<table cellspacing="0" cellpadding="0" border="0" ><tr><td valign="top" style="font: inherit;">Dear Matteo<br>In your tutorial about LDA_U, you mentioned that the perturbation should be isolated. and&nbsp; said that we normally use supercell of increasing size up to when convergence of is reached. I don't understand how do you increase the size of the cell and how do you compute the U in larger supercell.? I mean how do you apply perturbation to bigger size or how to extrapolate the U to converge?<br>Another question according to prior email for computation of U by common method(not linear response) is : you mentioned that espresso cann't directly change the occupation of special orbital by hand, but it is possible to do by using constraint on total number of electron(total occupation) indirectly . Do you mean that( for example for Ti atom that has 2 electron in d orbital) ,what can i do is change the of number of nelup and neldw&nbsp; without change in
 nelec? or I have to do another switch? Would you please tell me in more details.<br>Thank you very much for your aids<br>Ali Kazempour<br>
Physics department, Isfahan University of Technology<br>
84156 Isfahan, Iran.            Tel-1:  +98 311 391 3733<br>
Fax: +98 311 391 2376      Tel-2:  +98 311 391 2375<br><br>--- On <b>Tue, 8/4/09, Matteo Cococcioni <i>&lt;matteo@umn.edu&gt;</i></b> wrote:<br><blockquote style="border-left: 2px solid rgb(16, 16, 255); margin-left: 5px; padding-left: 5px;"><br>From: Matteo Cococcioni &lt;matteo@umn.edu&gt;<br>Subject: Re: [Pw_forum] U for Oxygen<br>To: "PWSCF Forum" &lt;pw_forum@pwscf.org&gt;<br>Date: Tuesday, August 4, 2009, 9:52 AM<br><br><div class="plainMail"><br>this is what comes out from the calculation (U(O) &gt; U(Ni)). whether it's <br>reasonable to apply U on O or not, I'm not sure.<br><br>I think the U shouldn't be used on states that are completely occupied.<br><br>Matteo<br><br><br><br>ali kazempour wrote:<br>&gt; Dear Matteo<br>&gt; This means that hubbard parameter for oxygen is large (grater than <br>&gt; Ni). Is it correct or originate from another challenge?<br>&gt; thanks<br>&gt;<br>&gt; Ali Kazempour<br>&gt; Physics department, Isfahan University
 of Technology<br>&gt; 84156 Isfahan, Iran. Tel-1: +98 311 391 3733<br>&gt; Fax: +98 311 391 2376 Tel-2: +98 311 391 2375<br>&gt;<br>&gt; --- On *Tue, 8/4/09, Matteo Cococcioni /&lt;<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a>&gt;/* wrote:<br>&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;From: Matteo Cococcioni &lt;<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Subject: Re: [Pw_forum] U for Oxygen<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;To: "PWSCF Forum" &lt;<a ymailto="mailto:pw_forum@pwscf.org" href="/mc/compose?to=pw_forum@pwscf.org">pw_forum@pwscf.org</a>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Date: Tuesday, August 4, 2009, 6:59 AM<br>&gt;<br>&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Dear Ali,<br>&gt;<br>&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;ali kazempour wrote:<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; Dear Matteo<br>&gt;&nbsp;
 &nbsp;&nbsp;&nbsp;&gt; I used your comment and followed&nbsp;&nbsp;&nbsp;your example for NiO and I<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;succeed <br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; to get U for TiO2. But in file Ur16.out I see three column .<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;I don't have any three column section in any of my files. so if<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;you want<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;me to help you you have to report at least a section of your output<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;file. This is a good habit actually in consideration of the fact that<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;this is not a private conversation (I would have asked you in any<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;case<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;as I don't remember by heart the format of any file) between me<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;and you<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;but a message to the forum and you want
 other people to understand<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;what you are saying.<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; The first is number of atom in supercell , second is U value for Ni<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; atom and I don,t know&nbsp; the third? does the third belong to&nbsp; U value<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; for Oxygen or not? Because the third column kas big value?<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; another question is for isolated atom Hubbard parameter. What do<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;you<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; expect about the value of U for Isolated atom? Is this big value<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; (greater than 10 eV) or&nbsp; it would be tiny value?<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;I think it will be meaningless for the reasons I explained in my
 last<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;email. To obtain a meaningful number for the isolated atom you<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;probably<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;have to abandon the idea to use a perturbation in the potential and<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;change instead the number of electrons directly. This means you will<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;have to do calculations on atoms having fractional charges which<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;requires additional corrections if you use a code designed for<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;periodic<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;systems (crystals). However I have never done this isolated atom<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;calculation of U before so I don't have more details about it.<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;I see (from a private email) you are thinking of computing U for the<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;isolated atom and to obtain a value for the crystal just
