[Pw_forum] nspin values

Thu Jan 11 07:44:44 CET 2018

Dear Dr. Daniele,

 Thank you very much for your detailed reply to my question regarding
Fourier interpolation for DFPT.

 I am trying to replicate the results of your article, "Thermoelastic
properties of alpha-iron from first-principles" published in PRB 91,
104105, in 2015.

 I used the pseudo potential Fe.pbe-spn-rrkjus psl.0.2.1.UPF, and other
values of parameter (e.g. wavefunction kinetic-energy cut off of 90 Ry
etc.) as mentioned in the first two paragraphs of the section : III.
COMPUTATIONAL DETAILS AND PSEUDOPOTENTIAL SELECTION.

I have coupled thermo_pw.0.8.0 with QE-6.2. The script file I used is:

cat > thermo_control << EOF
 &INPUT_THERMO
  what='scf_disp',
  nq1_d=128,
  nq2_d=128,
  nq3_d=128,
 /
EOF
cat > fe.scf_disp.in << EOF
 &control
   calculation = 'scf'
   prefix='fe',
   pseudo_dir = '$PSEUDO_DIR/',
   outdir='$TMP_DIR/'
 /
 &system

ibrav = 3, celldm(1) =5.217, nat= 1, ntyp= 1,
    ecutwfc = 90.0,ecutrho = 1080.0,
 occupations='smearing', smearing='marzari-vanderbilt', degauss=0.005
    nspin = 2
    starting_magnetization(1) = 0.5
        /
 &electrons
    conv_thr = 1.0e-16,
 mixing_beta = 0.2
 /
ATOMIC_SPECIES
 Fe 55.847 Fe.pbe-spn-rrkjus_psl.0.2.1.UPF
ATOMIC_POSITIONS (alat)
 Fe 0.0 0.0 0.0
K_POINTS {automatic}
 24 24 24 1 1 1
EOF
cat > ph_control << EOF

 &inputph
  tr2_ph = 1.0e-16,
  prefix='fe',
  fildyn='fe.dyn',
  ldisp=.TRUE.
  nq1=4, nq2=4, nq3=4,

/
EOF

With the above input the generated fe.dyn.g1.1 dynamic matrix has negative
values of frequencies.

 I guess I am not using the correct value of nspin. as according to pw.x
input description, nspin = 2, is for:

 spin-polarized calculation, LSDA
  (manetization along z axis)

However, for the GGA functional (Fe.pbe-spn-rrkjus_psl.0.2.1.UPF) I do not
know what is the value of nspin I should be using.

So, my question is kindly let me know the value of nspin I should use.
Also, if possible, please let me know if any other parameters I have input
has incorrect values or needs improvement.

Thanking you,
Best regards,
Krishnendu

------------------------------------------------------------------------------------------------------------------

Date: Fri, 5 Jan 2018 13:42:57 +0000
From: Dragoni Daniele <daniele.dragoni at epfl.ch>
Subject: Re: [Pw_forum] Fourier interpolation for DFPT
To: PWSCF Forum <pw_forum at pwscf.org>
Message-ID: <1515159777313.75745 at epfl.ch>
Content-Type: text/plain; charset="iso-8859-1"

Dear Krishnendu,

in order to perform the so-called Fourier-interpolation you need first to
compute phonon frequencies with DPTF on a proper grid of q-points using
ph.x. From the dynamical matrix produced by the phonon code you can then
compute the interatomic force constants (IFC) in real space using the q2r.x
code. From there, it is possible to recompute the phonon frequencies at any
arbitrary q-point by means of the matdyn.x code. At this stage, you can
specify the q-mesh (or a list of specific q-points) that is required for
calculating converged values of the quantities you are interested in (in
our case a 21x21x21 mesh was sufficient). For more specific information,
try to have a look at this very useful document by Paolo Giannozzi
http://www.fisica.uniud.it/~giannozz/QE-Tutorial/tutorial_disp.html

Both q2r.x and matdyn.x should be automatically compiled by typing "make
ph" from the main quantum Espresso directory, as stated (for example) here
http://www.afs.enea.it/software/qu_esp/Doc/ph_6.0.0.pdf

Examples of q2r and matdyn input files can be found here

http://www.fisica.uniud.it/~giannozz/QE-Tutorial/examples_disp.tgz

Have also a look at this tutorial (Exercise 2 in particular)

http://www.tcl.t.u-tokyo.ac.jp/pdf/2014_lm/3rd_day_pm.pdf

In particular you can read:

##q2r input template

&INPUT
    fildyn = ***,
    flfrc  = ***,
    zasr   = ***,    (we use crystal)
 /

##

## matdyn input template

&INPUT
    ntyp  = ***,    (we have 1 atom type)
    flvec = ***,
    flfrc = ***,
    asr   = ***,    (we use crystal)
    amass(1) = ***,
/
***               (integer # of q points for which you want compute phonon
frequencies)
x1 x2 x3  0 (here goes a list of q points coordinates followed by a growing
integer number N counting the q-points. In our case, this list should
correspond to a 21x21x21 MP grid)

x1 x2 x3  1
x1 x2 x3  2

...

...

##

Replace *** with the correct input card. An exhaustive description of the
input cards available for the two codes can be found in the header of the
source codes q2r.f90 and matdyn.f90 in /yourQEfolder/PHonon/PH directory.

Hope this helps!

Best regards,

   Daniele

__________________________________________

Daniele Francesco Dragoni
Laboratoire de th?orie et simulation des mat?riaux
Swiss Federal Institute of Technology
EPFL STI IMX THEOS
ME D2 1020
Station 9
CH-1015, Lausanne
+41 21 69 31956
<tel:+41216931956>

________________________________
From: pw_forum-bounces at pwscf.org <pw_forum-bounces at pwscf.org> on behalf of
Krishnendu Mukherjee <krishnendu.mukherjee789 at gmail.com>
Sent: Tuesday, January 2, 2018 10:31 AM
To: pw_forum at pwscf.org
Subject: [Pw_forum] Fourier interpolation for DFPT

Dear experts,

 I am interested in calculating the phonon calculation of bcc iron ferrite.
I am following the article titled "Thermoelastic properties of alpha-iron
from first-principles" authored by Daniele Dragoni, Davide Ceresoli and
Nicola Marzari, published in PRB 91, 104105. In that article it is
mentioned that,

"Phonon calculations were carried out for each deformation within DFPT. The
dynamical matrix and its eigenvalues are calculated on an 4x4x4 mesh of
special points in the BZ and Fourier-interpolated on an extended 21x21x21
grid for integration of thermodynamic quantities."

 Now I gave the nq1=4, nq2=4, nq3=4 at fe.ph.in<http://fe.ph.in> according
to the values mentioned in the article.

 But I do not know the parameter for which I have to set the values
21x21x21. Can you kindly let me know the parameter where I have to put the
values to set the Fourier-interpolation extension grid for integration of
thermodynamic quantities.

-- 
Dr. Krishnendu Mukherjee,

Principal Scientist,
CSIR-NML,
Jamshedpur.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180111/d37eb081/attachment.html>