[Pw_forum] The output of d3x
Dear Hua Bao,
you can understand the meaning of the several fields from the loop that
actually writes those files, in write_d3dyn.f90:
do i = 1, 3 * nat
if (wrmode (i) ) then
write (iudyn, '(/,12x,"modo:",i5,/)') i
nc = (i - 1) / 3 + 1
kcar = i - 3 * (nc - 1)
do na = 1, nat
do nb = 1, nat
write (iudyn, '(2i3)') na, nb
do icar = 1, 3
write (iudyn, '(3e24.12)') (phi (kcar, icar, jcar, nc,
na, nb) &
, jcar = 1, 3)
enddo
enddo
enddo
endif
enddo
phi has 6 indexes that correspond to the modes (cartesian index + atomic
index) of the 3 phonons involved in the process. I.e. phi(1,2,3, 5,6,7)
corresponds to (atom5_x, atom6_y, atom7_z)
Keep in mind that when doing a calculation with d3.x at q the first phonon
corresponds to the Gamma point, the second to -q and the third to q.
More specifically, inside the file, the index that follows "modo:" is the
mode for Gamma (modo is just italian for mode), it is composite
cartesian+atom, ie. 1=atom1_x, 2=atom1_y, 3=atom1_z, 4=atom2_x, ...
Then, the 2 isolated integers ("1 1" in your examples) are the indexes of
the atoms for the second and third mode.
Finally, the 3x3 matrix of *complex* numbers follows. Every *two* lines
the second cartesian index increments by one.
While the the third index increases by one every couple of numbers (each
are the real and imaginary part of one element). Yes, the second one has
the real part in a line and the imaginary in the next...
It sounds a bit complicated, but it is only complicated to explain in
text. If you study the code for a few minutes it should be clear.
bests
On Wed, 08 Feb 2012 22:56:48 +0100, Hua Bao wrote:
> Dear All,
>
> I am working on the anharmonic force constant calculation following
> example
> 14.
> Since there is not much description of the output format (let me know if
> I
> am wrong), I have to guess what the parameters are.
>
> Based on Debernardi's thesis and the discussions,
> I guess the code calculates C(0,q,-q), but this quantity is also a
> function of p, p', p'', which are the components of atomic displacements.
>
> I have read the code a little bit. The d3 matrix is stored in the
> quantity
> phi, which is a function of atom index and direction index( i guess?).
>
> However, I could not understand the output format as follows
>
>modo:1
>
> 1 1
> -0.138777878078E-15 0.E+00 -0.832667268469E-16
> 0.E+00 -0.277555756156E-16 0.E+00
> -0.416333634234E-16 0.E+00 0.832667268469E-16
> 0.E+00 0.383064069101E+00 0.E+00
> -0.111022302463E-15 0.E+00 0.383064069101E+00
> 0.E+00 -0.277555756156E-16 0.E+0
>
> I cannot understand "modo" and (1,1).
> I suspect (1 1) means atom index, but there are p, p', p''. Which two
> (1,1)
> stands for?
>
> Also, in the README file. It shows something like C_{x,y,z}
> (0,0,0|1,1,1).
> I can understand (0,0,0) are the three q-points, but what does this 1,1,1
> stands for? It does not look like phonon branch index or atom index.
>
> I have been struggling on this for quite some time. I really appreciate
> if
> anyone can help me to clarify this issue.
>
> Thanks,
> Hua
--
Lorenzo Paulatto IdR @ IMPMC/UPMC CNRS & Universit? Paris 6
phone: +33 (0)1 44275 084 / skype: paulatz
www: http://www.impmc.upmc.fr/~paulatto/
mail: 23-24/4?14 Bo?te courrier 115, 4 place Jussieu 75252 Paris C?dex 05
[Pw_forum] large system parallel execution problem
Dear QE users and developers, I'm trying to optimize structure of super cell containing 433 magnetic Fe atoms on Cray xt4 machine. The program runs electronic structure self consistency and calculate forces but stops at first structure optimization step. Do you know possible reason. It works for smaller systems. I've attached archive with input and output files together with pbatch I used Thank you German Samolyuk ORNL, USA -- next part -- A non-text attachment was scrubbed... Name: 111.tar.gz Type: application/x-gzip Size: 93666 bytes Desc: not available Url : http://www.democritos.it/pipermail/pw_forum/attachments/20120208/4dfa2796/attachment-0001.bin
[Pw_forum] The output of d3x
Dear All,
I am working on the anharmonic force constant calculation following example
14.
