[Pw_forum] time-reversal symmetry broken
Dear All, I have a question concerning the breaking of time-reversal symmetry within PWscf. For my proposes I have modified the code and included a (periodic) vector potential into the Hamiltonian. The code gives very good results for the quantity that I am interested in for molecules, i.e. gamma-point calculations. However, when I try to do calculations for periodic solids, the results worsen. I think the code (unless you do spin-orbit coupling) assumes time- reversal symmetry. In principle, including a vector potential breaks time-reversal symmetry. But, in the cases of my gamma-point calculations that did not cause any problems ( psi_(-k)=psi_(k)* for k=gamma ) and hence, I get good results for molecules. However, maybe this is the problem with my calculations for extended systems: including the vector potential breaks time-reversal symmetry, and in general psi_(-k) is not psi_(k)*, but the code is not "aware" that the symmetry has been broken. Thus, my question is: is there a simple way to tell the code that time-reversal symmetry is broken without getting into spin-orbit stuff? Thanks, Timo Thonhauser, MIT __ Dr. TIMO THONHAUSER Department of Materials Science and Engineering (617) 253 6026 MIT, 13-4084, 77 Massachusetts Avenue thonhauser at mit.edu Cambridge, MA 02139-4307, USAwww.mit.edu/~thonhaus __
[Pw_forum] still can not get the modes with the dynmat.x
hi Marcel: i am just looking for the file .and at last ,i found it .it was put in my home directory.hehe! and my another question is whether i can specify the destination of the file ,like in the currently directory? and i have a third question is the interactively process,just called serial process, just now ,i tried it .i can not work,the load of cpu is zero so ,i do not know why ! i need your help thank you 2007/11/20, Marcel Mohr : > > Hi Yuehua > > Try to run dynmat interactively without using the queue manager ! > > There should be a files called dynmat.out, dynmat.mold, dynmat.x... > See in example09. > > Nobody can guess, where your computer puts them ;) > > Cheers Marcel > > > Marcel Mohr Institut f?r Festk?rperphysik, TU Berlin > marcel(at)physik.tu-berlin.de Sekr. EW 5-4 > TEL: +49-30-314 24442 Hardenbergstr. 36 > FAX: +49-30-314 27705 10623 Berlin > > > On Tue, 20 Nov 2007, xu yuehua wrote: > > > i use dynmat.x to add lo-to split ,and want to get the new frequencies > and > > modes, and only the frequencies is present . > > my infile now: > > > > &input > > > fildyn='/raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/*.dynG', > > asr='one-dim', > > q(1)=0,q(2)=0,q(3)=12, > > filout= 'dynmat.out'/ > > > > > > > > but there is not dynmat.out .only the file like pwscf.sh.o8645 > > and in pwscf.sh.o8645 > > > > there is : > > > > ue Nov 20 17:29:58 CST 2007 > > > > Reading Dynamical Matrix from file > > > /raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/fourringwater.dynG > > asr rotation axis in 1D system=3 > > Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =0.815659E-01 > > Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=0.697493E-01 > > > > Polarizability (A^3 units) > > multiply by 0.971817 for Clausius-Mossotti correction > > 5.6254140.0006580.000133 > > 0.0006585.6241280.000114 > > 0.0001330.0001148.860593 > > > > IR cross sections are in (D/A)^2/amu units > > # mode [cm-1] [THz] IR > >1 0.000.0. > >2 0.000.0. > >3 0.000.0. > >4 0.000.0. > >5 49.161.47380.0064 > >6 63.881.91510.0043 > >7 68.562.05540.0016 > >8106.293.18650.0002 > >9156.914.70400.0002 > > 10252.937.58260.2436 > > 11254.667.63440.4708 > > 12257.937.73240.6958 > > 13564.27 16.91630.4540 > > 14631.40 18.92870.1605 > > 15632.64 18.96591.7223 > > 16633.03 18.97771.8109 > > 17705.34 21.14551.3172 > > 18750.19 22.49003.3584 > > 19750.80 22.50843.3466 > > 20793.99 23.80300.0001 > > 21854.32 25.61173.6181 > > > > ... > > > > > > > > > > no mode ! .i am puzzled .and where is my error ? > > > > > > -- > > Xu Yuehua > > physics Department of Nanjing university > > China > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > > -- Xu Yuehua physics Department of Nanjing university China -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20071120/5c67212b/attachment.htm
[Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation
Hi Nicola, Thanks for your feedback. I was aware of your page on Quasiamore and actually have visited it in the past, and should have had a look before I sent in my forum posting. Re the PS3, I told my teenage gamer today I'm going to get a unit to experiment with (I've already begun downloading Yellow Dog Linux). He was very happy, because if I fail porting PWscf, he gets it! Over the last couple of days, my research reflects exactly what you mentioned on the detailed performance of the switch. Also, there may be some issues regarding Fortran compilers, but I've got some friends at IBM that could help with this. What concerns me most is the 256 MB on board memory limit, and I've found no hints online on how to hack this, soldering iron or otherwise. Ciao, -Paul Paul M. Grant, PhD Principal, W2AGZ Technologies Visiting Scholar, Applied Physics, Stanford University EPRI Science Fellow (Retired) IBM Research Staff Member Emeritus w2agz at pacbell.net http://www.w2agz.com ? ? -Original Message- From: pw_forum-bounces at pwscf.org [mailto:pw_forum-boun...