<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">The best theoretical approach is the
one mentioned by Laurence in his last message. <br>
<br>
About your last question ... It is clearly a matter of taste
because whatever we do, an approach consisting to choose the Ueff
value by ajusting it for one property is "semi-empirical". I never
did such an asjustment of Ueff ... just based on the cell volume
or structural properties and then use the Ueff for electronic
properties. <br>
However, it could make sense, because the structural properties
are the consequence of the electronic structure (orbital
interactions) and thus a good description of the structural
properties should imply a good description of the electronic
structure ... However, you can also artificially stabilize the
experimental volume for bad reasons!!! It is why I prefer to have
one more property (optical, magnetic, NMR, ...) to compare with my
calculation in order to validate the Ueff value. <br>
<br>
I should mentioned that depending on what you are doing the Ueff
value could be different! Indeed fixing the Ueff based on magnetic
properties or optical properties will not lead to the same Ueff
value ... This is mainly related to the fact that when we adjust
the Ueff value based on an experimental value we correct not only
the deficiency of DFT to properly describe the localized d- or
f-states, but more ... In that sense adjusting the Ueff value on
an experimental value is not correct but it remains a pragmatic
approach (with clear limitations). <br>
<br>
In the other hand, the Ueff value deduded from the constrained
approach will lead to results that will not perfectly reproduce
the experiments, but these results will correspond to the best you
can do using DFT with a "correct" Ueff correction. It is why I
consider that it remains a matter of taste and why I feel that you
need other experimental data to insure that your Ueff parameter is
"relevant in a certain context". <br>
<br>
Best Regards<br>
<br>
Xavier<br>
<br>
<br>
Le 3/1/2013 9:12 PM, Zsolt Rak a écrit :<br>
</div>
<blockquote
cite="mid:CACQkguoaX9m+9+yMZfdHVpno6z-SkqP7SXZB7ANSPSuupZp3PA@mail.gmail.com"
type="cite">Dear Xavier, you said that you choose the Ueff value
to reproduce one property and then you use this Ueff value to
predict the others. My question is can I choose Ueff to reproduce
a structural property (such as the lattice constant or cell
volume) and than use that value of Ueff to predict electronic
properties? <br>
<br>
<br>
<br>
<div class="gmail_quote">On Fri, Mar 1, 2013 at 3:04 PM, Laurence
Marks <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:L-marks@northwestern.edu" target="_blank">L-marks@northwestern.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
If all you have are the atomic positions (and lattice
constant) then the only unbiased approach would be to
calculate the U, see <a moz-do-not-send="true"
href="http://www.wien2k.at/reg_user/textbooks/Constraint_U.pdf"
target="_blank">http://www.wien2k.at/reg_user/textbooks/Constraint_U.pdf</a>.
<div>
<br>
</div>
<div>N.B., the DFT+U method is variational for a fixed U, but
I don't think it is variational as a function(al) of U.<br>
<br>
<div class="gmail_quote">
<div class="im">On Fri, Mar 1, 2013 at 1:58 PM, Zsolt Rak
<span dir="ltr"><<a moz-do-not-send="true"
href="mailto:zsolt.rak@gmail.com" target="_blank">zsolt.rak@gmail.com</a>></span>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>
<div class="im">
Let's suppose that the atomic positions (and lattice
constants) are the only information that I have. Is
it physically justified to use LDA+U/GGA+U to
optimize the volume? or to tune the U value to
reproduce the experimental lattice constants? Also,
is the DFT+U method based on the variational
principle?<br>
<br>
</div>
<div>
<div class="h5">
<div class="gmail_quote">On Fri, Mar 1, 2013 at
2:43 PM, Laurence Marks <span dir="ltr">
<<a moz-do-not-send="true"
href="mailto:L-marks@northwestern.edu"
target="_blank">L-marks@northwestern.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
You need more a-priori information than this,
for instance the bulk energy of related
compounds for which a U is relevant. Tuning
the U to reproduce known data is not by itself
spectacular science.<br>
<br>
<div class="gmail_quote">
<div>On Fri, Mar 1, 2013 at 1:24 PM, Zsolt
Rak <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:zsolt.rak@gmail.com"
target="_blank">zsolt.rak@gmail.com</a>></span>
wrote:<br>
</div>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;border-left:1px
#ccc solid;padding-left:1ex">
<div>
<div>I want to calculate the most
accurate bulk energy and the a-priori
information I have are the atomic
positions.<br>
<br>
<br>
<br>
</div>
<div class="gmail_quote">
<div>On Fri, Mar 1, 2013 at 1:59 PM,
Laurence Marks <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:L-marks@northwestern.edu"
target="_blank">L-marks@northwestern.edu</a>></span>
wrote:<br>
</div>
<blockquote class="gmail_quote"
style="margin:0 0 0
.8ex;border-left:1px #ccc
solid;padding-left:1ex">
<div>
<div>My two cents. Both LDA+U and
GGA+U are wrong. That said, for
f-/d- systems they are often
better than LDA/GGA for some
properties. The question you
should ask yourself is what
property are you trying to
measure/predict, and what
a-priori information (reference
state) do you have that can be
used?
