<rss xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd" version="2.0">
  <channel>
    <title>MaplePrimes - answers and comments on Question, A matter of trust?</title>
    <link>http://www.mapleprimes.com/questions/94834-A-Matter-Of-Trust</link>
    <language>en-us</language>
    <copyright>2026 Maplesoft, A Division of Waterloo Maple Inc.</copyright>
    <generator>Maplesoft Document System</generator>
    <lastBuildDate>Fri, 12 Jun 2026 03:43:36 GMT</lastBuildDate>
    <pubDate>Fri, 12 Jun 2026 03:43:36 GMT</pubDate>
    <itunes:subtitle />
    <itunes:summary />
    <description>The latest answers and comments added to the Question, A matter of trust?</description>
    <image>
      <url>http://www.mapleprimes.com/images/mapleprimeswhite.jpg</url>
      <title>MaplePrimes - answers and comments on Question, A matter of trust?</title>
      <link>http://www.mapleprimes.com/questions/94834-A-Matter-Of-Trust</link>
    </image>
    <item>
      <title>overlooked</title>
      <link>http://www.mapleprimes.com/questions/94834-A-Matter-Of-Trust?ref=Feed:MaplePrimes:A matter of trust?:Comments#answer94849</link>
      <itunes:summary>&lt;p&gt;You are overlooking the effects of round-off error during the floating-point computation (ie. during the evaluations of diff(sCARA4,mu) at floating-point values). Remember that the Digits environment variable controls the working precision and is not a target accuracy tolerance.&lt;/p&gt;
&lt;pre&gt;&amp;gt; restart:&lt;br&gt;&lt;br&gt;&amp;gt; sCARA4 := -ln(-(mu/sigma^2)^(mu^2/(mu-sigma^2))*(sigma^2/mu)^(mu^2/(mu-sigma^2))&lt;br&gt;&amp;gt;    +(sigma^2/mu)^(mu^2/(mu-sigma^2))*((exp(phi)*sigma^2+mu-sigma^2)&lt;br&gt;&amp;gt;    *exp(-phi)/sigma^2)^(mu^2/(mu-sigma^2))+1)/phi:&lt;br&gt;&lt;br&gt;&amp;gt; dsCARA4:=diff(sCARA4,mu):&lt;br&gt;&lt;br&gt;&amp;gt; eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          -2.499130625&lt;br&gt;&lt;br&gt;&amp;gt; evalf[20](eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]));&lt;br&gt;                     0.99919104807402453122&lt;br&gt;&lt;/pre&gt;
&lt;p&gt;Note also that, even though simplification (using `simplify`, with or without assumptions) seems to help below, there is no guarantee in general that it will generate a form of a symbolic expression with better conditioning during floating-point evaluation.&lt;/p&gt;
&lt;pre&gt;&amp;gt; eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          -2.499130625&lt;br&gt;&amp;gt; eval(simplify(dsCARA4),[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          1.204768329&lt;br&gt;&amp;gt; eval(simplify(dsCARA4),[mu=29.38,phi=1.0,sigma=1.0]) assuming positive;&lt;br&gt;                          0.9991907685&lt;br&gt;&lt;/pre&gt;
&lt;p&gt;You can also look at &lt;a href='http://www.maplesoft.com/support/help/search.aspx?term=evalr' target='_new'&gt;?evalr&lt;/a&gt;&lt;/p&gt;
&lt;p&gt;You are overlooking analysis during the simplification of sCARA4. You might compare output from these,&lt;/p&gt;
&lt;pre&gt;simplify(sCARA4);&lt;br&gt;&lt;br&gt;simplify(sCARA4) assuming positive;
&lt;p&gt;acer&lt;/p&gt;&lt;/pre&gt;</itunes:summary>
      <description>&lt;p&gt;You are overlooking the effects of round-off error during the floating-point computation (ie. during the evaluations of diff(sCARA4,mu) at floating-point values). Remember that the Digits environment variable controls the working precision and is not a target accuracy tolerance.&lt;/p&gt;