@pwscf.org] On Behalf Of Nicola Marzari Sent: Tuesday, November 20, 2007 8:02 AM To: PWSCF Forum Subject: Re: [Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation Hi Paul, let us know if you get PWSCF on a PS3 ! I generally echo the other suggestions - the speed of the memory bus is critical, having fast memory (matched with the bus) is good, having the fastest clock on current Intel chip is good, having two cores is a 25-50% improvement, and having 4 cores is a negligible improvement on 2. If you plan to link blades in a cluster, it's vital to have fast, low latency communication - if you use gigabit, you might get good parallelism with as many as 8 blades, or as little as 2, depending on the quality of your switch, using a very recent mpi implementation, and the quality of your ethernet controllers on board. We have some tests on different platforms on http://quasiamore.mit.edu/pmwiki/index.php?n=Main.CP90Timings nicola Paul M. Grant wrote: > To All Forum Members: > > I?m planning on building a new Linux box (or boxes) to explore highly > correlated systems (e-p coupling plus LDA+U), and am seeking the > collective experience and advice of the PWscf community on a suitable, > inexpensive (< 2000 USD, MB+CPUs+RAM, exclusive of power supplies, > enclosures, and accessories) hardware platform. I emphasize that the > principal purpose of this new box would be exploratory, or for > development, not production. > > I?ve built several past platforms, both Windows and Linux based, using > server boards manufactured by Supermicro and have had generally good > experience and service (the factory is only 15 miles from where I > live). Currently, I use two machines with dual Xeon processors (single > core, 32 bits), one with 1 GHz cpus, 1 GB RAM, the other 2.4 GHz and > 3.25 GB RAM, both with bus speeds of 133 MHz, the newest 3 years old. > However, occasionally I run PWscf exercises on my little Thinkpad X41 > tablet (single processor, 1.5 GHz, 1.5 GB), and the scf computation will > run 3-5 times faster than on the other machines! I suspect this rather > surprising result is because the Thinkpad has a 400 MHz bus clock speed. > > One option I?m considering is using a ?gaming? or server class > motherboard with dual 2.33 GHz quad-core 64-bit processors, a 1333 MHz > FSB, and 16 GB RAM. Having said this, I?m not sure PWscf (and the > Fortran compilers available) can handle all this parallelism efficiently > on a single motherboard. I?ve noticed when running pw.x, the CPU > activity ?flips? between processors every several seconds, instead of > sharing each at 90-100% full time. > > On the other hand, one could consider building a small MPI-connected > cluster for about the same amount of money. > > When IBM announced a couple of years ago the incredible performance > details about the Cell processor that would go into Playstation 3, I > thought, ?Wow, maybe the future of computational physics rests with > gamers.? I?m sure most of you know this is actually beginning to > happen, spurred on by the fact that the PS3 is ?open architecture? and > can run a Linux distro. Moreover, there apparently are ?open software? > numerical analysis tools available from IBM. At least four US > universities are experimenting with off-the-shelf PS3 clusters, perhaps > one of the more interesting is at UMass, > _http://gravity.phy.umassd.edu/ps3.html_. In the last week or so, Sony > lowered the entry level price of the PS3 to 400 USD. So, a cluster of > four with a cheap switch could be purchased for about the same price at > the single motherboard configuration I mentioned above. > > My teenager, a gamer, tells me the PS3 has problems. He says it?s > unreliable and overheats and only has 256 MB RAM on board (he owns a > Wii, which outsells the PS3 in the
[Pw_forum] Pseudo_Generation/unknown pseudo error
Dear All, Here is the input / &input title='F', zed=9.0, rel=0, beta=0.5, rlderiv=2.0, eminld=-4.0, emaxld=4.0, deld=0.02d0, nld=4, iswitch=3, dft='BLYP', / 4 1S 1 0 2.0 1 2S 2 0 2.0 1 2P 2 1 5.0 1 3D 3 2 -1.0 1 &inputp pseudotype=3, lloc=2, file_pseudopw='FBLYP.VAN_AK', / 5 2S 1 0 2.00 0.00 1.10 1.70 2S 1 0 0.00 0.05 1.10 1.70 2P 2 1 5.00 0.00 1.10 1.70 2P 2 1 0.00 0.05 1.10 1.70 3D 3 2 -2.00 0.15 1.50 1.50 regards fariba nazari IASBS > On Tue, November 20, 2007 18:25, nazari at iasbs.ac.ir wrote: >> I face with unknown pseudo potential error. > > There are many things that can go wrong; conflicting input directive may > result in quite obscure error messages. Anyway if you provide the error > message, or the input file, it is more likely than someone will be able to > solve you problem. > > Another hint is to have a look in atomic_doc/pseudo_library/ where you can > find many examples of working input files. > > goodbye > > -- > Lorenzo Paulatto > +39 040 3787 511 > http://people.sissa.it/~paulatto/ > > > > SISSA Webmail https://webmail.sissa.it/ > Powered by SquirrelMail http://www.squirrelmail.org/ > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum >
[Pw_forum] Problem with nbnd in clusters
On Nov 20, 2007, at 15:55 , Cyrille Barreteau wrote: > Does anyone has experience about the behaviour > of the various diagonalization schemes? > Which one is better, in which case? the so-called conjugate-gradient diagonalization basically never fails; it uses less memory but it is much slower than the block Davidson algorithm. The latter is also quite robust but there is a tiny probability to end up in the wrong ground state. The only such case I know happens in some high-symmetry perovskites, if you start from superposition of atomic states, with occupied states only, and if the phase of the moon is close to 42. The problem disappears if you add a few more states, or if you start from random wavefunctions. I am quite surprised to hear that there is a case in which you get good results with 12 states and bad with 20. I wouldn't be surprised by the opposite. Could you please post one jobs showing the problem? Paolo --- Paolo Giannozzi, Dept of Physics, University of Udine via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] Pseudo_Generation/unknown pseudo error
Dear All, I wish to generate a pseudo potential. I have used the instruction in Metodi Numerici in struttura Elettronica. But when I applied the pseudo potential I face with unknown pseudo potential error. Also I did search in archive and find one answer but it doesn't work. Would you please let me know How I can use the result of step 1 and 2 of the mentioned text for producing pseudo potential. Because I wrote my input file for pseudo generation by comparing with oxygen input in pseudo-gen folder and run f.in in this folder. Regards Fariba Nazari IASBS
[Pw_forum] still can not get the modes with the dynmat.x
i use dynmat.x to add lo-to split ,and want to get the new frequencies and modes, and only the frequencies is present . my infile now: &input fildyn='/raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/*.dynG', asr='one-dim', q(1)=0,q(2)=0,q(3)=12, filout= 'dynmat.out'/ but there is not dynmat.out .only the file like pwscf.sh.o8645 and in pwscf.sh.o8645 there is : ue Nov 20 17:29:58 CST 2007 Reading Dynamical Matrix from file /raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/fourringwater.dynG asr rotation axis in 1D system=3 Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =0.815659E-01 Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=0.697493E-01 Polarizability (A^3 units) multiply by 0.971817 for Clausius-Mossotti correction 5.6254140.0006580.000133 0.0006585.6241280.000114 0.0001330.0001148.860593 IR cross sections are in (D/A)^2/amu units # mode [cm-1] [THz] IR 1 0.000.0. 2 0.000.0. 3 0.000.0. 4 0.000.0. 5 49.161.47380.0064 6 63.881.91510.0043 7 68.562.05540.0016 8106.293.18650.0002 9156.914.70400.0002 10252.937.58260.2436 11254.667.63440.4708 12257.937.73240.6958 13564.27 16.91630.4540 14631.40 18.92870.1605 15632.64 18.96591.7223 16633.03 18.97771.8109 17705.34 21.14551.3172 18750.19 22.49003.3584 19750.80 22.50843.3466 20793.99 23.80300.0001 21854.32 25.61173.6181 ... no mode ! .i am puzzled .and where is my error ? -- Xu Yuehua physics Department of Nanjing university China -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20071120/20d45094/attachment.htm
[Pw_forum] Problem with nbnd in clusters
Hi Nicola Of course my (iron) cluster is magnetic and my calculation is spin polarized. I also did play with the various smearings (fd,mp,mv) and the broadening but the result is quite insensitive to this parameter. I really have the feeling that for some reasons as Matteo mentionned davidson with 20 bands (who knows why...) gets trapped in some local minimum just for numerical reasons... But this trap seems quite stable since it remains even when changing the smearing. It was just bad (or good) luck. cyrille Nicola Marzari wrote: >Hi Cyrille, > >your tm dimer will likely require spin polarization, and >will have several scf solutions that correspond to different >S_z (that you can fix in several ways, either specifying the >number of spin up and spin down, or using a fermi_up and >fermi_down energy, to specify n_up - n_down but still treating >the system with a smearing and fractional occupations). > >Even for a given S_z, there will be different scf solutions >that have different spatial symmetry. > >So, in your case I would definitely switch spin-polarization, >and maybe play around with small and large smearings, to see what >happens, and look at the states, and the occupations. Anyhow, if >you get to a scf solution, that is a "good" solution for the GGA >problem, although might not be very physical. > >Also, have a look at Kulik et al PRL 2006. > > nicola > > >Cyrille Barreteau wrote: > > >>Dear pwscf_ers >> >>I am doing a very simple calculation on a transition metal >>dimer and I have a encoutered a strange problem (I love >>strange problems:-) >> >>I have done a first calculation with the default value >>of nbnd, ie nbnd=nelec/2*20% (=12 in my case) >> >>But since it is known that in clusters it is often good >>to increase nbnd I have performed another calculation with >>larger nbnd (=20). The result is quite different from >>the one at nbnd=12. >> >>I have