From cbarreteau at cea.fr Mon May 3 17:26:59 2004 From: cbarreteau at cea.fr (Cyrille Barreteau) Date: Mon, 03 May 2004 17:26:59 +0200 Subject: [Pw_forum] Projection of plane wave functions into atomic orbitals References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> Message-ID: <409664C3.903@cea.fr> Dear pwscf users, I have a question concerning the problem of projection of PW wave function on atomic orbitals. example 8 shows the case of bulk Ni with LDA pseudo pot NiUS.RRKJ3.UPF which uses two wave functions s and d (see below) 2 6 Number of Wavefunctions, Number of Projectors Wavefunctions nl l occ 4S 0 1.00 3D 2 9.00 The result of the projection with projwfc.x gives the following decomposition: Atom # 1: total charge = 9.8849, s, p, d, f = 0.9347 0.0000 8.9501 Spilling Parameter: 0.0115 My question is the following: where have the p electrons gone? It seems that eventhough the spilling parameter is not fantastic, it is not that bad and from the charge distribution it seems that the p electrons have in a way been projected on s orbitals! How is it possible? thanks for your help. cyrille ============================================================ | Cyrille Barreteau | phone:+33 (0)1 69 08 29 51 | | CEA Saclay | | | DSM/DRECAM/SPCSI | fax : +33 (0)1 69 08 84 46 | | Batiment 462 | | | 91191 Gif sur Yvette Cedex | email: cbarreteau at cea.fr | | FRANCE | | | ~~~~~~~~~~~~~~~~~~~~~~~~ | | web : http://www-drecam.cea.fr/spcsi/groupe1.htm | ============================================================ From francesco.antoniella at aquila.infn.it Mon May 3 17:46:53 2004 From: francesco.antoniella at aquila.infn.it (Francesco Antoniella) Date: 03 May 2004 17:46:53 +0200 Subject: [Pw_forum] Projection of plane wave functions into atomic orbitals In-Reply-To: <409664C3.903@cea.fr> References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> <409664C3.903@cea.fr> Message-ID: <1083599216.1952.11.camel@altair> Il lun, 2004-05-03 alle 17:26, Cyrille Barreteau ha scritto: > > Dear pwscf users, > > I have a question concerning the problem of projection of PW wave function > on atomic orbitals. > example 8 shows the case of bulk Ni with LDA pseudo pot NiUS.RRKJ3.UPF > which uses two wave functions s and d (see below In your pseudo the p channel is the local channel, i.e. you have no projectors for the p because it has no nonlocal part at all for the p pseudo. This mean that the projecting proceudure cannot find the p-like charge. It is better to use a directly separated pseudo as the vanderbilt pseudo to overcome this problem. Francesco > > 2 6 Number of Wavefunctions, Number of Projectors > Wavefunctions nl l occ > 4S 0 1.00 > 3D 2 9.00 > > The result of the projection with projwfc.x gives the following decomposition: > > Atom # 1: total charge = 9.8849, s, p, d, f = 0.9347 0.0000 8.9501 > Spilling Parameter: 0.0115 > > > > My question is the following: where have the p electrons gone? > > It seems that eventhough the spilling parameter is not fantastic, it is > not that bad and from the charge distribution it seems that the p electrons > have in a way been projected on s orbitals! How is it possible? > > thanks for your help. > > cyrille > > > > > > ============================================================ > | Cyrille Barreteau | phone:+33 (0)1 69 08 29 51 | > | CEA Saclay | | > | DSM/DRECAM/SPCSI | fax : +33 (0)1 69 08 84 46 | > | Batiment 462 | | > | 91191 Gif sur Yvette Cedex | email: cbarreteau at cea.fr | > | FRANCE | | > | ~~~~~~~~~~~~~~~~~~~~~~~~ | > | web : http://www-drecam.cea.fr/spcsi/groupe1.htm | > ============================================================ > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > From degironc at sissa.it Mon May 3 17:39:29 2004 From: degironc at sissa.it (Stefano de Gironcoli) Date: Mon, 03 May 2004 17:39:29 +0200 Subject: [Pw_forum] Projection of plane wave functions into atomic orbitals References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> <409664C3.903@cea.fr> Message-ID: <409667B1.9020702@sissa.it> > I