<div dir="ltr">Dear PWSCF users, <br>
<br>
The following message might have been lost, so I permit myself to
reiterate it: would someone have an idea why the finite displacement
method as implemented in quantum espresso v 5.1. gives wrong matrices of
interatomic force constants when supercell options are taken? <br>
<br>
Also, I was wondering if an example of input would be available for
using fd_ef.x (calculation of dielectric constants and effective charges
by finite displacement). <br>
<br>
<div>Thanks in advance for any help,<br><br></div>
Best regards,<br>
<br>
Merlin Méheut, PhD, <br>
Géosciences et Environnement Toulouse,
OMP, <br>
Université Paul Sabatier Toulouse III,<br>
14 avenue Edouard Belin, 31400 Toulouse, France
<br>
phone +33 (0)5 61 33 26 17, fax +33 (0)5 61 33 25 60
<br>
<a href="https://www.researchgate.net/profile/Merlin_Meheut">https://www.researchgate.net/profile/Merlin_Meheut</a><br>
<div class="gmail_extra"><br><div class="gmail_quote">2016-04-26 16:54 GMT+01:00 Merlin Meheut <span dir="ltr"><<a href="mailto:merlin.meheut@gmail.com" target="_blank">merlin.meheut@gmail.com</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Dear QE users and developpers,
<br>
<br>I would like to compute phonon properties using the finite displacement
method, as proposed in the example of quantum espresso version 5.1
(PHonon/FD/example). For frequencies at q=0, the answer is reasonably
close to the results of DFPT, but for q-vectors different from zero, I
find very negative frequencies, and very different from DFPT.
<br>
<br>Follows a description of my calculations. Any help would be thoroughly
appreciated.
<br>
<br>My system is forsterite Mg2SiO4 (orthorhombic Pbnm):
<br>
<br>&system
<br> ibrav =0, celldm(1)=8.985459,
<br> nat =28, ntyp = 3, ecutwfc =${a}.0, ecutrho = ${b}.0
<br>/&end
<br>
<br>CELL_PARAMETERS (alat= 8.98545900)
<br> 1.011800931 0.000000000 0.000000000
<br> 0.000000000 2.175636534 0.000000000
<br> 0.000000000 0.000000000 1.273688789
<br>
<br>ATOMIC_POSITIONS (crystal)
<br>Mg -0.000000000 -0.000000000 -0.000000000
<br>Mg 0.500000000 0.500000000 -0.000000000
<br>Mg -0.000000000 0.000000000 0.500000000
<br>Mg 0.500000000 0.500000000 0.500000000
<br>Mg 0.991704573 0.277641327 0.250000000
<br>Mg 0.491704573 0.222358673 0.750000000
<br>Mg 0.008295427 0.722358673 0.750000000
<br>Mg 0.508295427 0.777641327 0.250000000
<br>Si 0.425939676 0.093656753 0.250000000
<br>Si 0.925939676 0.406343247 0.750000000
<br>Si 0.574060324 0.906343247 0.750000000
<br>Si 0.074060324 0.593656753 0.250000000
<br>O 0.765422396 0.091530172 0.250000000
<br>O 0.265422396 0.408469828 0.750000000
<br>O 0.234577604 0.908469828 0.750000000
<br>O 0.734577604 0.591530172 0.250000000
<br>O 0.222541031 0.447081956 0.250000000
<br>O 0.722541031 0.052918044 0.750000000
<br>O 0.777458969 0.552918044 0.750000000
<br>O 0.277458969 0.947081956 0.250000000
<br>O 0.277386450 0.162817013 0.033263088
<br>O 0.777386450 0.337182987 0.966736912
<br>O 0.722613550 0.837182987 0.533263088
<br>O 0.222613550 0.662817013 0.466736912
<br>O 0.722613550 0.837182987 0.966736912
<br>O 0.222613550 0.662817013 0.033263088
<br>O 0.277386450 0.162817013 0.466736912
<br>O 0.777386450 0.337182987 0.533263088
<br>
<br>On one had I have made a phonon (perturbation theory) calculation on a
2x2x2 grid of the Brillouin zone, deliberately without Born charges and
dielectric tensor (since I do not do it with finite displacements):
<br>
<br>&inputph
<br> amass(1)=23.9850,
<br> amass(2)=27.9769,
<br> amass(3)=15.9949,
<br> alpha_mix(1) = 0.7,
<br> ldisp=.true., nq1=2, nq2=2, nq3=2,
<br> tr2_ph = 1.0D-16,
<br> prefix='FOR',
<br> fildyn='mat.$PREFIX',
<br> lraman=.false.,
<br> epsil =.false.,
<br> trans =.true.,
<br> zue = .false.,
<br> outdir='<i><span>/</span>tmpdir<span>/</span></i>$LOGNAME/'
<br>/&end
<br>0.0 0.0 0.0
<br>
<br>Then I obtained an ifc matrix with q2r and re-calculated the frequencies
with matdyn on the 8 q-points (finding the same frequencies as directly
from the individual dynamical matrices).