then increased again nbnd up to 25 and then I recover >>the result obtained for nbnd=12 >> >>More problematic is the local density of states (I know >>it is not really a dos but a bunch of dirac peaks). >>In a dimer with the z axis along the direction connecting >>the two atoms, the xy and x^2-y^2 dos should be degenerate >>(if the supercell box is large enough). >>In fact the xy and x^2-y^2 are degenerate if nbnd=12 or 25 >>but there is a rather large splitting if nbnd=20. >> >>I am quite sure the result for nbnd=20 is not correct but >>what is the origin of this problem? >> >>Maybe I could try to use another diagonalization scheme.. >> >> thanks for reading my strange problems >> >> cyrille >> >> >> >> > > > > -- == Cyrille Barreteau | phone : +33 (0)1 69 08 29 51 CEA Saclay | fax : +33 (0)1 69 08 84 46 DSM/DRECAM/SPCSI | email cyrille.barreteau at cea.fr Batiment 462 | 91191 Gif sur Yvette Cedex FRANCE http://www-drecam.cea.fr/spcsi/index.php http://www-drecam.cea.fr/Images/Pisp/cbarreteau/cbarreteau_fr.html http://www-drecam.cea.fr/Phocea/Membres/Cours/index.php ==
[Pw_forum] Pseudo_Generation/unknown pseudo error
I regret to say that your input file works for me (QE-3.2). Also the output PP looks quite ok (as much as you can say without testing it). What version of QE are you using? P.S. I would just change this line: file_pseudopw='FBLYP.VAN_AK', to file_pseudopw='FBLYP.VAN_AK.UPF', in order to use "new" file format. -- Lorenzo Paulatto +39 040 3787 511 http://people.sissa.it/~paulatto/ SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/
[Pw_forum] about the influence of lo-to split on the vibrational model
i am very sorry .just now ,i did not find the dynmat.f90. now .i am reading . thank you for your kind criticism and your suggestion 2007/11/20, Paolo Giannozzi : > > > On Nov 20, 2007, at 4:16 , xu yuehua wrote: > > > and there is not the vector at all!! > > [...] > > and what is wrong ? i do not know > > please give me a detailed advise > > your attitude is wrong. Had you followed my previous > advice (not "advise") you would have found this: > > ! filout character output file containing frequencies and modes > !(default: filout='dynmat.out') > ! filmol character as above, in a format suitable for 'molden' > !(default: filmol='dynmat.mold') > ! filxsf character as above, in axsf format suitable for xcrysden > !(default: filmol='dynmat.axsf') > --- > Paolo Giannozzi, Dept of Physics, University of Udine > via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > -- Xu Yuehua physics Department of Nanjing university China -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20071120/d4676105/attachment.htm
[Pw_forum] Problem with nbnd in clusters
I just wanted to add that changing the diagonalization scheme to conjugate gradient 'cg' instead of the default 'david' seems to solve the problem in my case. I now find the same result with nbnd=12,20,25.. Does anyone has experience about the behaviour of the various diagonalization schemes? Which one is better, in which case? cyrille Cyrille Barreteau wrote: >Dear pwscf_ers > >I am doing a very simple calculation on a transition metal >dimer and I have a encoutered a strange problem (I love >strange problems:-) > >I have done a first calculation with the default value >of nbnd, ie nbnd=nelec/2*20% (=12 in my case) > >But since it is known that in clusters it is often good >to increase nbnd I have performed another calculation with >larger nbnd (=20). The result is quite different from >the one at nbnd=12. > >I have then increased again nbnd up to 25 and then I recover >the result obtained for nbnd=12 > >More problematic is the local density of states (I know >it is not really a dos but a bunch of dirac peaks). >In a dimer with the z axis along the direction connecting >the two atoms, the xy and x^2-y^2 dos should be degenerate >(if the supercell box is large enough). >In fact the xy and x^2-y^2 are degenerate if nbnd=12 or 25 >but there is a rather large splitting if nbnd=20. > >I am quite sure the result for nbnd=20 is not correct but >what is the origin of this problem? > >Maybe I could try to use another diagonalization scheme.. > > thanks for reading my strange problems > > cyrille > > > > -- == Cyrille Barreteau | phone : +33 (0)1 69 08 29 51 CEA Saclay | fax : +33 (0)1 69 08 84 46 DSM/DRECAM/SPCSI | email cyrille.barreteau at cea.fr Batiment 462 | 91191 Gif sur Yvette Cedex FRANCE http://www-drecam.cea.fr/spcsi/index.php http://www-drecam.cea.fr/Images/Pisp/cbarreteau/cbarreteau_fr.html http://www-drecam.cea.fr/Phocea/Membres/Cours/index.php ==
[Pw_forum] about the influence of lo-to split on the vibrational model