<div><br>
</div>
<div>For instance, if I want to
calculate a surface energy
then I would tune the U to
give the most accurate bulk
energy treating this as my
a-priori information;
similarly if I wanted to
calculate the elastic behavior
of a defect I would tune to
the bulk elastic constants. In
my opinion this is the only
justifiable approach.
<div>
<div><br>
<br>
<div class="gmail_quote">On
Fri, Mar 1, 2013 at
12:47 PM, Zsolt Rak <span
dir="ltr">
<<a
moz-do-not-send="true"
href="mailto:zsolt.rak@gmail.com" target="_blank">zsolt.rak@gmail.com</a>></span>
wrote:<br>
<blockquote
class="gmail_quote"
style="margin:0 0 0
.8ex;border-left:1px
#ccc
solid;padding-left:1ex">
<div>Dear wien2k
users, <br>
<br>
I am calculating the
properties of
several f-electron
compounds. I would
like to ask the
users' opinion about
the volume
optimization in an
f- or d-electron
system: which way is
better (or
physically
justified), with
LDA/GGA or with
LDA+U/GGA+U? In my
opinion, the
LDA+U/GGA+U
techniques were
developed to correct
band energies of
localized states, so
there is no
fundamental physical
reason to use
LDA+U/GGA+U methods
for volume
optimization.
However, we observe
a change in the
lattice parameters
when we go from
LDA/GGA to
LDA+U/GGA+U. Also,
from a brief search
of the literature we
found that, in many
cases, people tune
the Hubbard-U
parameter to
reproduce the
experimental lattice
constants.
<br>
I would appreciate
further thoughts and
insights into this
issue.<br>
<br>
Thank you, <br>
Zs </div>
</blockquote>
</div>
<br>
<br clear="all">
<div><br>
</div>
</div>
</div>
<span><font color="#888888">--
<br>
Professor Laurence Marks<br>
Department of Materials
Science and Engineering<br>
Northwestern University<br>
<a moz-do-not-send="true"
href="http://www.numis.northwestern.edu" target="_blank">www.numis.northwestern.edu</a>
<a moz-do-not-send="true"
href="tel:1-847-491-3996" value="+18474913996" target="_blank">1-847-491-3996</a><br>
"Research is to see what
everybody else has seen,
and to think what nobody
else has thought"<br>
Albert Szent-Gyorgi </font></span></div>
<br>
</div>
</div>
<div>_______________________________________________<br>
Wien mailing list<br>
<a moz-do-not-send="true"
href="mailto:Wien@zeus.theochem.tuwien.ac.at"
target="_blank">Wien@zeus.theochem.tuwien.ac.at</a><br>
<a moz-do-not-send="true"
href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien"
target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
<br>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
</div>
<br>
<br clear="all">
<div>
<div>
<div><br>
</div>
-- <br>
Professor Laurence Marks<br>
Department of Materials Science and
Engineering<br>
Northwestern University<br>
<a moz-do-not-send="true"
href="http://www.numis.northwestern.edu"
target="_blank">www.numis.northwestern.edu</a>
<a moz-do-not-send="true"
href="tel:1-847-491-3996"
value="+18474913996" target="_blank">1-847-491-3996</a><br>
"Research is to see what everybody else
has seen, and to think what nobody else
has thought"<br>
Albert Szent-Gyorgi </div>
</div>
<br>
_______________________________________________<br>
Wien mailing list<br>
<a moz-do-not-send="true"
href="mailto:Wien@zeus.theochem.tuwien.ac.at"
target="_blank">Wien@zeus.theochem.tuwien.ac.at</a><br>
<a moz-do-not-send="true"
href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien"
target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
<br>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</blockquote>
</div>
<div>
<div class="h5"><br>
<br clear="all">
<div><br>
</div>
-- <br>
Professor Laurence Marks<br>
Department of Materials Science and Engineering<br>
Northwestern University<br>
<a moz-do-not-send="true"
href="http://www.numis.northwestern.edu"
target="_blank">www.numis.northwestern.edu</a> <a
moz-do-not-send="true" href="tel:1-847-491-3996"
value="+18474913996" target="_blank">1-847-491-3996</a><br>
"Research is to see what everybody else has seen, and to
think what nobody else has thought"<br>
Albert Szent-Gyorgi
</div>
</div>
</div>
<br>
_______________________________________________<br>
Wien mailing list<br>
<a moz-do-not-send="true"
href="mailto:Wien@zeus.theochem.tuwien.ac.at">Wien@zeus.theochem.tuwien.ac.at</a><br>
<a moz-do-not-send="true"
href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien"
target="_blank">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a><br>
<br>
</blockquote>
</div>
<br>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
Wien mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Wien@zeus.theochem.tuwien.ac.at">Wien@zeus.theochem.tuwien.ac.at</a>
<a class="moz-txt-link-freetext" href="http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien">http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien</a>
</pre>
</blockquote>
<br>
</body>
</html>