<div> </div>
<div> </div>
<div> Dear Miguel,</div>
<div> </div>
<div> Yes, you are right. The linear response calculation needs </div>
<div> much time even for calculating phonons at Gamma point. </div>
<div> Only for Gamma point phonon one </div>
<div> can use "Frozen</div>
<div> phonon" method because</div>
<div> it is much faster.</div>
<div> </div>
<div> The calculations can be got stalled for many reasons.</div>
<div> You have not mentioned whether at last </div>
<div> (I mean staying in the silence mode for few weeks) your jobs</div>
<div> properly (without any error) get over or not???</div>
<div> </div>
<div> If yes, then it can be due to the problem in your </div>
<div> parallel computer - may be out of 16 processors </div>
<div> some particular processor was down for some time.</div>
<div> </div>
<div> If the answer of the above question is no, then the</div>
<div> memory of your computer is not sufficient </div>
<div> enough to tackle a large system with lot of atoms</div>
<div> because it needs large memory and as well as</div>
<div> generates a huge tmp file.</div>
<div> </div>
<div> All input parameters </div>
<div> (deguass, la2F, filedvscf ......) also should be given correctly for </div>
<div> electron-phonon calculation. </div>
<div> </div>
<div> regards,</div>
<div> SKS</div>
<div> JNCASR</div>
<div> Bangalore<br><br> </div>
<div><span class="gmail_quote">On 8/8/08, <b class="gmail_sendername">Miguel Martínez</b> <<a href="mailto:miguel.martinez@ehu.es">miguel.martinez@ehu.es</a>> wrote:</span>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">Dear everybody,<br><br>I've been running into a few issues as of late, when running some<br>calculations, it seems the calculations have stalled for a pretty long<br>
period of time.<br><br>The one that troubles me less, taking into account that it's pretty<br>expensive, is an electron-phonon calculation. This is a pretty large<br>system (20 atoms in the primitive cell) and computing the phonons at<br>
gamma took ages (well two days using 4 processors). It would be great<br>if some output were written just to remind you the code is still<br>working, maybe something related to convergence. But I find somewhat<br>disturbing that no line was written since the electron phonon<br>
calculation began.<br><br>My other issue is a very similar stalling in a Raman calculation. No,<br>it's not the same structure. Here, I lack from experience, since this<br>is the first Raman calculation I've done outside the examples.<br>
<br>Self-consistent input is:<br>&control<br> calculation='scf'<br> restart_mode='from_scratch',<br> prefix='sih4',<br> tstress = .true.<br> tprnfor = .true.<br> pseudo_dir = '/home/wmbmacam/pseudo/',<br>
outdir='/scratch/wmbmacam/fdd2-raman/'<br>/<br>&system<br> ibrav= 10, celldm(1) = 12.65898481, celldm(2) = 1.07100858,<br>celldm(3) = 0.621208699, nat=10, ntyp= 2,<br> ecutwfc = 55.0<br> nr1= 72, nr2= 72, nr3= 72<br>
/<br>&electrons<br> conv_thr = 1.0d-8<br> mixing_beta = 0.7<br>/<br>ATOMIC_SPECIES<br>Si 28.0855 Si.pbe-rrkj.UPF<br>H 1.0079 H.pbe-vbc.UPF<br>ATOMIC_POSITIONS {crystal}<br>Si 0.614368447799 0.385631552201 0.614368447799<br>
Si 0.864368447799 0.635631552201 0.864368447799<br>H 0.442372877617 0.836174882136 0.364427192728<br>H 0.364427192728 0.357025047520 0.442372877617<br>H 0.892974952480 0.885572807272 0.413825117864<br>H 0.413825117864 0.807627122383 0.892974952480<br>
H 0.378768583324 0.338682246661 0.003482898532<br>H 0.003482898532 0.279066271483 0.378768583324<br>H 0.970933728517 0.246517101468 0.911317753339<br>H 0.911317753339 0.871231416676 0.970933728517<br>K_POINTS {automatic}<br>
12 12 12 1 1 1<br><br><br>The phonon input is:<br><br>phonons of Sih4 at Gamma<br>&inputph<br>tr2_ph=1.0d-16,<br>prefix='sih4',<br>epsil=.true.,<br>trans=.true.,<br>lraman=.true.,<br>elop=.true.,<br>amass(1)=28.0855,<br>
amass(2)=1.00794,<br>outdir='/scratch/wmbmacam/fdd2-raman/',<br>fildyn='sih4.dynG',<br>fildrho='sih4.drho',<br>/<br>0.0 0.0 0.0<br><br><br>The calculations stalled right after the effective charges calculations.<br>
After that, it spent 24 hours without writing anything (it needed 10<br>hours to compute the effective charges). No convergence, no threshold,<br>no crashes, no nothing.<br><br>Is this behaviour normal? Or are my compiled binaries just telling me I<br>
should stop bothering them in August?<br><br>Finally, I just wanted to note that the amasses in example 15 are<br>seemingly wrong. While the masses in the scf run are fine, the ones<br>used in the phonon calculations correspond to GaAs instead of AlAs.<br>
<br>Regards,<br><br>Miguel<br><br>--<br>----------------------------------------<br>Miguel Martínez Canales<br> Dto. Física de la Materia Condensada<br> UPV/EHU<br> Facultad de Ciencia y Tecnología<br> Apdo. 644<br> 48080 Bilbao (Spain)<br>
Fax: +34 94 601 3500<br>Tlf: +34 94 601 5326<br>----------------------------------------<br><br>"If you have an apple and I have an apple and<br>we exchange these apples then you and I will<br>still each have one apple. But if you have an<br>
idea and I have an idea and we exchange these<br>ideas, then each of us will have two ideas."<br><br>George Bernard Shaw<br><br>_______________________________________________<br>Pw_forum mailing list<br><a href="mailto:Pw_forum@pwscf.org">Pw_forum@pwscf.org</a><br>
<a href="http://www.democritos.it/mailman/listinfo/pw_forum">http://www.democritos.it/mailman/listinfo/pw_forum</a><br></blockquote></div><br>