<br><br><div><span class="gmail_quote">On 10/23/08, <b class="gmail_sendername">Paolo Giannozzi</b> &lt;<a href="mailto:giannozz@democritos.it">giannozz@democritos.it</a>&gt; wrote:</span><blockquote class="gmail_quote" style="margin-top: 0; margin-right: 0; margin-bottom: 0; margin-left: 0; margin-left: 0.80ex; border-left-color: #cccccc; border-left-width: 1px; border-left-style: solid; padding-left: 1ex">
Alexej Mazheika wrote:<br><br> &gt; Yes, really. I use the version 4.0.1, there are such options in this<br> &gt; version.<br><br><br>well, no. More exactly, there are, but they are fake. Try<br><br>&nbsp;&nbsp; diagonalization=&#39;whatever funny algorithm you can think of&#39;<br>
<br> and the code will work. Of course it will not use whatever funny<br> algorithm you inserted there, but will use the default, i.e.<br> Davidson diagonalization (with parallel subspace diagonalization).<br> Only the following keywords do something different from the default:<br>
 &#39;cg&#39;, &#39;cg-serial&#39;, &#39;cg+serial&#39;, &#39;david-serial&#39;, &#39;david+serial&#39;.<br> The documentation is rather unclear on this point.<br> Anyway: you used Davidson, and that&#39;s fine.<br><br> Paolo<br>
<br>--<br><br>Paolo Giannozzi, Democritos and University of Udine, Italy<br> _______________________________________________</blockquote><div><br></div></div>I&nbsp;use&nbsp;the&nbsp;program&nbsp;PWGUI&nbsp;for&nbsp;the building&nbsp;of&nbsp;my&nbsp;input&nbsp;files.&nbsp;There&nbsp;is&nbsp;the keyword &#39;diis&#39; in PWGUI, and so I thought that  such&nbsp;algorithm&nbsp;is&nbsp;present&nbsp;in&nbsp;Espresso-4.0.1&nbsp;too.&nbsp;During&nbsp;the&nbsp;calculations&nbsp;it&nbsp;does&nbsp;not&nbsp;write&nbsp;anything&nbsp;about&nbsp;&#39;unrecognized&nbsp;word&#39;&nbsp;both in&nbsp;output&nbsp;file&nbsp;and&nbsp;in&nbsp;Linux-console,&nbsp;so&nbsp;I&nbsp;really&nbsp;considered&nbsp;that&nbsp;the&nbsp;&#39;diis&#39;&nbsp;algorithm&nbsp;is&nbsp;embedded&nbsp;in&nbsp;Espresso.&nbsp;Then&nbsp;it&nbsp;means&nbsp;that&nbsp;the&nbsp;changing&nbsp;of&nbsp;occupation and mixing&nbsp;data&nbsp;helped&nbsp;to&nbsp;solve&nbsp;the&nbsp;problem&nbsp;with&nbsp;covergance.&nbsp;That&#39;s&nbsp;fine.<br>
<br>  Alex Mazheika, Research Institute for phys. and chem. problems, Minsk, Belarus<br>