Spaces:
Runtime error
Runtime error
| <!-- | |
| Copyright 2000-2022 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. | |
| --> | |
| <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 GNU 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; | |
| } | |
| /* 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; | |
| } | |
| .highlight { background: #dddddd } | |
| .nowrap { white-space: nowrap } | |
| /* | |
| dl.faq { counter-reset: faq } | |
| dl.faq dt:before | |
| { | |
| counter-increment: faq; | |
| content: counter(faq) ". "; | |
| } | |
| */ | |
| dl.faq dt { background: #dddddd } | |
| dl.faq dd | |
| { | |
| border-left: 4px solid; | |
| border-color: transparent; | |
| margin-left: 0em; | |
| padding-left: 1.5em; | |
| } | |
| dl.faq dt:target + dd | |
| { | |
| border-left-color: #aaaaaa; | |
| } | |
| var.env { font-style: normal } | |
| /*]]>*/</style> | |
| </head> | |
| <body> | |
| <h1>Frequently Asked Questions about <cite><acronym>GNU</acronym> <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="https://www.mpfr.org/faq.html">https://www.mpfr.org/faq.html</a>. | |
| Please look at this version if possible.</p> | |
| <ol> | |
| <li><a href="#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>?</a></li> | |
| <li><a href="#mpf2mpfr">How to convert my program written using | |
| <cite><acronym>MPF</acronym></cite> to | |
| <cite><acronym>MPFR</acronym></cite>?</a></li> | |
| <li><a href="#no_libgmp">At configure time, I get the error: <q>libgmp not found or uses a different ABI.</q></a></li> | |
| <li><a href="#undef_ref1">I get undefined reference to <code>__gmp_get_memory_functions</code>.</a></li> | |
| <li><a href="#undef_ref2">When I link my program with | |
| <cite><acronym>MPFR</acronym></cite>, I get undefined reference | |
| to <code>__gmpXXXX</code>.</a></li> | |
| <li><a href="#crash_high_prec">My program crashes with high precisions.</a></li> | |
| <li><a href="#accuracy">Though I have increased the precision, the results | |
| are not more accurate.</a></li> | |
| <li><a href="#detect_mpfr">How can I detect <cite><acronym>MPFR</acronym></cite> | |
| installation using <cite>autoconf</cite> or <cite>pkg-config</cite>?</a></li> | |
| <li><a href="#cite">How to cite <cite><acronym>MPFR</acronym></cite> in a | |
| scientific publication?</a></li> | |
| <li><a href="#fpic">When I build <cite><acronym>MPFR</acronym></cite>, I get | |
| an error asking me to recompile with <samp>-fPIC</samp>.</a></li> | |
| </ol> | |
| <dl class="faq"> | |
| <dt id="mpfr_vs_mpf">1. 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>As a consequence, <cite><acronym>MPFR</acronym></cite> uses a | |
| base-2 exponent, whereas in <cite><acronym>MPF</acronym></cite>, this | |
| is a base-2<sup>32</sup> or base-2<sup>64</sup> exponent, depending on | |
| the limb size. For this reason (and other internal ones), the maximum | |
| exponent range in <cite><acronym>MPFR</acronym></cite> is different | |
| (and smaller, if the exponent is represented by the same type as in | |
| <cite><acronym>MPF</acronym></cite>).</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">2. 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>MPFR_RNDN</code> for rounding to nearest, | |
| <code>MPFR_RNDZ</code> for rounding toward zero, <code>MPFR_RNDU</code> | |
| for rounding toward positive infinity, <code>MPFR_RNDD</code> for rounding | |
| toward negative infinity). You can also define macros as follows: | |
| <code class="block-code">#define mpf_add(a, b, c) mpfr_add(a, b, c, MPFR_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. If the program uses <cite><acronym>MPF</acronym></cite> | |
| internals (such as direct access to <code>__mpf_struct</code> members), | |
| additional changes will be needed.</p></dd> | |
| <dt id="no_libgmp">3. At configure time, I get the error: <q>libgmp not found or uses a different ABI.</q></dt> | |
| <dd><p>This test (<samp>checking for __gmpz_init in -lgmp</samp>) comes | |
| after the <samp>gmp.h</samp> detection. The failure occurs either because | |
| the <cite><acronym>GMP</acronym></cite> library could not be found | |
| (as it is not in the provided library search paths) or because the | |