Since there is not much description of the output format (let me know if I
am wrong), I have to guess what the parameters are.
Based on Debernardi's thesis and the discussions,
I guess the code calculates C(0,q,-q), but this quantity is also a
function of p, p', p'', which are the components of atomic displacements.
I have read the code a little bit. The d3 matrix is stored in the quantity
phi, which is a function of atom index and direction index( i guess?).
However, I could not understand the output format as follows
modo:1
1 1
-0.138777878078E-15 0.E+00 -0.832667268469E-16
0.E+00 -0.277555756156E-16 0.E+00
-0.416333634234E-16 0.E+00 0.832667268469E-16
0.E+00 0.383064069101E+00 0.E+00
-0.111022302463E-15 0.E+00 0.383064069101E+00
0.E+00 -0.277555756156E-16 0.E+0
I cannot understand "modo" and (1,1).
I suspect (1 1) means atom index, but there are p, p', p''. Which two (1,1)
stands for?
Also, in the README file. It shows something like C_{x,y,z} (0,0,0|1,1,1).
I can understand (0,0,0) are the three q-points, but what does this 1,1,1
stands for? It does not look like phonon branch index or atom index.
I have been struggling on this for quite some time. I really appreciate if
anyone can help me to clarify this issue.
Thanks,
Hua
-- next part --
An HTML attachment was scrubbed...
URL:
http://www.democritos.it/pipermail/pw_forum/attachments/20120208/8af9892f/attachment.htm
[Pw_forum] The iotk error occurrd in ph.x
Dear QE developer and users I am using an old cluster for calculation, where mpich2 1.2.1 was compiled by gortran 4.1.2 in Fedora 7 x86_64. The pw.x calculations were fine, but when I was executing ph.x, the iotk error occurred (no matter multi-nodes or single-core, this error occurred as soon as the input file read by ph.x) : # FROM IOTK LIBRARY, VERSION 1.2.0 # UNRECOVERABLE ERROR (ierr=1) # ERROR IN: iotk_scan_end (iotk_scan.f90:241) # CVS Revision: 1.23 # foundl # ERROR IN: iotk_close_read (iotk_files.f90:746) # CVS Revision: 1.20 As I known, such a problem is due to the bug of compiler, i.e., gfortran 4.1.2. However, since I just have the normal user authority in this cluster, I cannot upgrade gfortran or install other fortran compiler. Even mpich2 was installed in /usr, so I cannot re-compile it. In such a case, do I have to give up the phonon calculaitons in this cluster ? -- GAO Zhe CMC Lab, Materials Science & Engineering Department, Seoul National University, South Korea -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20120208/37513e5d/attachment.htm
[Pw_forum] Segfault in ph.x (QE 4.3.2, 4.3.1)
On Wed, 08 Feb 2012 12:04:00 +0100, Ij?s Mari wrote: > Atomic displacements: > There are 2 irreducible representations > > Representation 1 3 modes -A_1g G_1 G_1+ To be done > > Representation 2 3 modes -A_1g G_1 G_1+ To be done Dear Ij?s, the cutoff you are using is extremely high, are you sure you're not just running out of memory? Have you tried running with lower cutoffs and got the some error? Anyway, gfortran often miscompiles the phonon code, which version are you using? bests -- Lorenzo Paulatto IdR @ IMPMC/CNRS & Universit? Paris 6 phone: +33 (0)1 44275 084 / skype: paulatz www: http://www-int.impmc.upmc.fr/~paulatto/ mail: 23-24/4?16 Bo?te courrier 115, 4 place Jussieu 75252 Paris C?dex 05
[Pw_forum] Uranium pseudopotenial