hi Paolo: oh .there is a file named dynmatrix.f90 .and i got confused . i list my in file and my out file here : &input fildyn='/raid/xyh/pwscf/*/*.dynG',asr='one-dim',q(1)=0,q(2)=0,q(3)=12/ AND THE OUTFILE HERE: Thu Oct 25 11:06:00 CST 2007 Reading Dynamical Matrix from file /raid/xyh/pwscf/*/*.dynG asr rotation axis in 1D system=3 Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =0.815659E-01 Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=0.697493E-01 Polarizability (A^3 units) multiply by 0.971817 for Clausius-Mossotti correction 5.6254140.0006580.000133 0.0006585.6241280.000114 0.0001330.0001148.860593 IR cross sections are in (D/A)^2/amu units # mode [cm-1] [THz] IR 1 0.000.0. 2 0.000.0. 3 0.000.0. 4 0.000.0. 5 49.161.4738 0.0064 6 63.881.91510.0043 7 68.562.05540.0016 8106.293.18650.0002 9156.914.70400.0002 10252.937.58260.2436 11254.66 7.63440.4708 and there is not the vector at all!! and the job script: #!/bin/bash # #$ -cwd #$ -j y #$ -S /bin/bash # MPI_DIR=/home/software/mpich/intel/bin date $MPI_DIR/mpirun -np $NSLOTS -machinefile $TMPDIR/machines -nolocal /home/xyh/intel_espresso-3.2/bin/dynmat.xhttp://www.democritos.it/pipermail/pw_forum/attachments/20071120/68ebc118/attachment.htm
[Pw_forum] Pseudo_Generation/unknown pseudo error
On Tue, November 20, 2007 18:25, nazari at iasbs.ac.ir wrote: > I face with unknown pseudo potential error. There are many things that can go wrong; conflicting input directive may result in quite obscure error messages. Anyway if you provide the error message, or the input file, it is more likely than someone will be able to solve you problem. Another hint is to have a look in atomic_doc/pseudo_library/ where you can find many examples of working input files. goodbye -- Lorenzo Paulatto +39 040 3787 511 http://people.sissa.it/~paulatto/ SISSA Webmail https://webmail.sissa.it/ Powered by SquirrelMail http://www.squirrelmail.org/
[Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation
Actually newer Woodcrest/Clovertown Xeon platform have dual independent FSBs, but this does not help much. Memory bandwidth available on each FSB is very limited - between 1/2 and 1/3 of what is available to modern desktop CPUs. There is no good "server" computational platform today, unfortunately . AMD's CPUs have weak FPU and Intel's have weak memory subsystem. Intel- based desktop single-socket platform is the best choice for now. On Tuesday 20 November 2007 14:33, Paolo Giannozzi wrote: > On Nov 20, 2007, at 4:07 , Paul M. Grant wrote: > > Currently, I use two machines with dual Xeon processors > > dual Xeon? very bad choice. They have a single memory > bus, so if you run two processes on the same motherboard, > they spend most of the time waiting for access to memory. > > P. > --- > Paolo Giannozzi, Dept of Physics, University of Udine > via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum
[Pw_forum] Problem with nbnd in clusters
Dear pwscf_ers I am doing a very simple calculation on a transition metal dimer and I have a encoutered a strange problem (I love strange problems:-) I have done a first calculation with the default value of nbnd, ie nbnd=nelec/2*20% (=12 in my case) But since it is known that in clusters it is often good to increase nbnd I have performed another calculation with larger nbnd (=20). The result is quite different from the one at nbnd=12. I have then increased again nbnd up to 25 and then I recover the result obtained for nbnd=12 More problematic is the local density of states (I know it is not really a dos but a bunch of dirac peaks). In a dimer with the z axis along the direction connecting the two atoms, the xy and x^2-y^2 dos should be degenerate (if the supercell box is large enough). In fact the xy and x^2-y^2 are degenerate if nbnd=12 or 25 but there is a rather large splitting if nbnd=20. I am quite sure the result for nbnd=20 is not correct but what is the origin of this problem? Maybe I could try to use another diagonalization scheme.. thanks for reading my strange problems cyrille -- == Cyrille Barreteau | phone : +33 (0)1 69 08 29 51 CEA Saclay | fax : +33 (0)1 69 08 84 46 DSM/DRECAM/SPCSI | email cyrille.barreteau at cea.fr Batiment 462 | 91191 Gif sur Yvette Cedex FRANCE http://www-drecam.cea.fr/spcsi/index.php http://www-drecam.cea.fr/Images/Pisp/cbarreteau/cbarreteau_fr.html http://www-drecam.cea.fr/Phocea/Membres/Cours/index.php ==
[Pw_forum] about the influence of lo-to split on the vibrational model
hi Paolo: oh .there is a file named dynmatrix.f90 .and i got confused . i list my in file and my out file here : &input fildyn='/raid/xyh/pwscf/*/*.dynG',asr='one-dim',q(1)=0,q(2)=0,q(3)=12/ AND THE OUTFILE HERE: Thu Oct 25 11:06:00 CST 2007 Reading Dynamical Matrix from file /raid/xyh/pwscf/*/*.dynG asr rotation axis in 1D system=3 Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =0.815659E-01 Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=0.697493E-01 Polarizability (A^3 units) multiply by 0.971817 for Clausius-Mossotti correction 5.6254140.0006580.000133 0.0006585.6241280.000114 0.0001330.0001148.860593 IR cross sections are in (D/A)^2/amu units # mode [cm-1] [THz] IR 1 0.000.0. 2 0.000.0. 3 0.000.0. 4 0.000.0. 5 49.161.47380.0064 6 63.881.91510.0043 7 68.562.05540.0016 8106.293.18650.0002 9156.914.70400.0002 10252.937.58260.2436 11254.667.63440.4708 and there is not the vector at all!! and the job script: #!/bin/bash # #$ -cwd #$ -j y #$ -S /bin/bash # MPI_DIR=/home/software/mpich/intel/bin date $MPI_DIR/mpirun -np $NSLOTS -machinefile $TMPDIR/machines -nolocal /home/xyh/intel_espresso-3.2/bin/dynmat.x: > > > On Nov 19, 2007, at 13:45 , xu yuehua wrote: > > > the mode has been recalculate,but there is only new frequency in > > outfile, > > no new vibrational vector present which is my wanted. how can i get > > it ? > > see the header of file dynmat.f90 > -- > Paolo Giannozzi, Dept of Physics, University of Udine > via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > -- lalala lalala -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20071120/204595f0/attachment.htm