<br>For example, for q= (0.5,0,0) I find the following frequencies:
<br>
<br> 152.7355 152.7355 159.2373 159.2373 159.6323 159.6323
<br> 171.3805 171.3805 177.2753 177.2753 191.6439 191.6439
<br> 202.0364 202.0364 244.3864 244.3864 244.9694 244.9694
<br> 264.4823 264.4823 277.6309 277.6309 279.3910 279.3910
<br> 287.4779 287.4779 288.9345 288.9345 295.9524 295.9524
<br> 307.4176 307.4176 309.7795 309.7795 329.0229 329.0229
<br> 332.7833 332.7833 339.4032 339.4032 346.7481 346.7481
<br> 371.5853 371.5853 403.7708 403.7708 423.2761 423.2761
<br> 436.9743 436.9743 458.7147 458.7147 459.0536 459.0536
<br> 482.4561 482.4561 493.6059 493.6059 501.4041 501.4041
<br> 530.9486 530.9486 550.9678 550.9678 560.3782 560.3782
<br> 603.1617 603.1617 786.0294 786.0294 800.2429 800.2429
<br> 820.2436 820.2436 842.1800 842.1800 846.5773 846.5773
<br> 854.6477 854.6477 947.1591 947.1591 992.4286 992.4286
<br>
<br>On the other hand, I realized a finite displacement calculation with
<a href="http://fd.in" target="_blank">fd.in</a> below:
<br>
<br>&inputfd
<br> fd_prefix = 'FOR'
<br> fd_outdir = '<i><span>/</span>tmpdir/mmeheut<span>/</span></i>'
<br> fd_outfile = 'displaced'
<br> fd_outfile_dir = './fd_files'
<br> nrx1 = 2
<br> nrx2 = 2
<br> nrx3 = 2
<br> innx = 1
<br> de = 0.01
<br>/
<br>
<br>and <a href="http://fd_ifc.in" target="_blank">fd_ifc.in</a>
<br>
<br>&input
<br> prefix='FOR'
<br> nrx1=2,
<br> nrx2=2,
<br> nrx3=2,
<br> de=0.01,
<br> file_force='./fd_files/force',
<br> file_out='./For-rel-pbeUS.80Ryx4-222111.fd222',
<br> innx=1
<br> nodispsym=.false.
<br> noatsym=.false.
<br> verbose=.true.
<br> hex=.false.
<br>/
<br>
<br>the ifc matrix I obtain is a bit different from DFPT but it is difficult
to make clearer statement. When I use matdyn to obtain frequencies, at
q=0 it is correct, but at q= (0.5,0,0) for example, I have:
<br>
<br> -549.4798 -549.4798 -542.0251 -542.0251 -270.8103 -270.8103
<br> -270.3831 -270.3831 -268.2961 -268.2961 -255.8122 -255.8122
<br> -229.8735 -229.8735 -224.8183 -224.8183 -219.7591 -219.7591
<br> -205.2850 -205.2850 -200.4448 -200.4448 -197.4146 -197.4146
<br> 131.2067 131.2067 143.8600 143.8600 155.3084 155.3084
<br> 163.2342 163.2342 181.7740 181.7740 200.2658 200.2658
<br> 213.1614 213.1614 243.0599 243.0599 254.4463 254.4463
<br> 257.8945 257.8945 270.8913 270.8913 271.1355 271.1355
<br> 277.1344 277.1344 280.2588 280.2588 296.9960 296.9960
<br> 339.1382 339.1382 344.2783 344.2783 346.3596 346.3596
<br> 360.5026 360.5026 369.8420 369.8420 387.3814 387.3814
<br> 457.0229 457.0229 491.8080 491.8080 512.6835 512.6835
<br> 576.2465 576.2465 588.0433 588.0433 596.5263 596.5263
<br> 608.6300 608.6300 623.7900 623.7900 628.1992 628.1992
<br>
<br>The symetry does seem correct, but the frequencies are completely off..
<br>
<br>Thanks again for any help,
<br>
<br>Sincerely,
<br>
<br> Merlin Méheut
<br><span class="HOEnZb"><font color="#888888">
<br><div><span>-- <br></span>Merlin Méheut, Géosciences et Environnement Toulouse,
<br>OMP, 14 avenue Edouard Belin, 31400 Toulouse, France
<br>
<br>phone <a href="tel:%2B33%20%280%295%2061%2033%2026%2017" value="+33561332617" target="_blank">+33 (0)5 61 33 26 17</a>, fax <a href="tel:%2B33%20%280%295%2061%2033%2025%2060" value="+33561332560" target="_blank">+33 (0)5 61 33 25 60</a>
</div><div><div dir="ltr"><br></div></div>
</font></span></div>
</blockquote></div><br><br clear="all"><br>-- <br><div class="gmail_signature"><div dir="ltr">Merlin Méheut<br>adresse labo:<br>GET - OMP - Université Paul Sabatier <br>14 avenue Edouard Belin<br>31400 Toulouse<br>FRANCE <br>tel: (+33) 5 61 33 26 17<br><br></div></div>
</div></div>