<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Marc:<br>
<br>
_num.scm has some code to deal with computing the correct inexact
real for (sqrt x) when x is an exact number and (sqrt x) is not an
exact number.<br>
<br>
The simplest thing to do would be (sqrt (exact->inexact x)), but
I feared that the double rounding could lead to incorrect (but very
close) answers. (There are also issues of unnecessary overflows and
underflows with that simple expression.)<br>
<br>
So I ran some random tests to see (a) whether the Gambit code always
gave correctly-rounded answers (it did), and (b) how often it
happened that double rounding caused (sqrt (exact->inexact x)) to
give incorrect answers.<br>
<br>
In just 100 tests of (sqrt x) where x is a rational with numerator
and denominator between 1 and 100, the tests found the following
examples where (sqrt (exact->inexact x)) gives incorrect
answers. In each row the numbers are: x, (sqrt (exact->inexact
x)), and correctly-rounded (sqrt x):<br>
<br>
<blockquote><tt>(double-rounded-result-wrong 9/70 .35856858280031806
.3585685828003181)</tt><br>
<tt>(double-rounded-result-wrong 33/49 .8206518066482897
.8206518066482898)</tt><br>
<tt>(double-rounded-result-wrong 87/91 .977775002771065
.9777750027710649)</tt><br>
<tt>(double-rounded-result-wrong 100/67 1.2216944435630523
1.2216944435630521)</tt><br>
<tt>(double-rounded-result-wrong 77/18 2.068278940998476
2.0682789409984763)</tt><br>
<tt>(double-rounded-result-wrong 3/7 .6546536707079771
.6546536707079772)</tt><br>
<tt>(double-rounded-result-wrong 48/7 2.6186146828319083
2.618614682831909)</tt><br>
<tt>(double-rounded-result-wrong 20/51 .6262242910851494
.6262242910851495)</tt><br>
<tt>(double-rounded-result-wrong 25/11 1.5075567228888183
1.507556722888818)</tt><br>
<tt>(double-rounded-result-wrong 38/59 .8025383458814721
.8025383458814722)</tt><br>
<tt>(double-rounded-result-wrong 3/7 .6546536707079771
.6546536707079772)</tt><br>
<tt>(double-rounded-result-wrong 25/11 1.5075567228888183
1.507556722888818)</tt><br>
</blockquote>
<br>
So even for something as simple as (sqrt 3/7) the formula (sqrt
(exact->inexact 3/7)) gives an unnecessarily inaccurate answer.<br>
<br>
And for integers we have the example<br>
<br>
<blockquote><tt>(double-rounded-result-wrong 5680341405563297078633
7.536803968237e10 7.536803968236998e10)</tt><br>
</blockquote>
<br>
Pretty cool. Maybe not so important, but still cool!<br>
<br>
Brad<br>
</body>
</html>