| <cite><acronym>GMP</acronym></cite> library that was found does not have | |
| the expected <acronym title="Application Binary Interface">ABI</acronym> | |
| (<abbr>e.g.</abbr> 32-bit <abbr>vs</abbr> 64-bit). The former problem can be | |
| due to the fact that a static build of <cite><acronym>MPFR</acronym></cite> | |
| was requested while only a shared <cite><acronym>GMP</acronym></cite> library | |
| is installed (or the opposite, but another error can also show up in this | |
| case, see the <a href="#fpic">question about <samp>-fPIC</samp></a>). The | |
| latter problem can have several causes:</p> | |
| <ul> | |
| <li>A wrong libgmp library has been picked up. This can occur if you have | |
| several <cite><acronym>GMP</acronym></cite> versions installed on the | |
| machine and something is wrong with the provided library search paths.</li> | |
| <li>Wrong compiler options (<samp>CFLAGS</samp>) were given. In general, the | |
| presence or absence of the <samp>-m64</samp> compiler option must match the | |
| library <acronym title="Application Binary Interface">ABI</acronym>.</li> | |
| <li>A wrong <samp>gmp.h</samp> file has been picked up (if you have several | |
| <cite><acronym>GMP</acronym></cite> versions installed). Indeed, by default, | |
| <cite><acronym>MPFR</acronym></cite> gets the compiler options from the | |
| <samp>gmp.h</samp> file (with <cite><acronym>GMP</acronym></cite> 4.2.3 | |
| or later); this is needed because <cite><acronym>GMP</acronym></cite> does | |
| not necessarily use the default <acronym>ABI</acronym>. The consequence is | |
| that if the <samp>gmp.h</samp> file is associated with a library using a | |
| different <acronym>ABI</acronym>, the <acronym>ABI</acronym>-related options | |
| will be incorrect. Hence the failure.</li> | |
| </ul> | |
| <p>Note: The <samp>config.log</samp> output gives more information | |
| than the error message. In particular, see the output of the test: | |
| <samp>checking for CC and CFLAGS in gmp.h</samp>; it should give you | |
| the default compiler options (from <samp>gmp.h</samp>).</p> | |
| <p>See also the answer to the <a href="#undef_ref1">next question</a>.</p></dd> | |
| <dt id="undef_ref1">4. I get undefined reference to <code>__gmp_get_memory_functions</code>.</dt> | |
| <dd><p>Note: this was mainly a problem when upgrading from | |
| <cite><acronym>GMP</acronym></cite> 4.1.4 to a later version, | |
| but information given below may still be useful in other cases, | |
| when several <cite><acronym>GMP</acronym></cite> libraries are | |
| installed on the same machine.</p> | |
| <p>If you get such an error, in particular when running | |
| <samp>make check</samp>, then this probably means that you are using | |
| the header file from <cite><acronym>GMP</acronym></cite> 4.2.x but the | |
| <cite><acronym>GMP</acronym></cite> 4.1.4 library. This can happen if | |
| several <cite><acronym>GMP</acronym></cite> versions are installed on | |
| your machine (<abbr>e.g.</abbr>, one provided by the system in | |
| <samp>/usr/{include,lib}</samp> and a new one installed by the owner or | |
| administrator of the machine in <samp>/usr/local/{include,lib}</samp>) | |
| and your include and library search paths are inconsistent. On various | |
| <acronym>GNU</acronym>/Linux machines, this is unfortunately the case | |
| by default (<samp>/usr/local/include</samp> is in the default include | |
| search path, but <samp>/usr/local/lib</samp> is <em>not</em> in the | |
| default library search path). Typical errors are: | |
| <samp class="block-code">undefined reference to `__gmp_get_memory_functions'</samp> | |
| in <samp>make check</samp>. The best solution is to add | |
| <samp>/usr/local/include</samp> to your <var class="env">C_INCLUDE_PATH</var> | |
| environment variable and to add <samp>/usr/local/lib</samp> to your | |
| <var class="env">LIBRARY_PATH</var> and <var class="env">LD_LIBRARY_PATH</var> | |
| environment variables (and/or <var class="env">LD_RUN_PATH</var>). | |
| Alternatively, you can use <samp>--with-gmp*</samp> configure options, | |
| <abbr>e.g.</abbr> <samp>--with-gmp=/usr/local</samp>, but <strong>this is | |
| not guaranteed to work</strong> (in particular with <samp>gcc</samp> and | |
