xref: /openbsd-src/lib/libm/man/exp.3 (revision b2ea75c1b17e1a9a339660e7ed45cd24946b230e)
1.\"	$OpenBSD: exp.3,v 1.10 2000/04/15 02:15:28 aaron Exp $
2.\"
3.\" Copyright (c) 1985, 1991 Regents of the University of California.
4.\" All rights reserved.
5.\"
6.\" Redistribution and use in source and binary forms, with or without
7.\" modification, are permitted provided that the following conditions
8.\" are met:
9.\" 1. Redistributions of source code must retain the above copyright
10.\"    notice, this list of conditions and the following disclaimer.
11.\" 2. Redistributions in binary form must reproduce the above copyright
12.\"    notice, this list of conditions and the following disclaimer in the
13.\"    documentation and/or other materials provided with the distribution.
14.\" 3. All advertising materials mentioning features or use of this software
15.\"    must display the following acknowledgement:
16.\"	This product includes software developed by the University of
17.\"	California, Berkeley and its contributors.
18.\" 4. Neither the name of the University nor the names of its contributors
19.\"    may be used to endorse or promote products derived from this software
20.\"    without specific prior written permission.
21.\"
22.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32.\" SUCH DAMAGE.
33.\"
34.\"     from: @(#)exp.3	6.12 (Berkeley) 7/31/91
35.\"
36.Dd July 31, 1991
37.Dt EXP 3
38.Os
39.Sh NAME
40.Nm exp ,
41.Nm expf ,
42.Nm expm1 ,
43.Nm expm1f ,
44.Nm log ,
45.Nm logf ,
46.Nm log10 ,
47.Nm log10f ,
48.Nm log1p ,
49.Nm log1pf ,
50.Nm pow ,
51.Nm powf
52.Nd exponential, logarithm, power functions
53.Sh SYNOPSIS
54.Fd #include <math.h>
55.Ft double
56.Fn exp "double x"
57.Ft float
58.Fn expf "float x"
59.Ft double
60.Fn expm1 "double x"
61.Ft float
62.Fn expm1f "float x"
63.Ft double
64.Fn log "double x"
65.Ft float
66.Fn logf "float x"
67.Ft double
68.Fn log10 "double x"
69.Ft float
70.Fn log10f "float x"
71.Ft double
72.Fn log1p "double x"
73.Ft float
74.Fn log1pf "float x"
75.Ft double
76.Fn pow "double x" "double y"
77.Ft float
78.Fn powf "float x" "float y"
79.Sh DESCRIPTION
80The
81.Fn exp
82function computes the exponential value of the given argument
83.Fa x .
84.Pp
85The
86.Fn expm1
87function computes the value exp(x)\-1 accurately even for tiny argument
88.Fa x .
89.Pp
90The
91.Fn log
92function computes the value of the natural logarithm of argument
93.Fa x .
94.Pp
95The
96.Fn log10
97function computes the value of the logarithm of argument
98.Fa x
99to base 10.
100.Pp
101The
102.Fn log1p
103function computes
104the value of log(1+x) accurately even for tiny argument
105.Fa x .
106.Pp
107The
108.Fn pow
109computes the value
110of
111.Ar x
112to the exponent
113.Ar y .
114.Sh ERROR (due to Roundoff etc.)
115exp(x), log(x), expm1(x) and log1p(x) are accurate to within
116an
117.Em ulp ,
118and log10(x) to within about 2
119.Em ulps ;
120an
121.Em ulp
122is one
123.Em Unit
124in the
125.Em Last
126.Em Place .
127The error in
128.Fn pow x y
129is below about 2
130.Em ulps
131when its
132magnitude is moderate, but increases as
133.Fn pow x y
134approaches
135the over/underflow thresholds until almost as many bits could be
136lost as are occupied by the floating\-point format's exponent
137field; that is 8 bits for
138.Tn "VAX D"
139and 11 bits for IEEE 754 Double.
140No such drastic loss has been exposed by testing; the worst
141errors observed have been below 20
142.Em ulps
143for
144.Tn "VAX D" ,
145300
146.Em ulps
147for
148.Tn IEEE
149754 Double.
150Moderate values of
151.Fn pow
152are accurate enough that
153.Fn pow integer integer
154is exact until it is bigger than 2**56 on a
155.Tn VAX ,
1562**53 for
157.Tn IEEE
158754.