[Pw_forum] still can not get the modes with the dynmat.x
Hi Yuehua Try to run dynmat interactively without using the queue manager ! There should be a files called dynmat.out, dynmat.mold, dynmat.x... See in example09. Nobody can guess, where your computer puts them ;) Cheers Marcel Marcel Mohr Institut f?r Festk?rperphysik, TU Berlin marcel(at)physik.tu-berlin.de Sekr. EW 5-4 TEL: +49-30-314 24442 Hardenbergstr. 36 FAX: +49-30-314 27705 10623 Berlin On Tue, 20 Nov 2007, xu yuehua wrote: > i use dynmat.x to add lo-to split ,and want to get the new frequencies and > modes, and only the frequencies is present . > my infile now: > > &input > fildyn='/raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/*.dynG', > asr='one-dim', > q(1)=0,q(2)=0,q(3)=12, > filout= 'dynmat.out'/ > > > > but there is not dynmat.out .only the file like pwscf.sh.o8645 > and in pwscf.sh.o8645 > > there is : > > ue Nov 20 17:29:58 CST 2007 > > Reading Dynamical Matrix from file > /raid/xyh/pwscf/fourringwater/pho_scf_relax_fromlqf_via_vaspz/fourringwater.dynG > asr rotation axis in 1D system=3 > Acoustic Sum Rule: || Z*(ASR) - Z*(orig)|| =0.815659E-01 > Acoustic Sum Rule: ||dyn(ASR) - dyn(orig)||=0.697493E-01 > > Polarizability (A^3 units) > multiply by 0.971817 for Clausius-Mossotti correction > 5.6254140.0006580.000133 > 0.0006585.6241280.000114 > 0.0001330.0001148.860593 > > IR cross sections are in (D/A)^2/amu units > # mode [cm-1] [THz] IR >1 0.000.0. >2 0.000.0. >3 0.000.0. >4 0.000.0. >5 49.161.47380.0064 >6 63.881.91510.0043 >7 68.562.05540.0016 >8106.293.18650.0002 >9156.914.70400.0002 > 10252.937.58260.2436 > 11254.667.63440.4708 > 12257.937.73240.6958 > 13564.27 16.91630.4540 > 14631.40 18.92870.1605 > 15632.64 18.96591.7223 > 16633.03 18.97771.8109 > 17705.34 21.14551.3172 > 18750.19 22.49003.3584 > 19750.80 22.50843.3466 > 20793.99 23.80300.0001 > 21854.32 25.61173.6181 > > ... > > > > > no mode ! .i am puzzled .and where is my error ? > > > -- > Xu Yuehua > physics Department of Nanjing university > China >
[Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation
Hi Paul, let us know if you get PWSCF on a PS3 ! I generally echo the other suggestions - the speed of the memory bus is critical, having fast memory (matched with the bus) is good, having the fastest clock on current Intel chip is good, having two cores is a 25-50% improvement, and having 4 cores is a negligible improvement on 2. If you plan to link blades in a cluster, it's vital to have fast, low latency communication - if you use gigabit, you might get good parallelism with as many as 8 blades, or as little as 2, depending on the quality of your switch, using a very recent mpi implementation, and the quality of your ethernet controllers on board. We have some tests on different platforms on http://quasiamore.mit.edu/pmwiki/index.php?n=Main.CP90Timings nicola Paul M. Grant wrote: > To All Forum Members: > > I?m planning on building a new Linux box (or boxes) to explore highly > correlated systems (e-p coupling plus LDA+U), and am seeking the > collective experience and advice of the PWscf community on a suitable, > inexpensive (< 2000 USD, MB+CPUs+RAM, exclusive of power supplies, > enclosures, and accessories) hardware platform. I emphasize that the > principal purpose of this new box would be exploratory, or for > development, not production. > > I?ve built several past platforms, both Windows and Linux based, using > server boards manufactured by Supermicro and have had generally good > experience and service (the factory is only 15 miles from where I > live). Currently, I use two machines with dual Xeon processors (single > core, 32 bits), one with 1 GHz cpus, 1 GB RAM, the other 2.4 GHz and > 3.25 GB RAM, both with bus speeds of 133 MHz, the newest 3 years old. > However, occasionally I run PWscf exercises on my little Thinkpad X41 > tablet (single processor, 1.5 GHz, 1.5 GB), and the scf computation will > run 3-5 times faster than on the other machines! I suspect this rather > surprising result is because the Thinkpad has a 400 MHz bus clock speed. > > One option I?m considering is using a ?gaming? or server class > motherboard with dual 2.33 GHz quad-core 64-bit processors, a 1333 MHz > FSB, and 16 GB RAM. Having said this, I?m not sure PWscf (and the > Fortran compilers available) can handle all this parallelism efficiently > on a single motherboard. I?ve noticed when running pw.x, the CPU > activity ?flips? between processors every several seconds, instead of > sharing each at 90-100% full time. > > On the other hand, one could consider building a small MPI-connected > cluster for about the same amount of money. > > When IBM announced a couple of years ago the incredible performance > details about the Cell processor that would go into Playstation 3, I > thought, ?Wow, maybe the future of