Dear quantum espresso users, How I can add 6s state in the basis of Uranum pseudopotential? I could not find any example for two channels with the same L and different principal number, Thank you German Samolyuk, ORNL, USA the input I used to build pspot: title='U', zed=92., rel=1, iswitch=3, config='[Rn] 5f3.0 6d1.0 7s1.0', dft='PBE' rlderiv=2.80, eminld=-14.0, emaxld=4.0, deld=0.01, nld=4, / lloc=0, pseudotype=3, file_pseudopw='U.pbe-rrkjus.UPF', nlcc=.true., rcore=1.1, author='GDS', / 7 5F 4 3 3.00 0.0 1.060 1.600 5F 4 3 0.00 -.2 1.060 1.600 6P 2 1 6.00 0.0 1.300 1.700 6P 2 1 0.00 -.2 1.300 1.700 6D 3 2 1.00 0.0 2.400 2.600 6D 3 2 0.00 -.2 2.400 2.600 7S 1 0 1.00 0.0 3.500 3.500
[Pw_forum] pwscf: constrained total magnetic moment
Dear Gabriele,
I do two Fermi level fixed magnetic moment calculation.
There is print in subroutine electrons.f90 version 4.2.1 in line 462
WRITE( stdout, 9072 ) bfield(3),
and bfield(3) is determined as (ef_up-ef_dn)/2.
I gives some estimation for for "equivalent" magnetic field.
I understand that this is very rough estimate.
The question was how to recalulcate it to Tesla.
As I understand, it should be (ef_up-ef_dn)/(2mu_B).
Thanks,
German
ORNL
On Fri, Jan 6, 2012 at 3:26 AM, Gabriele Sclauzero
wrote:
> Dear German,
>
> ? ?In my previous answer I actually didn't get the point of your question.
> Now I think I've understood better your concern, but some points of your
> message still look obscure to me.
>
> Il giorno 05/gen/2012, alle ore 19.41, German Samolyuk ha scritto:
>
> Dear Gabriele,
>
> Thank you for your answer.
> I use version 4.2.1 and the input I used works for this version. For
> the 4.3.2 vesion the input should be as you mentioned
> tot_magnetization = some_number;. The results obtained with both
> version are the same.
>
>
> What results? Do you mean the same energy? Total magnetization? Both?
>
>
> The output from 4.2.1 version has value of "corresponding" magnetic
> field which according to text is defined as (e_f_up-e_f_dn)/2 in
> Rydbergs.
>
>
> Which text? Where did you find this definition? What is the "corresponding"
> magnetic field?
>
> My question was if there is any way to estimate corresponding external
> magnetic field?
>
>
> If you use the constrained magnetization method, the external magnetic field
> should vanish as you reach the ground state with the target magnetization
> (see the form of the penalty functional here below). This may not happen if
> your value of lambda is too small, because it needs to be large enough to
> make energetically favored the state with the target symmetry.?I don't think
> you should attach any physical importance to a nonzero?constraining
> field.?This said, things can be tricky, because if you increase lambda the
> charge density might become nearly impossible to??converge. If this happens,
> you should try to decrease mixing_beta and/or make a chain of several runs
> where you "restart" from the previous charge density by progressively
> increasing lambda.
>
> If you use the other method in version 4.3.2, there shouldn't be any
> external magnetic field.
>
> HTH
>
> GS
>
>
> Thank you,
>
> German
>
>
> On Thu, Jan 5, 2012 at 5:50 AM, Gabriele Sclauzero
> wrote:
>
> Dear German,
>
>
> ? ? have you checked the manuals? In the input description of PWscf I've
>
> found:
>
>
> ? ? ? ? ? ? ? ? ? ?'total':
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? total magnetization is constrained by
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? adding a penalty functional to the total energy:
>
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? LAMBDA * SUM_{i} ( magnetization(i) -
>
> fixed_magnetization(i) )**2
>
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? where the sum over i runs over the three
>
> components of
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? the magnetization. Lambda is a real number (see
>
> below).