 by<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;dividing by<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;the epsilon_inft. the method is proposed in<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;<a href="http://link.aps.org/doi/10.1103/PhysRevB.74.045202" target="_blank">http://link.aps.org/doi/10.1103/PhysRevB.74.045202</a><br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;(which I haven't re-read yet).<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;I see a couple of problems:<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;1) (not important for ZnO). suppose you have Fe3O4 with Fe2+ and<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Fe3+.<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;these Fe ions likely have different U. how do you obtain<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;a different number from the atomic code? would you do the calculation<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;for isolated Fe2+ and Fe3+?<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;2) how do you get epsilon? since it's a linear response property I<br>&gt;&nbsp;
 &nbsp;&nbsp;&nbsp;don't<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;think its calculation is any cheaper than the direct calculation of U.<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;regards,<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Matteo<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; should we choose for running the script for isolated atom only<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;gamma<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; point or we have to choose a fine mesh?<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; thanks a lot<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; Ali Kazempour<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; Physics department, Isfahan University of Technology<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; 84156 Isfahan, Iran. Tel-1: +98 311 391 3733<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; Fax: +98 311 391 2376 Tel-2: +98 311 391 2375<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt; --- On *Sun,
 8/2/09, Matteo Cococcioni /&lt;<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a><br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&lt;/mc/compose?to=<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a>&gt;&gt;/* wrote:<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;From: Matteo Cococcioni &lt;<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a><br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&lt;/mc/compose?to=<a ymailto="mailto:matteo@umn.edu" href="/mc/compose?to=matteo@umn.edu">matteo@umn.edu</a>&gt;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;Subject: Re: [Pw_forum] occupation<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;To: "PWSCF Forum" &lt;<a ymailto="mailto:pw_forum@pwscf.org"
 href="/mc/compose?to=pw_forum@pwscf.org">pw_forum@pwscf.org</a><br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&lt;/mc/compose?to=<a ymailto="mailto:pw_forum@pwscf.org" href="/mc/compose?to=pw_forum@pwscf.org">pw_forum@pwscf.org</a>&gt;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;Date: Sunday, August 2, 2009, 8:31 AM<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;Dear Ali,<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;1, -1, 0 are just spin indexes to tell the code what is the<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;relative<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;magnetization of the atoms (parallel, antiparallel, or none). if<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;your system is non magnetic put all 0. or set magn = .false.<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;and
 put<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;nothing.<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;you should look at the examples on Fe and Si as well and<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;learn from<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;there by comparison.<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;&nbsp; &nbsp;&nbsp;&nbsp;Matteo<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;&gt;<br>&gt;<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;-- <br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Matteo Cococcioni<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Department of Chemical Engineering and Materials Science,<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;University of Minnesota<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;421 Washington Av. SE<br>&gt;&nbsp;
 &nbsp;&nbsp;&nbsp;Minneapolis, MN 55455<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Tel. +1 612 624 9056&nbsp; &nbsp; Fax +1 612 626 7246<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;_______________________________________________<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;Pw_forum mailing list<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;<a ymailto="mailto:Pw_forum@pwscf.org" href="/mc/compose?to=Pw_forum@pwscf.org">Pw_forum@pwscf.org</a> &lt;/mc/compose?to=<a ymailto="mailto:Pw_forum@pwscf.org" href="/mc/compose?to=Pw_forum@pwscf.org">Pw_forum@pwscf.org</a>&gt;<br>&gt;&nbsp; &nbsp;&nbsp;&nbsp;<a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>&gt;<br>&gt;<br>&gt; ------------------------------------------------------------------------<br>&gt;<br>&gt; _______________________________________________<br>&gt; Pw_forum
 mailing list<br>&gt; <a ymailto="mailto:Pw_forum@pwscf.org" href="/mc/compose?to=Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>&gt; <a href="http://www.democritos.it/mailman/listinfo/pw_forum" target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br>&gt;&nbsp;&nbsp;&nbsp;<br><br><br>-- <br>%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br>Matteo Cococcioni<br>Department of Chemical Engineering and Materials Science,<br>University of Minnesota<br>421 Washington Av. SE<br>Minneapolis, MN 55455<br>Tel. +1 612 624 9056&nbsp; &nbsp; Fax +1 612 626 7246<br>%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br><br>_______________________________________________<br>Pw_forum mailing list<br><a ymailto="mailto:Pw_forum@pwscf.org" href="/mc/compose?to=Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br><a href="http://www.democritos.it/mailman/listinfo/pw_forum"
 target="_blank">http://www.democritos.it/mailman/listinfo/pw_forum</a><br></div></blockquote></td></tr></table><br>