<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <p>Dear Paolo,</p>
    <p>the "patch" is working. Now I can't calculate the dvscf files
      using images... I need to use the GRID way...<br>
      Not really a solution but at least the problem won't occur in the
      official version :)</p>
    <p>Yet, the 5th problem remains... Restarting an el-ph calculation
      does not work no matter if I use the<br>
      calculation from file (trans=.false.) or not. At least for me it
      does not work...<br>
    </p>
    <p>Regards</p>
    <p>Thomas<br>
    </p>
    <br>
    <div class="moz-cite-prefix">On 09/23/2016 06:21 PM, Paolo Giannozzi
      wrote:<br>
    </div>
    <blockquote
cite="mid:CAPMgbCuFvw5bxU9_auUMEKvreddQiMi_oDX-+dFDsbnTuSO9gw@mail.gmail.com"
      type="cite">
      <meta http-equiv="Content-Type" content="text/html;
        charset=windows-1252">
      <div dir="ltr">
        <div>Is the following patch doing (that is: stopping) the job?<br>
          ---<br>
          Index: /home/giannozz/trunk/espresso/PHonon/PH/phq_readin.f90<br>
===================================================================<br>
          --- /home/giannozz/trunk/espresso/PHonon/PH/phq_readin.f90   
          (revision 13008)<br>
          +++ /home/giannozz/trunk/espresso/PHonon/PH/phq_readin.f90   
          (working copy)<br>
          @@ -679,6 +679,8 @@<br>
           <br>
             IF(elph.and.nimage>1) call errore('phq_readin',&<br>
                  'el-ph with images not implemented',1)<br>
          +  IF( fildvscf /= ' ' .and. nimage > 1 ) call
          errore('phq_readin',&<br>
          +       'saving dvscf to file images not implemented',1)<br>
             <br>
             IF (elph.OR.fildvscf /= ' ') lqdir=.TRUE.<br>
          ---<br>
          <br>
        </div>
        <div>Paolo</div>
      </div>
      <div class="gmail_extra"><br>
        <div class="gmail_quote">On Fri, Sep 23, 2016 at 4:45 PM, Thomas
          Brumme <span dir="ltr"><<a moz-do-not-send="true"
              href="mailto:thomas.brumme@mpsd.mpg.de" target="_blank">thomas.brumme@mpsd.mpg.de</a>></span>
          wrote:<br>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div bgcolor="#FFFFFF" text="#000000">
              <p>I meanwhile had a discussion with Lorenzo Paulatto
                about a similar problem.<br>
              </p>
              <p>I think that it might be a rather specific problem. As
                soon as I parallelize only over<br>
                q points using start_q and last_q there is no problem -
                also for restarting.</p>
              <p>Using images I can, in principle, even create the full
                dvscf files, without having to<br>
                rerun the calculation without images, using split and
                cat on the different dvscf files<br>
                in the different temp folders. It's tedious but it
                works. Yet, in future I will use only<br>
                the parallelization over q points for the calculation of
                the dvscf.</p>
              <p>In summary, the parallelization for PH is not
                straightforward and I think that it<br>
                might help to store, e.g., the dvscf files for different
                representations separately.<br>
                But Lorenzo mentioned that system administrators
                complain if the number of<br>
                written files is large... It could be helpful if there
                would be a kind of summary<br>
                what can be done using images and what not... I.e. dvscf
                (and el-ph) does not<br>
                work if image parallelization is used, especially if the
                different representations<br>
                of one q point are split across different images. For
                el-ph the code does not<br>
                start, but maybe a similar check can be added for the
                dvscf files?</p>
              <p>Well, or maybe not, I don't know :)<br>
              </p>
              <div>
                <div class="h5"> <br>
                  <div>On 09/23/2016 04:24 PM, Paolo Giannozzi wrote:<br>
                  </div>
                  <blockquote type="cite">
                    <div dir="ltr">has anybody any idea? P.<br>
                    </div>
                    <div class="gmail_extra"><br>
                      <div class="gmail_quote">On Wed, Sep 14, 2016 at
                        1:30 PM, Thomas Brumme <span dir="ltr"><<a
                            moz-do-not-send="true"
                            href="mailto:thomas.brumme@mpsd.mpg.de"
                            target="_blank">thomas.brumme@mpsd.mpg.de</a>></span>
                        wrote:<br>
                        <blockquote class="gmail_quote" style="margin:0
                          0 0 .8ex;border-left:1px #ccc
                          solid;padding-left:1ex">Dear all,<br>
                          <br>
                          I think I found a bug in the image
                          parallelization of PH - or I'm doing<br>
                          something wrong.<br>
                          I used the version 5.4 but the problem is also
                          there if I use the 6.0 beta.<br>
                          Maybe someone remembers my email few days ago
                          to the normal email list<br>
                          concerning<br>
                          the parallelization using the GRID technique -
                          the problem I encounter<br>
                          here is essentially<br>
                          the same. As an example, I use a modified
                          run_example_1 of the<br>
                          Recover_example<br>
                          directory of PH.<br>
                          <br>
                          Description of the problem:<br>
                          <br>
                          0. (Following the example) I did an scf
                          calculation using 2 CPUs with:<br>
                          <br>
                            &control<br>
                               calculation='scf'<br>
                               restart_mode='from_scratch',<br>
                               prefix='aluminum',<br>
                               pseudo_dir = './',<br>
                               outdir='./tempdir/'<br>
                            /<br>
                            &system<br>
                               ibrav=  2, celldm(1) =7.5, nat= 1, ntyp=
                          1,<br>
                               ecutwfc =15.0,<br>
                               occupations='smearing',
                          smearing='methfessel-paxton', degauss=0.05,<br>
                               la2F = .true.,<br>
                            /<br>
                            &electrons<br>
                               conv_thr =  1.0d-8<br>
                               mixing_beta = 0.7<br>
                            /<br>