>
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? Noncolinear case only. Use "tot_magnetization"
>
> for LSDA
>
>
> I think that, in order to fix the total magnetization in LSDA calculations
>
> (nspin=2), you should simply set the tot_magnetization parameter
>
> (two-Fermi-energies method) and not specify the constrained_magnetization
>
> keyword. Which version of QE are you using?
>
>
>
> HTH
>
>
> GS
>
>
>
> Il giorno 03/gen/2012, alle ore 20.10, German Samolyuk ha scritto:
>
>
> Dear all,
>
>
> I do self-consistent calculation of supercell with 135 Fe atoms with
>
> constrained magnetization.
>
>
> nspin = 2,
>
> fixed_magnetization(3) = 360.0,
>
> constrained_magnetization = 'total' ,
>
>
> The output value of magnetic field:
>
>
> ??Magnetic field ???= ??0.0785401 Ry
>
>
>
> In order to make a rough estimation of corresponding magnetic field in
>
> Tesla I should divide this value by mu_B and difference between
>
> total magnetic moments of system with constrain and without it?
>
>
> Thank you
>
>
> German Samolyuk
>
> Oak Rigde National Laboratory
>
> ___
>
> Pw_forum mailing list
>
> Pw_forum at pwscf.org
>
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
>
> ? Gabriele Sclauzero,?EPFL SB ITP CSEA
>
> ?? PH H2 462, Station 3,?CH-1015 Lausanne
>
>
>
>
>
>
>
>
>
> ___
>
> Pw_forum mailing list
>
> Pw_forum at pwscf.org
>
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
>
> ? Gabriele Sclauzero,?EPFL SB ITP CSEA
> ?? PH H2 462, Station 3,?CH-1015 Lausanne
>
>
>
>
>
>
>
[Pw_forum] Pseudopotentials
Dear Henry, if with "hybrid" you mean with a fraction of exact exchange. There are no pseudos with EXX fraction. I saw some calcuations made with pseudo potentials built with OEP approximation, simil Hartree-Fock. If you think that with exx the semicore could play a role, you need to include the semicore in your calculations. bests Layla 2012/2/8 henry odhiambo > Hello everyone, > > Kindly, could someone, please, share with me GGA and Hybrid > pseudopotentials and Green's functions for Te, Sb and Ge. I would highly > appreciate. > > Regards, > Henry O. Otunga > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > > -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20120208/53d77ec3/attachment.htm
[Pw_forum] Identifying the phonon mode from eigen vector
On Feb 8, 2012, at 4:00 AM, Ajit Vallabhaneni wrote: > > Dear users and developers, > > I am trying to understand how the atoms will be displaced for each > of the six phonon modes in graphene at various q's from the eigen vectors. At > gamma point (0,0,0), the eigen vectors are symmetric with the real part being > of the same sign for acoustic modes and of opposite signs for optical modes > with zero imaginary part for all of them. But for a different q like the one > showed below, I can tell which are long, transv and out of plane (z) modes, > but i want to know if it is possible to distinguish optical and acoustic > modes (assuming frequencies and dispersion curves are not known). Dear Ajit, In general, the distinction between acoustic and optical modes is conventional. "acoustic modes" are all those whose frequency is continuously connected with the three zone-center zero modes whose eigen-displacements correspond to a rigid translation of the entire crystal. All the others are, by convention, "optic modes". There is no rigorous way of telling which is which in the most general crystal structure at the most general q-point in the Brillouin zone. If asked to bet, I would say that modes whose displacements in the unit cell are "more parallel" would have a more "optic character" than those whose displacements are "anti-parallel", but again this statement is not rigorous, nor entirely meaningful either ... > On a relevant note, i read some previous archives of 2009 discussing the > possibility of identifying the frequencies by it's branch, not just from > lowest to highest, i was wondering if that is possible at all without looking > at the eigen