<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
Hello Chris:<br>
<br>
Chris Saunders wrote:<br>
<blockquote cite="mid001401c72509$e9829110$28191347@mycomputer"
type="cite">
<pre wrap="">Hi Carlos
I am not a GMP expert but I am a programmer who uses GMP.
As to your first question GMP has three number types integer, real and
rational each of these types has its own separate representation in memory.
Each of these types can consume as much memory as the particular computer in
use will allow.
</pre>
</blockquote>
Well... it is not obvious to me the relationship between available
memory and the way a number is represented. <br>
To clarify a little bit: I am looking for something like "...<i> a GMP
real is represented with a mantissa of NMANT bits and an exponent of
NEXP bits, being NMANT user defined, and NEXP defined at compile time</i>..."<br>
<blockquote cite="mid001401c72509$e9829110$28191347@mycomputer"
type="cite">
<pre wrap="">
I am not certain of the answer to the second question but I believe the
answer is no.
</pre>
</blockquote>
OK, but how to estimate a reliable tolerance given the (yet unknown to
me) user defined parameters?<br>
<blockquote cite="mid001401c72509$e9829110$28191347@mycomputer"
type="cite">
<pre wrap="">
For the third question the answer is that the more memory your computer has
then the largest floating point number can be larger.
</pre>
</blockquote>
This might be true or not, depending on the way they are represented.
To say something, the largest representable double precision number is
unrelated with the available memory. Another thing is the largest total
number of double precision number which might fit into memory.<br>
I might agree that the above looks confusing: I am looking for the
Overflow Level (largest finite and positive representable number), the
Underflow level (i.e. smallest positive, non zero representable
number), etc. all of them as functions of the (yet unkown to me) user
defined parameters (like NMANT, NEXP, etc.)<br>
Regards<br>
Carlos<br>
<blockquote cite="mid001401c72509$e9829110$28191347@mycomputer"
type="cite">
<pre wrap="">
I hope this is clear enough and helpful.
Regards,
Chris Saunders
-----Original Message-----
From: <a class="moz-txt-link-abbreviated" href="mailto:gmp-discuss-bounces@swox.com">gmp-discuss-bounces@swox.com</a> [<a class="moz-txt-link-freetext" href="mailto:gmp-discuss-bounces@swox.com">mailto:gmp-discuss-bounces@swox.com</a>] On
Behalf Of Dr. Ing. Carlos López
Sent: Thursday, December 21, 2006 8:16 AM
To: <a class="moz-txt-link-abbreviated" href="mailto:gmp-discuss@swox.com">gmp-discuss@swox.com</a>
Subject: Some basic questions
Hello everybody:
I would like to know if there is a place where to find some basic
information
about GMP. By "basic" I meant non-programmer info, but general one.
I am looking for:
.- How a number is represented in GMP
.- Is there something equivalent to the machine epsilon in GMP? If not, how
may I estimate a "tolerance" from the parameters which define the GMP
environment?
.- Which is the largest representable floating point? An the underflow
level?
Regards
Carlos
_______________________________________________
gmp-discuss mailing list
<a class="moz-txt-link-abbreviated" href="mailto:gmp-discuss@swox.com">gmp-discuss@swox.com</a>
<a class="moz-txt-link-freetext" href="https://gmplib.org/mailman/listinfo/gmp-discuss">https://gmplib.org/mailman/listinfo/gmp-discuss</a>
</pre>
</blockquote>
</body>
</html>