                          ATOMIC_SPECIES<br>
                            Al  26.98 Al.pz-vbc.UPF<br>
                          ATOMIC_POSITIONS<br>
                            Al 0.00 0.00 0.00<br>
                          K_POINTS {automatic}<br>
                            16 16 16  0 0 0<br>
                          <br>
                          <br>
                          1. I'll do the scf calculation using 2 CPUS
                          and:<br>
                          <br>
                            &control<br>
                               calculation='scf'<br>
                               restart_mode='from_scratch',<br>
                               prefix='aluminum',<br>
                               pseudo_dir = './',<br>
                               outdir='./tempdir/'<br>
                            /<br>
                            &system<br>
                               ibrav=  2, celldm(1) =7.5, nat= 1, ntyp=
                          1,<br>
                               ecutwfc =15.0,<br>
                               occupations='smearing',
                          smearing='methfessel-paxton', degauss=0.05<br>
                            /<br>
                            &electrons<br>
                               conv_thr =  1.0d-8<br>
                               mixing_beta = 0.7<br>
                            /<br>
                          ATOMIC_SPECIES<br>
                            Al  26.98 Al.pz-vbc.UPF<br>
                          ATOMIC_POSITIONS<br>
                            Al 0.00 0.00 0.00<br>
                          K_POINTS {automatic}<br>
                            8 8 8  0 0 0<br>
                          <br>
                          <br>
                          2. I'll do a phonon calculation including
                          storing the dvscf files and<br>
                          using images.<br>
                          More specifically I used:<br>
                          <br>
                          mpirun -np 4 ph.x -ni 2 < <a
                            moz-do-not-send="true"
                            href="http://al.elph.in" rel="noreferrer"
                            target="_blank">al.elph.in</a><br>
                          <br>
                          with <a moz-do-not-send="true"
                            href="http://al.elph.in" rel="noreferrer"
                            target="_blank">al.elph.in</a> given by:<br>
                          <br>
                          Electron-phonon coefficients for Al<br>
                            &inputph<br>
                             tr2_ph=1.0d-10,<br>
                             prefix='aluminum',<br>
                             fildvscf='aldv',<br>
                             amass(1)=26.98,<br>
                             outdir='./tempdir/',<br>
                             fildyn='al.dyn',<br>
                          !  electron_phonon='interpolated'<wbr>,<br>
                          !  el_ph_sigma=0.005,<br>
                          !  el_ph_nsigma=10,<br>
                          !  recover=.true.<br>
                          !  trans=.false.,<br>
                             ldisp=.true.<br>
                             max_seconds=6,<br>
                             nq1=4, nq2=4, nq3=4<br>
                            /<br>
                          <br>
                          I used max_seconds in order to simulate the
                          finite run time we have on<br>
                          our HPC.<br>
                          Restarting with recover=.true. works fine...
                          I.e. I used:<br>
                          <br>
                          Electron-phonon coefficients for Al<br>
                            &inputph<br>
                             tr2_ph=1.0d-10,<br>
                             prefix='aluminum',<br>
                             fildvscf='aldv',<br>
                             amass(1)=26.98,<br>
                             outdir='./tempdir/',<br>
                             fildyn='al.dyn',<br>
                          !  electron_phonon='interpolated'<wbr>,<br>
                          !  el_ph_sigma=0.005,<br>
                          !  el_ph_nsigma=10,<br>
                             recover=.true.<br>
                          !  trans=.false.,<br>
                             ldisp=.true.<br>
                             max_seconds=6,<br>
                             nq1=4, nq2=4, nq3=4<br>
                            /<br>
                          <br>
                          <br>
                          3. Now I want to collect all data using no
                          images:<br>
                          <br>
                          mpirun -np 2 ph.x < <a
                            moz-do-not-send="true"
                            href="http://al.elph.in" rel="noreferrer"
                            target="_blank">al.elph.in</a><br>
                          <br>
                          with the same input file as given in 2.<br>
                          <br>
                          I'll get the error "Possibly too few bands at
                          point ..." once the code<br>
                          wants to<br>
                          recalculate the wave functions for the q
                          points which were calculated<br>
                          only on<br>
                          the second image, i.e., for q points 6, 7, and
                          8.<br>
                          <br>
                          If I check the charge_density.dat files in the
                          subfolders of the q<br>
                          points in the<br>
                          _ph0 directory I find that they're empty.
                          Thus, I copied the q<br>
                          subfolders of the<br>
                          second image by hand to the folder of the
                          first image using:<br>
                          <br>
                          cp -r _ph1/aluminum.q_* _ph0/<br>
                          <br>
                          If I now restart without images, using the
                          input of 2. it works...<br>
                          Everything is fine...<br>
                          <br>
                          <br>
                          4. Now I can also calculate the el-ph
                          parameters using the input:<br>
                          <br>
                          Electron-phonon coefficients for Al<br>
                            &inputph<br>
                             tr2_ph=1.0d-10,<br>
                             prefix='aluminum',<br>
                             fildvscf='aldv',<br>
                             amass(1)=26.98,<br>
                             outdir='./tempdir/',<br>
                             fildyn='al.dyn',<br>
                             electron_phonon='<wbr>interpolated',<br>
                             el_ph_sigma=0.005,<br>
                             el_ph_nsigma=10,<br>