vectors at each q. As said, the only meaningful statement can be made on the branch to which a mode belongs. If the frequency of the branch vanishes at q=0, then the entore branch is (conventionally) acoustic, otherwise it is deemed to be optic ... Hope this helps ... SB --- Stefano Baroni - SISSA & DEMOCRITOS National Simulation Center - Trieste http://stefano.baroni.me [+39] 040 3787 406 (tel) -528 (fax) / stefanobaroni (skype) La morale est une logique de l'action comme la logique est une morale de la pens?e - Jean Piaget -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20120208/5dbe87a5/attachment.htm
[Pw_forum] Segfault in ph.x (QE 4.3.2, 4.3.1)
Dear everyone, I'm trying to calculate the vibrational modes for diamond. I have compiled QE using gfortran and gcc, and pw.x runs just fine. When running ph.x, however, the execution stops with a segmentation fault (in my system, all information I get is: [1]+ Segmentation fault: 11 ../espresso-4.3.1/bin/ph.x < ph_diamond.in > ph_diamond.out). In the output file, the last lines are: Atomic displacements: There are 2 irreducible representations Representation 1 3 modes -A_1g G_1 G_1+ To be done Representation 2 3 modes -A_1g G_1 G_1+ To be done This is similar in both 4.3.1 and 4.3.2. I tried to search the archives but didn't find other reports matching this problem. I'm attaching the input files for my scf and ph calculation, if they happen to be of any help. Please let me know if any other information is required. Best wishes, Mari Ij?s Department of Applied Physics Aalto University School of Science -- next part -- A non-text attachment was scrubbed... Name: ph_diamond.in Type: application/octet-stream Size: 342 bytes Desc: ph_diamond.in Url : http://www.democritos.it/pipermail/pw_forum/attachments/20120208/98dcd41b/attachment.obj -- next part -- A non-text attachment was scrubbed... Name: scf_diamond.in Type: application/octet-stream Size: 1169 bytes Desc: scf_diamond.in Url : http://www.democritos.it/pipermail/pw_forum/attachments/20120208/98dcd41b/attachment-0001.obj
[Pw_forum] The iotk error occurrd in ph.x
2012/2/8 GAO Zhe : > Dear QE developer and users > I am using an old cluster for calculation, where mpich2 1.2.1 was compiled > by gortran 4.1.2 in Fedora 7 x86_64. > The pw.x calculations were fine, but when I was executing ph.x, the iotk > error occurred (no matter multi-nodes or single-core, this error occurred as > soon as the input file read by ph.x) : > # FROM IOTK LIBRARY, VERSION 1.2.0 > # UNRECOVERABLE ERROR (ierr=1) > # ERROR IN: iotk_scan_end (iotk_scan.f90:241) > # CVS Revision: 1.23 > # foundl > # ERROR IN: iotk_close_read (iotk_files.f90:746) > # CVS Revision: 1.20 > As I known, such a problem is due to the bug of compiler, i.e., gfortran > 4.1.2. However, since I just have the normal user authority in this cluster, > I cannot upgrade gfortran or install other fortran compiler. Even mpich2 was > installed in /usr, so I cannot re-compile it. yes, you can compile and install both the GCC compilers (including gfortran) and the MPI library without superuser privilege and override the system defaults. it just takes some effort and requires reading of the corresponding documentation. i have done both many times on very different machines without having root access. axel. > In such a case, do I have to give up the phonon calculaitons in this cluster > ? > > -- > GAO Zhe > CMC Lab, Materials Science & Engineering Department, > Seoul National University, South Korea > > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > -- Dr. Axel Kohlmeyer akohlmey at gmail.com ?http://goo.gl/1wk0 College of Science and Technology Temple University, Philadelphia PA, USA.
[Pw_forum] Pseudopotentials
Hello everyone, Kindly,? could someone, please, share with me GGA and Hybrid pseudopotentials and Green's functions for Te, Sb and Ge. I would highly appreciate. Regards, Henry O. Otunga -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20120208/f5311421/attachment.htm