have a question concerning the problem of projection of PW wave > function > on atomic orbitals. > example 8 shows the case of bulk Ni with LDA pseudo pot NiUS.RRKJ3.UPF > which uses two wave functions s and d (see below) > > 2 6 Number of Wavefunctions, Number of Projectors > Wavefunctions nl l occ > 4S 0 1.00 > 3D 2 9.00 > > The result of the projection with projwfc.x gives the following > decomposition: > > Atom # 1: total charge = 9.8849, s, p, d, f = 0.9347 0.0000 > 8.9501 > Spilling Parameter: 0.0115 > > My question is the following: where have the p electrons gone? The code projects the calculated wavefunctions on the atomic wavefunctions that are present in the pseudopotential file. In this case the 4s and 3d. There is no 4p wavefunction in the pseudopontential and there will not be any 4p projected density of states in the projected DOS. The fact that the spilling parameter is not that bad indicates that these wavefuntions are not really needed in order to describe the occupied manyfold. If some important atomic wavefunction were missing the spilling parameter would be much worse. It is possible that if the 4p wavefunctions were present in the pseudopotential somewhat different projected densities of states would result... (together with a smaller spilling parameter). This is because the 4p states of an atom are NOT ORTHOGONAL to the 4s/3d states of neighboring atoms so that part of the DOS that would be interpreted as 4p when this state is present are interpreted as 4s of 3d if it is absent... This would imply that some redundancy (overcompleteness) is present in the atomic-wavefunction basis-set and the choice operated by the code (orthogonalizing the atomic wavefunctions before making the projection ro get the so called Lowdin projections) is only one of the infinite (all arbitrary) ways this redundancy is removed. Stefano de Gironcoli From konstantin_kudin at yahoo.com Tue May 4 03:48:30 2004 From: konstantin_kudin at yahoo.com (Konstantin Kudin) Date: Mon, 3 May 2004 18:48:30 -0700 (PDT) Subject: [Pw_forum] NEB examples? In-Reply-To: <1083599216.1952.11.camel@altair> Message-ID: <20040504014830.14736.qmail@web21202.mail.yahoo.com> Can anyone e-mail to me an input for some simple system to run with the NEB method (nudged elastic band) ? I tried to find one among the PWSCF tests, but did not see any ... Thanks! Konstantin __________________________________ Do you Yahoo!? Win a $20,000 Career Makeover at Yahoo! HotJobs http://hotjobs.sweepstakes.yahoo.com/careermakeover From cbarreteau at cea.fr Tue May 4 08:28:46 2004 From: cbarreteau at cea.fr (Cyrille Barreteau) Date: Tue, 04 May 2004 08:28:46 +0200 Subject: [Pw_forum] Projection of plane wave functions into atomic orbitals References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> <409664C3.903@cea.fr> <409667B1.9020702@sissa.it> Message-ID: <4097381E.1050407@cea.fr> Dear Francesco and Stefano, Thank you very much for your clear answer, it is extremely helpful. By the way I wanted to congratulate all the pwscf developpers for the impressive work you have done. It is really a very good code. Cyrille -- ============================================================ | Cyrille Barreteau | phone:+33 (0)1 69 08 29 51 | | CEA Saclay | | | DSM/DRECAM/SPCSI | fax : +33 (0)1 69 08 84 46 | | Batiment 462 | | | 91191 Gif sur Yvette Cedex | email: cbarreteau at cea.fr | | FRANCE | | | ~~~~~~~~~~~~~~~~~~~~~~~~ | | web : http://www-drecam.cea.fr/spcsi/groupe1.htm | ============================================================ >> I have a question concerning the problem of projection of PW wave >> function >> on atomic orbitals. >> example 8 shows the case of bulk Ni with LDA pseudo pot NiUS.RRKJ3.UPF >> which uses two wave functions s and d (see below) >> >> 2 6 Number of Wavefunctions, Number of Projectors >> Wavefunctions nl l occ >> 4S 0 1.00 >> 3D 2 9.00 >> >> The result of the projection with projwfc.x gives the following >> decomposition: >> >> Atom # 1: total charge = 9.8849, s, p, d, f = 0.9347 0.0000 >> 