<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
On 28/05/2012 8:14 AM, Igor Druz wrote:
<blockquote
cite="mid:CAPEHq69XDqzc8Q3-jSP6FmA6sM420AqvV9+dKjapBtVyKVTWDQ@mail.gmail.com"
type="cite">
<pre>><i> You need to choose a large .xtc precision, or use nstxout and filter the group
</i>><i> afterwards.</i></pre>
<pre><i>
</i></pre>
<pre><i>I tried xtc-precision=1000000. This gives the same result as the default value </i><i>xtc-precision=1000.</i></pre>
</blockquote>
<br>
That seems highly unlikely. Can you demonstrate that with gmxdump of
the two files?<br>
<br>
<blockquote
cite="mid:CAPEHq69XDqzc8Q3-jSP6FmA6sM420AqvV9+dKjapBtVyKVTWDQ@mail.gmail.com"
type="cite">
<pre><i> Yet, the results are different from the analysis of the trr file.</i></pre>
</blockquote>
<br>
Necessarily.<br>
<br>
Of course, there may not be any value in saving the coordinates in
high precision. The errors from incomplete sampling and an
approximate model physics will dominate at some point.<br>
<br>
Mark<br>
</body>
</html>