                          !  recover=.true.<br>
                             trans=.false.,<br>
                             ldisp=.true.<br>
                          !  max_seconds=6,<br>
                             nq1=4, nq2=4, nq3=4<br>
                            /<br>
                          <br>
                          <br>
                          5. Another problem I encounter is the
                          following... Suppose the run time<br>
                          is not enough to<br>
                          finish the el-ph calculations, i.e., instead
                          of the input in 4. I use:<br>
                          <br>
                          Electron-phonon coefficients for Al<br>
                            &inputph<br>
                             tr2_ph=1.0d-10,<br>
                             prefix='aluminum',<br>
                             fildvscf='aldv',<br>
                             amass(1)=26.98,<br>
                             outdir='./tempdir/',<br>
                             fildyn='al.dyn',<br>
                             electron_phonon='<wbr>interpolated',<br>
                             el_ph_sigma=0.005,<br>
                             el_ph_nsigma=10,<br>
                          !  recover=.true.<br>
                             trans=.false.,<br>
                             ldisp=.true.<br>
                             max_seconds=6,<br>
                             nq1=4, nq2=4, nq3=4<br>
                            /<br>
                          <br>
                          The code will stop at a certain point (in my
                          case the 4th q point). If I<br>
                          now restart the calculation<br>
                          using:<br>
                          <br>
                          Electron-phonon coefficients for Al<br>
                            &inputph<br>
                             tr2_ph=1.0d-10,<br>
                             prefix='aluminum',<br>
                             fildvscf='aldv',<br>
                             amass(1)=26.98,<br>
                             outdir='./tempdir/',<br>
                             fildyn='al.dyn',<br>
                             electron_phonon='<wbr>interpolated',<br>
                             el_ph_sigma=0.005,<br>
                             el_ph_nsigma=10,<br>
                             recover=.true.<br>
                             trans=.false.,<br>
                             ldisp=.true.<br>
                          !  max_seconds=6,<br>
                             nq1=4, nq2=4, nq3=4<br>
                            /<br>
                          <br>
                          I get (again) the error message "Possibly too
                          few bands at point ..."<br>
                          once the code wants to calculate<br>
                          the wave functions for the 4th q point (the
                          one it stopped before)...<br>
                          All other points are fine...<br>
                          <br>
                          <br>
                          I think that the whole problem is related to
                          the storing of the wave<br>
                          functions and the charge density.<br>
                          Maybe I'm doing something really wrong, but I
                          don't see any obvious<br>
                          error in the input... Also I don't<br>
                          see any input variable for ph which influences
                          the saving of wave<br>
                          functions...<br>
                          <br>
                          Regards<br>
                          <br>
                          Thomas<br>
                          <br>
                          --<br>
                          Dr. rer. nat. Thomas Brumme<br>
                          Max Planck Institute for the Structure and
                          Dynamics of Matter<br>
                          Luruper Chaussee 149<br>
                          22761 Hamburg<br>
                          <br>
                          Tel:  <a moz-do-not-send="true"
                            href="tel:%2B49%20%280%2940%208998%206557"
                            value="+494089986557" target="_blank">+49
                            (0)40 8998 6557</a><br>
                          <br>
                          email: <a moz-do-not-send="true"
                            href="mailto:Thomas.Brumme@mpsd.mpg.de"
                            target="_blank">Thomas.Brumme@mpsd.mpg.de</a><br>
                          <br>
                          ______________________________<wbr>_________________<br>
                          Q-e-developers mailing list<br>
                          <a moz-do-not-send="true"
                            href="mailto:Q-e-developers@qe-forge.org"
                            target="_blank">Q-e-developers@qe-forge.org</a><br>
                          <a moz-do-not-send="true"
                            href="http://qe-forge.org/mailman/listinfo/q-e-developers"
                            rel="noreferrer" target="_blank">http://qe-forge.org/mailman/li<wbr>stinfo/q-e-developers</a><br>
                        </blockquote>
                      </div>
                      <br>
                      <br clear="all">
                      <br>
                      -- <br>
                      <div data-smartmail="gmail_signature">
                        <div dir="ltr">
                          <div>
                            <div dir="ltr">
                              <div>Paolo Giannozzi, Dip. Scienze
                                Matematiche Informatiche e Fisiche,<br>
                                Univ. Udine, via delle Scienze 208,
                                33100 Udine, Italy<br>
                                Phone <a moz-do-not-send="true"
                                  href="tel:%2B39-0432-558216"
                                  value="+390432558216" target="_blank">+39-0432-558216</a>,
                                fax <a moz-do-not-send="true"
                                  href="tel:%2B39-0432-558222"
                                  value="+390432558222" target="_blank">+39-0432-558222</a><br>
                                <br>
                              </div>
                            </div>
                          </div>
                        </div>
                      </div>
                    </div>
                    <br>
                    <fieldset></fieldset>
                    <br>
                    <pre>______________________________<wbr>_________________
Q-e-developers mailing list
<a moz-do-not-send="true" href="mailto:Q-e-developers@qe-forge.org" target="_blank">Q-e-developers@qe-forge.org</a>
<a moz-do-not-send="true" href="http://qe-forge.org/mailman/listinfo/q-e-developers" target="_blank">http://qe-forge.org/mailman/<wbr>listinfo/q-e-developers</a>
</pre>
    </blockquote>
    