computational physics rests with > gamers.? I?m sure most of you know this is actually beginning to > happen, spurred on by the fact that the PS3 is ?open architecture? and > can run a Linux distro. Moreover, there apparently are ?open software? > numerical analysis tools available from IBM. At least four US > universities are experimenting with off-the-shelf PS3 clusters, perhaps > one of the more interesting is at UMass, > _http://gravity.phy.umassd.edu/ps3.html_. In the last week or so, Sony > lowered the entry level price of the PS3 to 400 USD. So, a cluster of > four with a cheap switch could be purchased for about the same price at > the single motherboard configuration I mentioned above. > > My teenager, a gamer, tells me the PS3 has problems. He says it?s > unreliable and overheats and only has 256 MB RAM on board (he owns a > Wii, which outsells the PS3 in the US by a factor of three). > > Has anybody tried porting PWscf to a PS3? > > Any and all advice is welcome. > -- - Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu
[Pw_forum] Problem with nbnd in clusters
Hi Cyrille, your tm dimer will likely require spin polarization, and will have several scf solutions that correspond to different S_z (that you can fix in several ways, either specifying the number of spin up and spin down, or using a fermi_up and fermi_down energy, to specify n_up - n_down but still treating the system with a smearing and fractional occupations). Even for a given S_z, there will be different scf solutions that have different spatial symmetry. So, in your case I would definitely switch spin-polarization, and maybe play around with small and large smearings, to see what happens, and look at the states, and the occupations. Anyhow, if you get to a scf solution, that is a "good" solution for the GGA problem, although might not be very physical. Also, have a look at Kulik et al PRL 2006. nicola Cyrille Barreteau wrote: > Dear pwscf_ers > > I am doing a very simple calculation on a transition metal > dimer and I have a encoutered a strange problem (I love > strange problems:-) > > I have done a first calculation with the default value > of nbnd, ie nbnd=nelec/2*20% (=12 in my case) > > But since it is known that in clusters it is often good > to increase nbnd I have performed another calculation with > larger nbnd (=20). The result is quite different from > the one at nbnd=12. > > I have then increased again nbnd up to 25 and then I recover > the result obtained for nbnd=12 > > More problematic is the local density of states (I know > it is not really a dos but a bunch of dirac peaks). > In a dimer with the z axis along the direction connecting > the two atoms, the xy and x^2-y^2 dos should be degenerate > (if the supercell box is large enough). > In fact the xy and x^2-y^2 are degenerate if nbnd=12 or 25 > but there is a rather large splitting if nbnd=20. > > I am quite sure the result for nbnd=20 is not correct but > what is the origin of this problem? > > Maybe I could try to use another diagonalization scheme.. > >thanks for reading my strange problems > > cyrille > > -- - Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu
[Pw_forum] Problem with nbnd in clusters
Dear Cyrille, Nicola Marzari's is a good advice for these systems. Another thing that you could check is starting from random wave functions (instead of atomic). I am not expert of diagonalization algorithms but maybe what happens is that davidson with 20 bands (who knows why...) gets trapped in some local minimum just for numerical reasons. So maybe some noise could help. How do the energies compare? Regards, Matteo Nicola Marzari wrote: > Hi Cyrille, > > your tm dimer will likely require spin polarization, and > will have several scf solutions that correspond to different > S_z (that you can fix in several ways, either specifying the > number of spin up and spin down, or using a fermi_up and > fermi_down energy, to specify n_up - n_down but still treating > the system with a smearing and fractional occupations). > > Even for a given S_z, there will be different scf solutions > that have different spatial symmetry. > > So, in your case I would definitely switch spin-polarization, > and maybe play around with small and large smearings, to see what > happens, and look at the states, and the occupations. Anyhow, if > you get to a scf solution, that is a "good" solution for the GGA > problem, although might not be very physical. > > Also, have a look at Kulik et al PRL 2006. > > nicola > > > Cyrille Barreteau wrote: > >> Dear pwscf_ers >> >> I am doing a very simple calculation on a transition metal >> dimer and I have a encoutered a strange problem (I love >> strange problems:-) >> >> I have done a first calculation with the default value >> of nbnd, ie nbnd=nelec/2*20% (=12 in my case) >> >> But since it is known that in clusters it is often good >> to increase nbnd I have performed another calculation with >> larger nbnd (=20). The result is quite different from >> the one at nbnd=12. >> >> I have then increased again nbnd up to 25 and then I recover >> the result obtained for nbnd=12 >> >> More problematic is the local density of states (I know >> it is