mpfr_div_2ui (tmp, tmp, 1, MPFR_RNDA); /* exact */
      /* tmp = atan(x)*u/(2*Pi) * (1 + theta4)^4 */
      /* since |(1 + theta4)^4 - 1| <= 8*|theta4| for prec >= 3,
         the relative error is less than 2^(4-prec) */
      MPFR_ASSERTD(!MPFR_IS_ZERO(tmp));
      if (MPFR_LIKELY (MPFR_CAN_ROUND (tmp, prec - 4,
                                       MPFR_PREC (y), rnd_mode)))
        break;
      MPFR_ZIV_NEXT (loop, prec);
      mpfr_set_prec (tmp, prec);
      mpfr_set_prec (pi, prec);
    }
  MPFR_ZIV_FREE (loop);

  inex = mpfr_set (y, tmp, rnd_mode);
  mpfr_clear (tmp);
  mpfr_clear (pi);

  MPFR_SAVE_EXPO_FREE (expo);
  return mpfr_check_range (y, inex, rnd_mode);
}

int
mpfr_atanpi (mpfr_ptr y, mpfr_srcptr x, mpfr_rnd_t rnd_mode)
{
  return mpfr_atanu (y, x, 2, rnd_mode);
}