GCC Code Coverage Report
Directory: ./ Exec Total Coverage
File: lib/libm/src/s_ctanl.c Lines: 0 43 0.0 %
Date: 2017-11-07 Branches: 0 8 0.0 %

Line Branch Exec Source
1
/*	$OpenBSD: s_ctanl.c,v 1.3 2011/07/20 21:02:51 martynas Exp $	*/
2
3
/*
4
 * Copyright (c) 2008 Stephen L. Moshier <steve@moshier.net>
5
 *
6
 * Permission to use, copy, modify, and distribute this software for any
7
 * purpose with or without fee is hereby granted, provided that the above
8
 * copyright notice and this permission notice appear in all copies.
9
 *
10
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17
 */
18
19
/*							ctanl()
20
 *
21
 *	Complex circular tangent
22
 *
23
 *
24
 *
25
 * SYNOPSIS:
26
 *
27
 * long double complex ctanl();
28
 * long double complex z, w;
29
 *
30
 * w = ctanl( z );
31
 *
32
 *
33
 *
34
 * DESCRIPTION:
35
 *
36
 * If
37
 *     z = x + iy,
38
 *
39
 * then
40
 *
41
 *           sin 2x  +  i sinh 2y
42
 *     w  =  --------------------.
43
 *            cos 2x  +  cosh 2y
44
 *
45
 * On the real axis the denominator is zero at odd multiples
46
 * of PI/2.  The denominator is evaluated by its Taylor
47
 * series near these points.
48
 *
49
 *
50
 * ACCURACY:
51
 *
52
 *                      Relative error:
53
 * arithmetic   domain     # trials      peak         rms
54
 *    DEC       -10,+10      5200       7.1e-17     1.6e-17
55
 *    IEEE      -10,+10     30000       7.2e-16     1.2e-16
56
 * Also tested by ctan * ccot = 1 and catan(ctan(z))  =  z.
57
 */
58
59
#include <complex.h>
60
#include <float.h>
61
#include <math.h>
62
63
#if	LDBL_MANT_DIG == 64
64
static const long double MACHEPL= 5.42101086242752217003726400434970855712890625E-20L;
65
#elif	LDBL_MANT_DIG == 113
66
static const long double MACHEPL = 9.629649721936179265279889712924636592690508e-35L;
67
#endif
68
69
static const long double PIL = 3.141592653589793238462643383279502884197169L;
70
static const long double DP1 = 3.14159265358979323829596852490908531763125L;
71
static const long double DP2 = 1.6667485837041756656403424829301998703007e-19L;
72
static const long double DP3 = 1.8830410776607851167459095484560349402753e-39L;
73
74
static long double
75
redupil(long double x)
76
{
77
	long double t;
78
	long i;
79
80
	t = x / PIL;
81
	if (t >= 0.0L)
82
		t += 0.5L;
83
	else
84
		t -= 0.5L;
85
86
	i = t;	/* the multiple */
87
	t = i;
88
	t = ((x - t * DP1) - t * DP2) - t * DP3;
89
	return (t);
90
}
91
92
static long double
93
ctansl(long double complex z)
94
{
95
	long double f, x, x2, y, y2, rn, t;
96
	long double d;
97
98
	x = fabsl(2.0L * creall(z));
99
	y = fabsl(2.0L * cimagl(z));
100
101
	x = redupil(x);
102
103
	x = x * x;
104
	y = y * y;
105
	x2 = 1.0L;
106
	y2 = 1.0L;
107
	f = 1.0L;
108
	rn = 0.0L;
109
	d = 0.0L;
110
	do {
111
		rn += 1.0L;
112
		f *= rn;
113
		rn += 1.0L;
114
		f *= rn;
115
		x2 *= x;
116
		y2 *= y;
117
		t = y2 + x2;
118
		t /= f;
119
		d += t;
120
121
		rn += 1.0L;
122
		f *= rn;
123
		rn += 1.0L;
124
		f *= rn;
125
		x2 *= x;
126
		y2 *= y;
127
		t = y2 - x2;
128
		t /= f;
129
		d += t;
130
	}
131
	while (fabsl(t/d) > MACHEPL);
132
	return(d);
133
}
134
135
long double complex
136
ctanl(long double complex z)
137
{
138
	long double complex w;
139
	long double d, x, y;
140
141
	x = creall(z);
142
	y = cimagl(z);
143
	d = cosl(2.0L * x) + coshl(2.0L * y);
144
145
	if (fabsl(d) < 0.25L) {
146
		d = fabsl(d);
147
		d = ctansl(z);
148
	}
149
	if (d == 0.0L) {
150
		/*mtherr( "ctan", OVERFLOW );*/
151
		w = LDBL_MAX + LDBL_MAX * I;
152
		return (w);
153
	}
154
155
	w = sinl(2.0L * x) / d + (sinhl(2.0L * y) / d) * I;
156
	return (w);
157
}