not really a dos but a bunch of dirac peaks). >> In a dimer with the z axis along the direction connecting >> the two atoms, the xy and x^2-y^2 dos should be degenerate >> (if the supercell box is large enough). >> In fact the xy and x^2-y^2 are degenerate if nbnd=12 or 25 >> but there is a rather large splitting if nbnd=20. >> >> I am quite sure the result for nbnd=20 is not correct but >> what is the origin of this problem? >> >> Maybe I could try to use another diagonalization scheme.. >> >>thanks for reading my strange problems >> >> cyrille >> >> >> > > > -- next part -- A non-text attachment was scrubbed... Name: matteo.vcf Type: text/x-vcard Size: 294 bytes Desc: not available Url : http://www.democritos.it/pipermail/pw_forum/attachments/20071120/29749b8b/attachment.vcf
[Pw_forum] stoping the pw.x
On Nov 19, 2007, at 15:07 , Rudra Banerjee wrote: > is it possible to stop a running pw job in between > and restart it with "restart" mode? yes, but it doesn't always work. You can either stop after a predefined time (see input variable "max_seconds") or when you deem it appropriate (the best way is to create a file "prefix".EXIT in the temporary directory where the other "prefix".* files are). How the code will restart depends on how much information is stored to file. Paolo --- Paolo Giannozzi, Dept of Physics, University of Udine via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation
On Nov 20, 2007, at 4:07 , Paul M. Grant wrote: > Currently, I use two machines with dual Xeon processors dual Xeon? very bad choice. They have a single memory bus, so if you run two processes on the same motherboard, they spend most of the time waiting for access to memory. P. --- Paolo Giannozzi, Dept of Physics, University of Udine via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] about the influence of lo-to split on the vibrational model
On Nov 20, 2007, at 4:16 , xu yuehua wrote: > and there is not the vector at all!! > [...] > and what is wrong ? i do not know > please give me a detailed advise your attitude is wrong. Had you followed my previous advice (not "advise") you would have found this: ! filout character output file containing frequencies and modes !(default: filout='dynmat.out') ! filmol character as above, in a format suitable for 'molden' !(default: filmol='dynmat.mold') ! filxsf character as above, in axsf format suitable for xcrysden !(default: filmol='dynmat.axsf') --- Paolo Giannozzi, Dept of Physics, University of Udine via delle Scienze 208, 33100 Udine, Italy Phone +39-0432-558216, fax +39-0432-558222
[Pw_forum] The calculated results using the pp file of Cs from Mr. Paolo.
Dear Paolo and all users, I got a good areeement between the theory and experiment by using the Cs pp file that you sent to me. The simulated results even reproduce the experiment data under very high pressure. But I just tested the one with PBE and 9 valence electrons. Thanks a lot! Best regards, Yan Li - ?? -- next part -- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20071120/cd6c7d0e/attachment-0001.htm
[Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation
Thanks for the feedback. The two dual Xeon boards I have were originally chosen to host my website. I long suspected that there was "thrashing" between the two processors when performing numerically intensive calculations. I assume the same would occur within the cores of a single processor as well? However, my understanding was this approach was supposed to better implement matrix manipulation for gaming applications. If so, one would assume improved performance for the fft operations at the "core" of pw.x. Paul M. Grant, PhD Principal, W2AGZ Technologies Visiting Scholar, Applied Physics, Stanford University EPRI Science Fellow (Retired) IBM Research Staff Member Emeritus w2agz at pacbell.net http://www.w2agz.com ? ? -Original Message- From: pw_forum-bounces at pwscf.org [mailto:pw_forum-boun...@pwscf.org] On Behalf Of Alexander Shaposhnikov Sent: Tuesday, November 20, 2007 1:02 AM To: PWSCF Forum Subject: Re: [Pw_forum] Seeking Advice on Small Hardware Platforms for PWscf Implementation Actually newer Woodcrest/Clovertown Xeon platform have dual independent FSBs, but this does not help much. Memory bandwidth available on each FSB is very limited - between 1/2 and 1/3 of what is available to modern desktop CPUs. There is no good "server" computational platform today, unfortunately . AMD's CPUs have weak FPU and Intel's have weak memory subsystem. Intel- based desktop single-socket platform is the best choice for now. On Tuesday 20 November 2007 14:33, Paolo Giannozzi wrote: > On Nov 20, 2007, at 4:07 , Paul M. Grant wrote: > > Currently, I use two machines with dual Xeon processors > > dual Xeon? very bad choice. They have a single memory > bus, so if you run two processes on the same motherboard, > they spend most of the time waiting for access to memory. > > P. > --- > Paolo Giannozzi, Dept of Physics, University of Udine > via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > ___ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum ___ Pw_forum mailing list Pw_forum at pwscf.org http://www.democritos.it/mailman/listinfo/pw_forum