8.9501 >> Spilling Parameter: 0.0115 >> >> My question is the following: where have the p electrons gone? In your pseudo the p channel is the local channel, i.e. you have no projectors for the p because it has no nonlocal part at all for the p pseudo. This mean that the projecting proceudure cannot find the p-like charge. It is better to use a directly separated pseudo as the vanderbilt pseudo to overcome this problem. Francesco Stefano de Gironcoli wrote: > >> I have a question concerning the problem of projection of PW wave >> function >> on atomic orbitals. >> example 8 shows the case of bulk Ni with LDA pseudo pot NiUS.RRKJ3.UPF >> which uses two wave functions s and d (see below) >> >> 2 6 Number of Wavefunctions, Number of Projectors >> Wavefunctions nl l occ >> 4S 0 1.00 >> 3D 2 9.00 >> >> The result of the projection with projwfc.x gives the following >> decomposition: >> >> Atom # 1: total charge = 9.8849, s, p, d, f = 0.9347 0.0000 >> 8.9501 >> Spilling Parameter: 0.0115 >> >> My question is the following: where have the p electrons gone? > > > > The code projects the calculated wavefunctions on the atomic > wavefunctions that are present in the pseudopotential file. In this case > the 4s and 3d. > There is no 4p wavefunction in the pseudopontential and there will not > be any 4p projected density of states in the projected DOS. > > The fact that the spilling parameter is not that bad indicates that > these wavefuntions are not really needed in order to describe the > occupied manyfold. > If some important atomic wavefunction were missing the spilling > parameter would be much worse. > > It is possible that if the 4p wavefunctions were present in the > pseudopotential somewhat different projected densities of states would > result... (together with a smaller spilling parameter). > This is because the 4p states of an atom are NOT ORTHOGONAL to the 4s/3d > states of neighboring atoms so that part of the DOS that would be > interpreted as 4p when this state is present are interpreted as 4s of 3d > if it is absent... > This would imply that some redundancy (overcompleteness) is present in > the atomic-wavefunction basis-set and the choice operated by the code > (orthogonalizing the atomic wavefunctions before making the projection > ro get the so called Lowdin projections) is only one of the infinite > (all arbitrary) ways this redundancy is removed. > > Stefano de Gironcoli > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum From dchrobak at us.edu.pl Tue May 4 11:01:26 2004 From: dchrobak at us.edu.pl (Dariusz Chrobak) Date: Tue, 04 May 2004 11:01:26 +0200 Subject: [Pw_forum] cpmd2upf.f90 file Message-ID: <40975BE6.8050407@us.edu.pl> Dear PWSCF users and developers, I have a problem with pseudopotential transformation from cpmd format to upf one. The following information appears during transformation: ... some &FIELD card missing ... but after small modification of the cpmd2upf.f90 file : (from line number 154) else if (matches ("&NLCC", trim(line)) ) then found=found+1 <---- added line nlcc_ = .true. read (iunps, '(a)') line if (.not. matches ("NUMERIC", trim(line)) ) & call errore('read_cpmd','core-correction not allowed',1) read(iunps, *) mesh_ allocate (rho_atc_(mesh_)) read(iunps, * ) (r_(i), rho_atc_(i), i=1,mesh_) compiled program seems to work well. What do you think about this change? Best regards Dariusz Chrobak From giannozz at nest.sns.it Tue May 4 11:58:50 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Tue, 4 May 2004 11:58:50 +0200 Subject: [Pw_forum] cpmd2upf.f90 file In-Reply-To: <200405041153.23815.giannozz@nest.sns.it> References: <40975BE6.8050407@us.edu.pl> <200405041153.23815.giannozz@nest.sns.it> Message-ID: <200405041158.50004.giannozz@nest.sns.it> On Tuesday 04 May 2004 11:01, Dariusz Chrobak wrote: > [...] after small modification of the cpmd2upf.f90 file : > > (from line number 154) > else if (matches ("&NLCC", trim(line)) ) then > found=found+1 <---- added line > nlcc_ = .true. > read (iunps, '(a)') line > if (.not. matches ("NUMERIC", trim(line)) ) & > call errore('read_cpmd','core-correction not allowed',1) > read(iunps, *) mesh_ > allocate (rho_atc_(mesh_)) > read(iunps, * ) (r_(i), rho_atc_(i), i=1,mesh_) > > compiled program seems to work well. > > What do you think about this change? you are right: the code performs a check on the number of fields found in the cpmd file, but the core-correction field wasn't counted. It will be fixed soon. Thank you Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From marzari at MIT.EDU Tue May 4 13:57:56 2004 From: marzari at MIT.EDU (Nicola Marzari) Date: Tue, 04 May 2004 07:57:56 -0400 Subject: [Pw_forum] ifc/linux allocation of more than 2 GB dynamical memory In-Reply-To: <4097381E.1050407@cea.fr> References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> <409664C3.903@cea.fr> <409667B1.9020702@sissa.it> <4097381E.1050407@cea.fr> Message-ID: <40978544.4040703@mit.edu> Dear All, we have recently realized that there is an issue running large jobs on Linux machines, and I was wondering if anyone had some suggestions. This is the problem: if a fortran code running under Linux (Red Hat 9) and using ifc (7.1) and mkl (6.0) tries to allocate a number of arrays that total over 2 GB, it ends with a runtime error. This happens no matter the amount of dynamical memory physically available (we tested this on a machine having ~4GB RAM), and happens exactly at the point in which the array that is being allocated overshoots 2GB . The Intel webpage has a comment somewhere mentioning this limitation, suggesting it is related to where the kernel allocates space for the linked libraries, but we have had no success trying their suggestions (e.g. compiling -static). In practice, this limitation is even more serious, since a running code (e.g. CP) that allocates ~< 2GB will only use, in runtime, an average value that is smaller (~1.25GB). Any helpful suggestions would be very welcome, nicola --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 617.2586534 marzari at mit.edu http://nnn.mit.edu From sbraccia at sissa.it Tue May 4 14:16:28 2004 From: sbraccia at sissa.it (sbraccia carlo) Date: Tue, 4 May 2004 14:16:28 +0200 Subject: [Pw_forum] NEB examples? In-Reply-To: <20040504014830.14736.qmail@web21202.mail.yahoo.com> References: <20040504014830.14736.qmail@web21202.mail.yahoo.com> Message-ID: <200405041416.29158.sbraccia@sissa.it> On Tuesday 04 May 2004 03:48 am, Konstantin Kudin wrote: > Can anyone e-mail to me an input for some simple system to run with the > NEB method (nudged elastic band) ? I tried to find one among the PWSCF > tests, but did not see any ... > > Thanks! > Konstantin Dear Konstantin, since version 2.0 in the pw_examples directory you can find a NEBexample sub-directory containing what you are looking for. carlo sbraccia From giovanni.cangiani at epfl.ch Tue May 4 14:27:57 2004 From: giovanni.cangiani at epfl.ch (Giovanni Cangiani) Date: Tue, 04 May 2004 14:27:57 +0200 Subject: [Pw_forum] ifc/linux allocation of more than 2 GB dynamical memory In-Reply-To: <40978544.4040703@mit.edu> References: <408FC85C.3090701@cea.fr> <200404281721.55376.giannozz@nest.sns.it> <408FCD46.1040707@cea.fr> <40909B9F.3010003@cea.fr> <1083238803.1517.12.camel@altair> <409664C3.903@cea.fr> <409667B1.9020702@sissa.it> <4097381E.1050407@cea.fr> <40978544.4040703@mit.edu> Message-ID: <1083673674.12432.76.camel@algopc151.epfl.ch> Dear Nicola, I hope I am wrong but I think the memory is limited to 2Gb due to the 32bit addressing (don't ask me why 2Gb and not 8Gb=2^33) . AFAIK there is no way to solve it unless you buy 64 bit CPU like intel itanium or AMD opteron. I am presently testing the program on an opteron based cluster with Portland compiler (there is no i386_64 version of the intel compiler) but I still have quite a lot troubles. giovanni as far as I know the limit of 2Gb is due to the 32bit addressing. On Tue, 2004-05-04 at 13:57, Nicola Marzari wrote: > Dear All, > > we have recently realized that there is an issue running large jobs > on Linux