exp10.c
Go to the documentation of this file.
1 /* exp10.c
2  *
3  * Base 10 exponential function
4  * (Common antilogarithm)
5  *
6  *
7  *
8  * SYNOPSIS:
9  *
10  * double x, y, exp10();
11  *
12  * y = exp10( x );
13  *
14  *
15  *
16  * DESCRIPTION:
17  *
18  * Returns 10 raised to the x power.
19  *
20  * Range reduction is accomplished by expressing the argument
21  * as 10**x = 2**n 10**f, with |f| < 0.5 log10(2).
22  * The Pade' form
23  *
24  * 1 + 2x P(x**2)/( Q(x**2) - P(x**2) )
25  *
26  * is used to approximate 10**f.
27  *
28  *
29  *
30  * ACCURACY:
31  *
32  * Relative error:
33  * arithmetic domain # trials peak rms
34  * IEEE -307,+307 30000 2.2e-16 5.5e-17
35  *
36  * ERROR MESSAGES:
37  *
38  * message condition value returned
39  * exp10 underflow x < -MAXL10 0.0
40  * exp10 overflow x > MAXL10 INFINITY
41  *
42  * IEEE arithmetic: MAXL10 = 308.2547155599167.
43  *
44  */
45 
46 /*
47  * Cephes Math Library Release 2.2: January, 1991
48  * Copyright 1984, 1991 by Stephen L. Moshier
49  * Direct inquiries to 30 Frost Street, Cambridge, MA 02140
50  */
51 
52 
53 #include "mconf.h"
54 
55 static double P[] = {
56  4.09962519798587023075E-2,
57  1.17452732554344059015E1,
58  4.06717289936872725516E2,
59  2.39423741207388267439E3,
60 };
61 
62 static double Q[] = {
63  /* 1.00000000000000000000E0, */
64  8.50936160849306532625E1,
65  1.27209271178345121210E3,
66  2.07960819286001865907E3,
67 };
68 
69 /* static double LOG102 = 3.01029995663981195214e-1; */
70 static double LOG210 = 3.32192809488736234787e0;
71 static double LG102A = 3.01025390625000000000E-1;
72 static double LG102B = 4.60503898119521373889E-6;
73 
74 /* static double MAXL10 = 38.230809449325611792; */
75 static double MAXL10 = 308.2547155599167;
76 
77 double exp10(double x)
78 {
79  double px, xx;
80  short n;
81 
82  if (cephes_isnan(x))
83  return (x);
84  if (x > MAXL10) {
85  return (INFINITY);
86  }
87 
88  if (x < -MAXL10) { /* Would like to use MINLOG but can't */
89  sf_error("exp10", SF_ERROR_UNDERFLOW, NULL);
90  return (0.0);
91  }
92 
93  /* Express 10**x = 10**g 2**n
94  * = 10**g 10**( n log10(2) )
95  * = 10**( g + n log10(2) )
96  */
97  px = floor(LOG210 * x + 0.5);
98  n = px;
99  x -= px * LG102A;
100  x -= px * LG102B;
101 
102  /* rational approximation for exponential
103  * of the fractional part:
104  * 10**x = 1 + 2x P(x**2)/( Q(x**2) - P(x**2) )
105  */
106  xx = x * x;
107  px = x * polevl(xx, P, 3);
108  x = px / (p1evl(xx, Q, 3) - px);
109  x = 1.0 + ldexp(x, 1);
110 
111  /* multiply by power of 2 */
112  x = ldexp(x, n);
113 
114  return (x);
115 }
cephes_isnan
#define cephes_isnan(x)
Definition: mconf.h:99
x
set noclip points set clip one set noclip two set bar set border lt lw set xdata set ydata set zdata set x2data set y2data set boxwidth set dummy x
Definition: gnuplot_common_settings.hh:12
SF_ERROR_UNDERFLOW
@ SF_ERROR_UNDERFLOW
Definition: sf_error.h:11
p1evl
static double p1evl(double x, const double coef[], int N)
Definition: polevl.h:97
n
int n
Definition: BiCGSTAB_simple.cpp:1
LG102B
static double LG102B
Definition: exp10.c:72
polevl
static double polevl(double x, const double coef[], int N)
Definition: polevl.h:65
Eigen::Quaternion
The quaternion class used to represent 3D orientations and rotations.
Definition: ForwardDeclarations.h:293
mconf.h
MAXL10
static double MAXL10
Definition: exp10.c:75
LG102A
static double LG102A
Definition: exp10.c:71
sf_error
void sf_error(const char *func_name, sf_error_t code, const char *fmt,...)
Definition: sf_error.c:41
exp10
double exp10(double x)
Definition: exp10.c:77
NULL
#define NULL
Definition: ccolamd.c:609
P
static double P[]
Definition: exp10.c:55
LOG210
static double LOG210
Definition: exp10.c:70
floor
const EIGEN_DEVICE_FUNC FloorReturnType floor() const
Definition: ArrayCwiseUnaryOps.h:481
px
RealScalar RealScalar * px
Definition: level1_cplx_impl.h:28


gtsam
Author(s):
autogenerated on Fri Nov 1 2024 03:32:30