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
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
|
<?xml version="1.0" encoding="iso-8859-1"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<!--
Copyright 2000, 2001, 2002, 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
Contributed by the Spaces project, INRIA Lorraine.
This file is part of the MPFR Library.
The MPFR Library is free software; you can redistribute it and/or modify it
under the terms of the GNU Lesser General Public License (either version 2.1
of the License, or, at your option, any later version) and the GNU General
Public License as published by the Free Software Foundation (most of MPFR is
under the former, some under the latter).
The 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 MPFR Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston,
MA 02110-1301, USA.
-->
<html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
<head><meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
<title>Frequently Asked Questions about MPFR</title>
<style type="text/css">/*<![CDATA[*/
/* Global stylesheet for visual media */
html, body
{
background: white;
color: black;
}
div.logo { float: right }
div.logo img { border: 0 }
div.footer img { border: 0 }
dt
{
margin-top: 2ex;
margin-bottom: 1ex;
font-weight: bolder;
background: #dddddd;
}
/* For testing: dd { background: #ddddff } */
table { margin: 0.5ex auto }
li { margin-top: 0.5ex; margin-bottom: 0.5ex }
dd + dd
{
margin-top: 0;
margin-bottom: 0;
padding-top: 0.5ex;
padding-bottom: 0;
}
li > p, dd > p
{
margin-top: 0;
margin-bottom: 0;
padding-top: 0.5ex;
padding-bottom: 0.5ex;
}
.block-code code, code.block-code,
.block-code samp, samp.block-code
{
display: block;
padding: 0.5ex 0;
margin-left: 2em;
}
dl.faq { counter-reset: faq }
dl.faq dt:before
{
counter-increment: faq;
content: counter(faq) ". ";
}
/*]]>*/</style>
</head>
<body>
<h1>Frequently Asked Questions about <cite><acronym>MPFR</acronym></cite></h1>
<p><strong>Important notice: Problems with a particular version of
<cite><acronym>MPFR</acronym></cite> are discussed in the corresponding
bugs page.</strong></p>
<p>The latest version of this <acronym>FAQ</acronym> is available at
<a href="http://www.mpfr.org/faq.html">http://www.mpfr.org/faq.html</a>.
Please look at this version if possible.</p>
<dl class="faq">
<dt id="mpfr_vs_mpf">What are the differences between
<cite><acronym>MPF</acronym></cite> from <cite><acronym>GMP</acronym></cite>
and <cite><acronym>MPFR</acronym></cite>?</dt>
<dd><p>The main differences are:</p>
<ul>
<li><p>The precision of a <cite><acronym>MPFR</acronym></cite> variable
is the <em>exact</em> number of bits used for its mantissa, whereas in
<cite><acronym>MPF</acronym></cite>, the precision requested by the user
is a minimum value (<cite><acronym>MPF</acronym></cite> generally uses a
higher precision). With the additional difference below, this implies that
the <cite><acronym>MPFR</acronym></cite> results do not depend on the
number of bits (16, 32, 64 or more) of the underlying architecture.</p></li>
<li><p><cite><acronym>MPFR</acronym></cite> provides an additional rounding
mode argument to its functions; furthermore, it is guaranteed that the
result of any operation is the nearest possible floating-point value from
the exact result (considering the input variables as exact values), taking
into account the precision of the destination variable and the rounding
mode. <cite><acronym>MPFR</acronym></cite> also says whether the rounded
result is above or below the exact result.</p></li>
<li><p><cite><acronym>MPFR</acronym></cite> supports much more functions
(in particular transcendental functions such as exponentials, logarithms,
trigonometric functions and so on) and special values: signed zeros,
infinities, not-a-number (NaN).</p></li>
</ul></dd>
<dt id="mpf2mpfr">How to convert my program written using
<cite><acronym>MPF</acronym></cite> to
<cite><acronym>MPFR</acronym></cite>?</dt>
<dd><p>You need to add <q><code>r</code></q> to the function names, and to
specify the rounding mode (<code>GMP_RNDN</code> for rounding to nearest,
<code>GMP_RNDZ</code> for rounding towards zero, <code>GMP_RNDU</code>
for rounding towards plus infinity, <code>GMP_RNDD</code> for rounding
towards minus infinity). You can also define macros as follows:
<code class="block-code">#define mpf_add(a, b, c) mpfr_add(a, b, c, GMP_RNDN)</code></p>
<p>The header file <samp>mpf2mpfr.h</samp> from the
<cite><acronym>MPFR</acronym></cite> distribution automatically