| system directories such as <samp>/usr</samp> or <samp>/usr/local</samp>), | |
| and other software that uses <cite><acronym>GMP</acronym></cite> and/or | |
| <cite><acronym>MPFR</acronym></cite> will need correct paths too; | |
| environment variables allow you to set them in a global way.</p> | |
| <p>Other information can be given in the <samp>INSTALL</samp> file and | |
| <samp>ld</samp> manual. Please look at them for more details. See also | |
| the <a href="#undef_ref2">next question</a>.</p></dd> | |
| <dt id="undef_ref2">5. 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 in any order inside a group: | |
| <samp class="block-code">gcc link.c -Wl,--start-group libgmp.a libmpfr.a -Wl,--end-group</samp> | |
| with the full path to the libraries, or | |
| <samp class="block-code">gcc link.c -Wl,--start-group -l:libgmp.a -l:libmpfr.a -Wl,--end-group</samp> | |
| to use the library path. | |
| See <samp>INSTALL</samp> file and <samp>ld</samp> manual for more | |
| details.</p> | |
| <p>If you used correct link options, but still get an error, this may mean | |
| that your include and library search paths are inconsistent. Please see the | |
| <a href="#undef_ref1">previous question</a>.</p></dd> | |
| <dt id="crash_high_prec">6. 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">7. 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, MPFR_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">8. 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. | |
| Limited <cite>pkg-config</cite> support has been added for | |
| <cite><acronym>MPFR</acronym></cite>�4.0.0; example:</p> | |
| <pre style="margin-left: 2em">cc myprogram.c $(pkg-config --cflags --libs mpfr)</pre></dd> | |
| <dt id="cite">9. How to cite <cite><acronym>MPFR</acronym></cite> in a | |
| scientific publication?</dt> | |
| <dd><p>To properly cite <cite><acronym>MPFR</acronym></cite> in a scientific | |
| publication, please cite the | |
| <a href="https://doi.org/10.1145/1236463.1236468"><acronym title="Association for Computing Machinery">ACM</acronym> | |
| <acronym title="Transactions on Mathematical Software">TOMS</acronym> | |
| paper</a> | |
| and/or the library web page | |
| <a href="https://www.mpfr.org/">https://www.mpfr.org</a>. If your publication | |
| is related to a particular release of <cite><acronym>MPFR</acronym></cite>, | |
| for example if you report timings, please also indicate the release number | |
| for future reference.</p></dd> | |
| <dt id="fpic">10. When I build <cite><acronym>MPFR</acronym></cite>, I get | |
| an error asking me to recompile with <samp>-fPIC</samp>.</dt> | |
| <dd><p>A typical error looks like:</p> | |
| <p><tt>/usr/bin/ld: <em>/path/to/</em>libgmp.a(realloc.o): relocation | |
| R_X86_64_32 against `.rodata.str1.1' can not be used when making a | |
| shared object; recompile with -fPIC<br /> | |
| <em>/path/to/</em>libgmp.a: could not read symbols: Bad value<br /> | |
| collect2: ld returned 1 exit status</tt></p> | |
| <p>The probable reason is that you tried to build | |
| <cite><acronym>MPFR</acronym></cite> with the shared library enabled (this | |
| is the default), while only a static <cite><acronym>GMP</acronym></cite> | |
| library could be found. To solve this problem, either rebuild and reinstall | |
| <cite><acronym>GMP</acronym></cite> without the <samp>--disable-shared</samp> | |
| configure option, or configure <cite><acronym>MPFR</acronym></cite> with | |
| <samp>--disable-shared</samp>. If you did this and still get the above | |
| error, the cause may be conflicting <cite><acronym>GMP</acronym></cite> | |
| versions installed on your system; please check that your search path | |
| settings are correct.</p> | |
| <p>Additional note about the last sentence: Under <acronym>GNU</acronym>/Linux | |
| (for instance), the linker takes the first library found in the library search | |
| path, whether it is dynamic or static. The default behavior under darwin is | |
| different, but <cite><acronym>MPFR</acronym></cite> will change it.</p></dd> | |
| <!-- Reference concerning darwin: see MPFR_LD_SEARCH_PATHS_FIRST | |
| in MPFR's configure.{ac,in} and acinclude.m4 --> | |
| </dl> | |
| </body> | |
| </html> | |