    <pre cols="72">-- 
Dr. rer. nat. Thomas Brumme
Max Planck Institute for the Structure and Dynamics of Matter
Luruper Chaussee 149
22761 Hamburg

Tel:  <a moz-do-not-send="true" href="tel:%2B49%20%280%2940%208998%206557" value="+494089986557" target="_blank">+49 (0)40 8998 6557</a>

email: <a moz-do-not-send="true" href="mailto:Thomas.Brumme@mpsd.mpg.de" target="_blank">Thomas.Brumme@mpsd.mpg.de</a>
</pre>
  </div></div></div>


______________________________<wbr>_________________

Q-e-developers mailing list

<a moz-do-not-send="true" href="mailto:Q-e-developers@qe-forge.org">Q-e-developers@qe-forge.org</a>

<a moz-do-not-send="true" href="http://qe-forge.org/mailman/listinfo/q-e-developers" rel="noreferrer" target="_blank">http://qe-forge.org/mailman/<wbr>listinfo/q-e-developers</a>


</blockquote></div>


-- 
<div class="gmail_signature" data-smartmail="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div>Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222

</div></div></div></div></div>
</div>


<fieldset class="mimeAttachmentHeader"></fieldset>
<pre wrap="">_______________________________________________
Q-e-developers mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Q-e-developers@qe-forge.org">Q-e-developers@qe-forge.org</a>
<a class="moz-txt-link-freetext" href="http://qe-forge.org/mailman/listinfo/q-e-developers">http://qe-forge.org/mailman/listinfo/q-e-developers</a>
</pre>

</blockquote>
<pre class="moz-signature" cols="72">-- 
Dr. rer. nat. Thomas Brumme
Max Planck Institute for the Structure and Dynamics of Matter
Luruper Chaussee 149
22761 Hamburg

Tel:  +49 (0)40 8998 6557

email: <a class="moz-txt-link-abbreviated" href="mailto:Thomas.Brumme@mpsd.mpg.de">Thomas.Brumme@mpsd.mpg.de</a>
</pre></body></html>