redefines all <cite><acronym>MPF</acronym></cite> functions in this
way, using the default <cite><acronym>MPFR</acronym></cite> rounding
mode. Thus you simply need to add the following line in all your files
using <cite><acronym>MPF</acronym></cite> functions:
<code class="block-code">#include <mpf2mpfr.h></code>
just after the <samp>gmp.h</samp> and <samp>mpfr.h</samp>
header files.</p></dd>
<dt id="undef_ref">When I link my program with
<cite><acronym>MPFR</acronym></cite>, I get undefined reference
to <code>__gmpXXXX</code>.</dt>
<dd><p>Link your program with <cite><acronym>GMP</acronym></cite>. Assuming
that your program is <samp>foo.c</samp>, you should link it using:
<samp class="block-code">cc link.c -lmpfr -lgmp</samp>
<cite><acronym>MPFR</acronym></cite> library reference (<samp>-lmpfr</samp>)
should be before <cite><acronym>GMP</acronym></cite>'s one
(<samp>-lgmp</samp>). Another solution is, with <acronym>GNU</acronym>
<samp>ld</samp>, to give all the libraries inside a group:
<samp class="block-code">gcc link.c -Wl,--start-group libgmp.a libmpfr.a -Wl,--end-group</samp>
See <samp>INSTALL</samp> file and <samp>ld</samp> manual for more
details.</p>
<p>Moreover, if several <cite><acronym>GMP</acronym></cite> versions
are installed (<abbr>e.g.</abbr>, one provided by the system and a
new one installed by some user), you must make sure that the include
and library search paths are consistent. Unfortunately, on various
<acronym>GNU</acronym>/Linux machines, they aren't by default.
Typical errors are:
<samp class="block-code">undefined reference to `__gmp_get_memory_functions'</samp>
in <samp>make check</samp> when <cite><acronym>GMP</acronym></cite> 4.1.4
is installed in <samp>/usr/{include,lib}</samp> (provided by the system)
and <cite><acronym>GMP</acronym></cite> 4.2.1 is installed in
<samp>/usr/local/{include,lib}</samp> (installed by the user with
<samp>configure</samp>, <samp>make</samp>, <samp>make install</samp>).</p></dd>
<dt id="crash_high_prec">My program crashes with high precisions.</dt>
<dd><p>Your stack size limit may be too small; indeed, by default,
<cite><acronym>GMP</acronym></cite> 4.1.4 and below allocates all
temporary results on the stack, and in very high precisions, this
limit may be reached. You can solve this problem in different ways:</p>
<ul>
<li><p>You can upgrade to <cite><acronym>GMP</acronym></cite> 4.2 (or above),
which now makes temporary allocations on the stack only when they are
small.</p></li>
<li><p>You can increase the stack size limit with the <samp>limit</samp>,
<samp>unlimit</samp> or <samp>ulimit</samp> command, depending on your
shell. This may fail on some systems, where the maximum stack size cannot
be increased above some value.</p></li>
<li><p>You can rebuild both <cite><acronym>GMP</acronym></cite> and
<cite><acronym>MPFR</acronym></cite> to use another allocation method.</p></li>
</ul></dd>
<dt id="accuracy">Though I have increased the precision, the results
are not more accurate.</dt>
<dd><p>The reason may be the use of C floating-point numbers. If you want
to store a floating-point constant to a <code>mpfr_t</code>, you should use
<code>mpfr_set_str</code> (or one of the <cite><acronym>MPFR</acronym></cite>
constant functions, such as <code>mpfr_const_pi</code> for π) instead
of <code>mpfr_set_d</code> or <code>mpfr_set_ld</code>. Otherwise the
floating-point constant will be first converted into a reduced-precision
(<abbr>e.g.</abbr>, 53-bit) binary number before
<cite><acronym>MPFR</acronym></cite> can work with it. This is the case
in particular for most exact decimal numbers, such as 0.17, which are
not exactly representable in binary.</p>
<p>Also remember that <cite><acronym>MPFR</acronym></cite> does not track
the accuracy of the results: copying a value <var>x</var> to <var>y</var>
with <code>mpfr_set (y, x, GMP_RNDN)</code> where the variable <var>y</var>
is more precise than the variable <var>x</var> will not make it more
accurate; the (binary) value will remain unchanged.</p></dd>
<dt id="detect_mpfr">How can I detect <cite><acronym>MPFR</acronym></cite>
installation using <cite>autoconf</cite> or <cite>pkg-config</cite>?</dt>
<dd><p>The <cite><acronym>MPFR</acronym></cite> team does not currently
recommend any <cite>autoconf</cite> code, but a section will later
be added to the <cite><acronym>MPFR</acronym></cite> manual. The
<cite><acronym>MPFR</acronym></cite> team does not wish to support
<cite>pkg-config</cite> yet.</p></dd>
</dl>
</body>
</html>
|