&lt;pre&gt;&amp;gt; restart:&lt;br&gt;&lt;br&gt;&amp;gt; sCARA4 := -ln(-(mu/sigma^2)^(mu^2/(mu-sigma^2))*(sigma^2/mu)^(mu^2/(mu-sigma^2))&lt;br&gt;&amp;gt;    +(sigma^2/mu)^(mu^2/(mu-sigma^2))*((exp(phi)*sigma^2+mu-sigma^2)&lt;br&gt;&amp;gt;    *exp(-phi)/sigma^2)^(mu^2/(mu-sigma^2))+1)/phi:&lt;br&gt;&lt;br&gt;&amp;gt; dsCARA4:=diff(sCARA4,mu):&lt;br&gt;&lt;br&gt;&amp;gt; eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          -2.499130625&lt;br&gt;&lt;br&gt;&amp;gt; evalf[20](eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]));&lt;br&gt;                     0.99919104807402453122&lt;br&gt;&lt;/pre&gt;
&lt;p&gt;Note also that, even though simplification (using `simplify`, with or without assumptions) seems to help below, there is no guarantee in general that it will generate a form of a symbolic expression with better conditioning during floating-point evaluation.&lt;/p&gt;
&lt;pre&gt;&amp;gt; eval(dsCARA4,[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          -2.499130625&lt;br&gt;&amp;gt; eval(simplify(dsCARA4),[mu=29.38,phi=1.0,sigma=1.0]);&lt;br&gt;                          1.204768329&lt;br&gt;&amp;gt; eval(simplify(dsCARA4),[mu=29.38,phi=1.0,sigma=1.0]) assuming positive;&lt;br&gt;                          0.9991907685&lt;br&gt;&lt;/pre&gt;
&lt;p&gt;You can also look at &lt;a href='http://www.maplesoft.com/support/help/search.aspx?term=evalr' target='_new'&gt;?evalr&lt;/a&gt;&lt;/p&gt;
&lt;p&gt;You are overlooking analysis during the simplification of sCARA4. You might compare output from these,&lt;/p&gt;
&lt;pre&gt;simplify(sCARA4);&lt;br&gt;&lt;br&gt;simplify(sCARA4) assuming positive;
&lt;p&gt;acer&lt;/p&gt;&lt;/pre&gt;</description>
      <guid>94849</guid>
      <pubDate>Tue, 06 Jul 2010 18:54:40 Z</pubDate>
      <itunes:author>acer</itunes:author>
      <author>acer</author>
    </item>
    <item>
      <title>learned something new</title>
      <link>http://www.mapleprimes.com/questions/94834-A-Matter-Of-Trust?ref=Feed:MaplePrimes:A matter of trust?:Comments#comment94884</link>
      <itunes:summary>&lt;p&gt;Thanks acer,&lt;/p&gt;
&lt;p&gt;I learned something new. In fact I didn&amp;acute;t know, that the Digits environment variable controls the working precision and that the effect of round-off error could be that significant.&lt;/p&gt;
&lt;p&gt;So may I also manipulate the working precision of the 'Explore' option by setting 'Digits' to another value (say 20) or is there another way of minimizing the round-off error effects when using 'Explore'?&lt;/p&gt;</itunes:summary>
      <description>&lt;p&gt;Thanks acer,&lt;/p&gt;
&lt;p&gt;I learned something new. In fact I didn&amp;acute;t know, that the Digits environment variable controls the working precision and that the effect of round-off error could be that significant.&lt;/p&gt;
&lt;p&gt;So may I also manipulate the working precision of the 'Explore' option by setting 'Digits' to another value (say 20) or is there another way of minimizing the round-off error effects when using 'Explore'?&lt;/p&gt;</description>
      <guid>94884</guid>
      <pubDate>Wed, 07 Jul 2010 14:00:01 Z</pubDate>
      <itunes:author>afeddersen</itunes:author>
      <author>afeddersen</author>
    </item>
  </channel>
</rss>