159.Sh RETURN VALUES
160These functions will return the appropriate computation unless an error
161occurs or an argument is out of range.
162The functions
163.Fn exp ,
164.Fn expm1
165and
166.Fn pow
167detect if the computed value will overflow,
168set the global variable
169.Va errno to
170.Er ERANGE
171and cause a reserved operand fault on a
172.Tn VAX
173or
174.Tn Tahoe .
175The function
176.Fn pow x y
177checks to see if
178.Fa x
179< 0 and
180.Fa y
181is not an integer, in the event this is true,
182the global variable
183.Va errno
184is set to
185.Er EDOM
186and on the
187.Tn VAX
188and
189.Tn Tahoe
190generate a reserved operand fault.
191On a
192.Tn VAX
193and
194.Tn Tahoe ,
195.Va errno
196is set to
197.Er EDOM
198and the reserved operand is returned
199by log unless
200.Fa x
201> 0, by
202.Fn log1p
203unless
204.Fa x
205> \-1.
206.Sh NOTES
207The functions exp(x)\-1 and log(1+x) are called
208expm1 and logp1 in
209.Tn BASIC
210on the Hewlett\-Packard
211.Tn HP Ns \-71B
212and
213.Tn APPLE
214Macintosh,
215.Tn EXP1
216and
217.Tn LN1
218in Pascal, exp1 and log1 in C
219on
220.Tn APPLE
221Macintoshes, where they have been provided to make
222sure financial calculations of ((1+x)**n\-1)/x, namely
223expm1(n\(**log1p(x))/x, will be accurate when x is tiny.
224They also provide accurate inverse hyperbolic functions.
225.Pp
226The function
227.Fn pow x 0
228returns x**0 = 1 for all x including x = 0,
229.if n \
230Infinity
231.if t \
232\(if
233(not found on a
234.Tn VAX ) ,
235and
236.Em NaN
237(the reserved
238operand on a
239.Tn VAX ) .  Previous implementations of pow may
240have defined x**0 to be undefined in some or all of these
241cases.  Here are reasons for returning x**0 = 1 always:
242.Bl -enum -width indent
243.It
244Any program that already tests whether x is zero (or
245infinite or \*(Na) before computing x**0 cannot care
246whether 0**0 = 1 or not. Any program that depends
247upon 0**0 to be invalid is dubious anyway since that
248expression's meaning and, if invalid, its consequences
249vary from one computer system to another.
250.It
251Some Algebra texts (e.g., Sigler's) define x**0 = 1 for
252all x, including x = 0.
253This is compatible with the convention that accepts a[0]
254as the value of polynomial
255.Bd -literal -offset indent
256p(x) = a[0]\(**x**0 + a[1]\(**x**1 + a[2]\(**x**2 +...+ a[n]\(**x**n
257.Ed
258.Pp
259at x = 0 rather than reject a[0]\(**0**0 as invalid.
260.It
261Analysts will accept 0**0 = 1 despite that x**y can
262approach anything or nothing as x and y approach 0
263independently.
264The reason for setting 0**0 = 1 anyway is this:
265.Bd -filled -offset indent
266If x(z) and y(z) are
267.Em any
268functions analytic (expandable
269in power series) in z around z = 0, and if there
270x(0) = y(0) = 0, then x(z)**y(z) \(-> 1 as z \(-> 0.
271.Ed
272.It
273If 0**0 = 1, then
274.if n \
275infinity**0 = 1/0**0 = 1 too; and
276.if t \
277\(if**0 = 1/0**0 = 1 too; and
278then \*(Na**0 = 1 too because x**0 = 1 for all finite
279and infinite x, i.e., independently of x.
280.El
281.Sh SEE ALSO
282.Xr infnan 3 ,
283.Xr math 3
284.Sh HISTORY
285A
286.Fn exp ,
287.Fn log
288and
289.Fn pow
290functions
291appeared in
292.At v6 .
293A
294.Fn log10
295function
296appeared in
297.At v7 .
298The
299.Fn log1p
300and
301.Fn expm1
302functions appeared in
303.Bx 4.3 .
304