machines, and I was wondering if anyone had some suggestions. > > This is the problem: if a fortran code running under Linux > (Red Hat 9) and using ifc (7.1) and mkl (6.0) tries to allocate > a number of arrays that total over 2 GB, it ends with a runtime error. > > This happens no matter the amount of dynamical memory physically > available (we tested this on a machine having ~4GB RAM), and happens > exactly at the point in which the array that is being allocated > overshoots 2GB . > > The Intel webpage has a comment somewhere mentioning this limitation, > suggesting it is related to where the kernel allocates space for the > linked libraries, but we have had no success trying their suggestions > (e.g. compiling -static). > > In practice, this limitation is even more serious, since a running > code (e.g. CP) that allocates ~< 2GB will only use, in runtime, an > average value that is smaller (~1.25GB). > > Any helpful suggestions would be very welcome, > > > nicola > > --------------------------------------------------------------------- > Prof Nicola Marzari Department of Materials Science and Engineering > 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA > tel 617.4522758 fax 617.2586534 marzari at mit.edu http://nnn.mit.edu > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum From giannozz at nest.sns.it Tue May 4 15:44:21 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Tue, 4 May 2004 15:44:21 +0200 Subject: [Pw_forum] ifc/linux allocation of more than 2 GB dynamical memory In-Reply-To: <40978544.4040703@mit.edu> References: <408FC85C.3090701@cea.fr> <4097381E.1050407@cea.fr> <40978544.4040703@mit.edu> Message-ID: <200405041544.21301.giannozz@nest.sns.it> On Tuesday 04 May 2004 13:57, Nicola Marzari wrote: > we have recently realized that there is an issue running large jobs > on Linux machines The issue of maximum file and RAM size on Linux/Intel 32-bit processors has surfaced here more than once. The code doesn't impose any explicit limit to the size of files or allocated arrays or total RAM [*]. The compiler, libraries, operating system, kernel, or a combination of the above, impose limits. For the specific case of the Intel compiler, I think that the only hope to find out what is going on is here: http://softwareforums.intel.com/ids/search?board.id=11&q=2Gb+limit Paolo [*] there is a limit coming from storage of pointers used by FFTW into 32-bit integers; one could easily use 64-bit integers instead. -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From konstantin_kudin at yahoo.com Tue May 4 16:44:35 2004 From: konstantin_kudin at yahoo.com (Konstantin Kudin) Date: Tue, 4 May 2004 07:44:35 -0700 (PDT) Subject: [Pw_forum] ifc/linux allocation of more than 2 GB dynamical memory In-Reply-To: <40978544.4040703@mit.edu> Message-ID: <20040504144435.31933.qmail@web21201.mail.yahoo.com> Dear Nicola Please read the thread from February "Error encountered while attempting to allocate a data object." In this post Serguei Patchkovskii explained how to maximize RAM under linux: http://www.democritos.it/pipermail/pw_forum/2004-January/000825.html In short, it appears that static linking of everything (mkl included!!!) could give more RAM. Since mkl 6.0 does not link this way, you need to use mkl 5.2 Konstantin --- Nicola Marzari wrote: > > > Dear All, > > we have recently realized that there is an issue running large jobs > on Linux machines, and I was wondering if anyone had some suggestions. > > This is the problem: if a fortran code running under Linux > (Red Hat 9) and using ifc (7.1) and mkl (6.0) tries to allocate > a number of arrays that total over 2 GB, it ends with a runtime error. > > This happens no matter the amount of dynamical memory physically > available (we tested this on a machine having ~4GB RAM), and happens > exactly at the point in which the array that is being allocated > overshoots 2GB . > > The Intel webpage has a comment somewhere mentioning this limitation, > suggesting it is related to where the kernel allocates space for the > linked libraries, but we have had no success trying their suggestions > (e.g. compiling -static). > > In practice, this limitation is even more serious, since a running > code (e.g. CP) that allocates ~< 2GB will only use, in runtime, an > average value that is smaller (~1.25GB). > > Any helpful suggestions would be very welcome, > > > nicola > > --------------------------------------------------------------------- > Prof Nicola Marzari Department of Materials Science and Engineering > 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA > tel 617.4522758 fax 617.2586534 marzari at mit.edu http://nnn.mit.edu > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum __________________________________ Do you Yahoo!? Win a $20,000 Career Makeover at Yahoo! HotJobs http://hotjobs.sweepstakes.yahoo.com/careermakeover From FB153746 at ATIL.CEA.FR Tue May 4 16:09:17 2004 From: FB153746 at ATIL.CEA.FR (=?iso-8859-1?Q?BOUYER_Fr=E9d=E9ric_153746?=) Date: Tue, 4 May 2004 16:09:17 +0200 Subject: [Pw_forum] Compilation error Message-ID: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BC@atil.valrho.cea.fr> Hello Everybody, I am a very new user of PWscf and would try to compile on our NEC machine. After making some changes in the make.sys file, the compilation starts as expected, but after some minutes, the compilation stops with the following errors (see below). Could anyone help me to find what is wrong ? Many thanks in advance. Best regards, Fr?d?ric Bouyer ___________________________________________________________________________ sxf90 -o q2r.x q2r.o rigid.o ../PW/pwcom.o ../PW/cft_3.o ../PW/checksym.o ../PW/coset.o ../PW/cryst_to_car.o ../PW/cubicsym.o ../PW/dsum.o ../PW/eqvect.o ../PW/error.o ../PW/error_handler.o ../PW/hexsym.o ../PW/invmat.o ../PW/irrek.o ../PW/kpoint_grid.o ../PW/mode_group.o ../PW/multable.o ../PW/random.o ../PW/sgam_at.o ../PW/sgam_ph.o ../PW/sgama.o ../PW/smallg_q.o ../PW/supercell.o ../PW/trnvecc.o ../PW/wsweight.o ../Modules/basic_algebra_routines.o ../Modules/constants.o ../Modules/io_global.o ../Modules/io_files.o ../Modules/fft_scalar.o ../Modules/formats.o ../Modules/kind.o ../Modules/mp.o ../Modules/mp_global.o ../Modules/parameters.o ../Modules/parser.o ../Modules/splinelib.o -float0 -P stack -sx6 -g -Wf" -init stack=zero heap=zero" /home/bouyer/PWSCF/flib/ptools.a /home/bouyer/PWSCF/flib/flib.a /home/bouyer/PWSCF/clib/clib.a -p -Wl" -f zero " -llapack -lblas -L/home/bouyer/JMFFT-7.2/ -ljmfft undefined first referenced symbol in file reciprocal_vectors_ ../PW/pwcom.o control_flags_ ../PW/pwcom.o fft_types_ ../PW/pwcom.o parallel_include_ ../Modules/mp.o shmem_include_ ../Modules/mp_global.o sxld fatal: symbol referencing errors. no output written to file q2r.x. ____________________________________________________________________________ _ From roma at srmp19.saclay.cea.fr Wed May 5 09:07:37 2004 From: roma at srmp19.saclay.cea.fr (Guido Roma) Date: Wed, 05 May 2004 09:07:37 +0200 Subject: [Pw_forum] Compilation error In-Reply-To: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BC@atil.valrho.cea.fr> References: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BC@atil.valrho.cea.fr> Message-ID: <1083740857.7320.8521.camel@srmp19.saclay.cea.fr> On Tue, 2004-05-04 at 16:09, BOUYER Fr?d?ric 153746 wrote: > Hello Everybody, > > I am a very new user of PWscf and would try to compile on our NEC machine. > > After making some changes in the make.sys file, the compilation starts as > expected, but after some minutes, the compilation stops with the following > errors (see below). Hello, As far as I know there's one thing that you should do in order to crosscompile the last version of PWSCF (2.0.2) for NEC: change CC to CCLOCAL in install/Makefile then, after ./configure.old sxcross, it should work out of the box. But I see that you're using jmfft, this means that you're probably using a previous version (1.3.0?). In that case you can probably solve your problems by adding to pwtools/Makefile the following lines: ../Modules/recvec.o \ ../Modules/control_flags.o \ ../Modules/parallel_include.o \ ../Modules/shmem_include.o \ ../Modules/fft_types.o A remark: jmfft is slower than NEC ASL libraries for FFT; but if you use ASL you have to be careful, because some versions of ASL contain a routine (zfc3cl) which is bugged; in that case PWSCF can give wrong results. A workaround is to substitute (in PW/cft_3.f90) this routine by My_zfc3cl here below. Best regards, Guido Roma _____________________________________________________________________ subroutine My_zfc3cl(f1,nr1,nr2,nr3,nrz1,nrz2,nrz3,ierr) use kinds, only : DP complex(kind=DP), dimension(nrz1,nrz2,nrz3) :: f1 f1(1:nrz1,1:nrz2,nr3+1:nrz3) = cmplx(0.0,0.0) f1(1:nrz1,nr2+1:nrz2,1:nrz3) = cmplx(0.0,0.0) f1(nr1+1:nrz1,1:nrz2,1:nrz3) = cmplx(0.0,0.0) ierr=0 end subroutine My_zfc3cl ______________________________________________________________________ > > Could anyone help me to find what is wrong ? > > Many thanks in advance. > > Best regards, > > Fr?d?ric Bouyer > ___________________________________________________________________________ > > sxf90 -o q2r.x q2r.o rigid.o ../PW/pwcom.o ../PW/cft_3.o .../PW/checksym.o > .../PW/coset.o ../PW/cryst_to_car.o ../PW/cubicsym.o ../PW/dsum.o > .../PW/eqvect.o ../PW/error.o ../PW/error_handler.o ../PW/hexsym.o > .../PW/invmat.o ../PW/irrek.o ../PW/kpoint_grid.o ../PW/mode_group.o > .../PW/multable.o ../PW/random.o ../PW/sgam_at.o ../PW/sgam_ph.o > .../PW/sgama.o ../PW/smallg_q.o ../PW/supercell.o ../PW/trnvecc.o > .../PW/wsweight.o ../Modules/basic_algebra_routines.o .../Modules/constants.o > .../Modules/io_global.o ../Modules/io_files.o ../Modules/fft_scalar.o > .../Modules/formats.o ../Modules/kind.o ../Modules/mp.o > .../Modules/mp_global.o ../Modules/parameters.o ../Modules/parser.o > .../Modules/splinelib.o -float0 -P stack -sx6 -g -Wf" -init stack=zero > heap=zero" /home/bouyer/PWSCF/flib/ptools.a /home/bouyer/PWSCF/flib/flib.a > /home/bouyer/PWSCF/clib/clib.a -p -Wl" -f zero " -llapack -lblas > -L/home/bouyer/JMFFT-7.2/ -ljmfft > undefined first referenced > symbol in file > reciprocal_vectors_ ../PW/pwcom.o > control_flags_ ../PW/pwcom.o > fft_types_ ../PW/pwcom.o > parallel_include_ ../Modules/mp.o > shmem_include_ ../Modules/mp_global.o > sxld fatal: symbol referencing errors. no output written to file q2r.x. > > ____________________________________________________________________________ > _ > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum From FB153746 at ATIL.CEA.FR Wed May 5 10:01:59 2004 From: FB153746 at ATIL.CEA.FR (=?iso-8859-1?Q?BOUYER_Fr=E9d=E9ric_153746?=) Date: Wed, 5 May 2004 10:01:59 +0200 Subject: [Pw_forum] Compilation error Message-ID: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BE@atil.valrho.cea.fr> Hello Guido, Thanks to your answer. I already made the changes you propose (change $CC to $CCLOCAL or use the cc compiler instead of the sxcc ... at the beginning ). I have used one of the three modes for the compilation (ASL, ... or FFT of jmfft libraries), but I still have the same problem during the compilation. It stops always at the same point with the same errors. By the way, many thanks for the changes in the zfc3cl routine. Fr?d?ric -----Message d'origine----- De?: Guido Roma [mailto:roma at srmp19.saclay.cea.fr] Envoy??: mercredi 5 mai 2004 09:08 ??: pw_forum at pwscf.org Objet?: Re: [Pw_forum] Compilation error On Tue, 2004-05-04 at 16:09, BOUYER Fr?d?ric 153746 wrote: > Hello Everybody, > > I am a very new user of PWscf and would try to compile on our NEC machine. > > After making some changes in the make.sys file, the compilation starts as > expected, but after some minutes, the compilation stops with the following > errors (see below). Hello, As far as I know there's one thing that you should do in order to crosscompile the last version of PWSCF (2.0.2) for NEC: change CC to CCLOCAL in install/Makefile then, after ./configure.old sxcross, it should work out of the box. But I see that you're using jmfft, this means that you're probably using a previous version (1.3.0?). In that case you can probably solve your problems by adding to pwtools/Makefile the following lines: .../Modules/recvec.o \ .../Modules/control_flags.o \ .../Modules/parallel_include.o \ .../Modules/shmem_include.o \ .../Modules/fft_types.o A remark: jmfft is slower than NEC ASL libraries for FFT; but if you use ASL you have to be careful, because some versions of ASL contain a routine (zfc3cl) which is bugged; in that case PWSCF can give wrong results. A workaround is to substitute (in PW/cft_3.f90) this routine by My_zfc3cl here below. Best regards, Guido Roma _____________________________________________________________________ subroutine My_zfc3cl(f1,nr1,nr2,nr3,nrz1,nrz2,nrz3,ierr) use kinds, only : DP complex(kind=DP), dimension(nrz1,nrz2,nrz3) :: f1 f1(1:nrz1,1:nrz2,nr3+1:nrz3) = cmplx(0.0,0.0) f1(1:nrz1,nr2+1:nrz2,1:nrz3) = cmplx(0.0,0.0) f1(nr1+1:nrz1,1:nrz2,1:nrz3) = cmplx(0.0,0.0) ierr=0 end subroutine My_zfc3cl ______________________________________________________________________ > > Could anyone help me to find what is wrong ? > > Many thanks in advance. > > Best regards, > > Fr?d?ric Bouyer > ___________________________________________________________________________ > > sxf90 -o q2r.x q2r.o rigid.o ../PW/pwcom.o ../PW/cft_3.o .../PW/checksym.o > .../PW/coset.o ../PW/cryst_to_car.o ../PW/cubicsym.o ../PW/dsum.o > .../PW/eqvect.o ../PW/error.o ../PW/error_handler.o ../PW/hexsym.o > .../PW/invmat.o ../PW/irrek.o ../PW/kpoint_grid.o ../PW/mode_group.o > .../PW/multable.o ../PW/random.o ../PW/sgam_at.o ../PW/sgam_ph.o > .../PW/sgama.o ../PW/smallg_q.o ../PW/supercell.o ../PW/trnvecc.o > .../PW/wsweight.o ../Modules/basic_algebra_routines.o .../Modules/constants.o > .../Modules/io_global.o ../Modules/io_files.o ../Modules/fft_scalar.o > .../Modules/formats.o ../Modules/kind.o ../Modules/mp.o > .../Modules/mp_global.o ../Modules/parameters.o ../Modules/parser.o > .../Modules/splinelib.o -float0 -P stack -sx6 -g -Wf" -init stack=zero > heap=zero" /home/bouyer/PWSCF/flib/ptools.a /home/bouyer/PWSCF/flib/flib.a > /home/bouyer/PWSCF/clib/clib.a -p -Wl" -f zero " -llapack -lblas > -L/home/bouyer/JMFFT-7.2/ -ljmfft > undefined first referenced > symbol in file > reciprocal_vectors_ ../PW/pwcom.o > control_flags_ ../PW/pwcom.o > fft_types_ ../PW/pwcom.o > parallel_include_ ../Modules/mp.o > shmem_include_ ../Modules/mp_global.o > sxld fatal: symbol referencing errors. no output written to file q2r.x. > > ____________________________________________________________________________ > _ > _______________________________________________ > 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 From a.soon at auckland.ac.nz Fri May 7 12:10:49 2004 From: a.soon at auckland.ac.nz (Aloysius Soon Sze Lok) Date: Fri, 7 May 2004 22:10:49 +1200 Subject: [Pw_forum] Problems starting pw.x Message-ID: <200405072210.49237.a.soon@auckland.ac.nz> Dear all, As a newbie with this code, I have attached my raw input file. There must be something quite wrong with it as I could not get it started. Advice needed. Many thanks. -- Aloysius Soon SL Postgraduate Student Structural & Computational Chemistry Chemistry Department The University Of Auckland, New Zealand Phone: +64 9 373 7599 ext 88291 Email: a.soon at auckland.ac.nz -------------- next part -------------- &control calculation='scf' restart_mode='from_scratch' title='uranium_dioxide_3_layered_slab' pseudo_dir='/DATA/programs/PWscf/pseudo' prefix='uo2' tstress = .true. tprnfor = .true. disk_io='high' / &system ibrav=14 a=3.8666, b=3.8666, c=18.9424, cosab=0.0, cosac=0.0, cosbc=-0.5 nat= 9 ntyp= 2 ecutwfc = 25.0 ecutrho = 150.0 occupations='fixed' / &electrons diagonalization='cg' diago_cg_maxiter=100 conv_thr = 1.0e-6 mixing_beta = 0.7 / ATOMIC_SPECIES U 238.0 U_MT_PBE.UPF O 16.00 O_US_PBE.UPF ATOMIC_POSITIONS angstrom U 0.0000 0.0000 0.0000 U 0.0000 2.2324 15.7853 U 1.9333 1.1162 3.1571 O 0.0000 0.0000 16.5746 O 0.0000 2.2324 3.9476 O 0.0000 2.2324 18.1525 O 0.0000 0.0000 2.3678 O 1.9333 1.1162 14.9948 O 1.9333 1.1162 0.7899 K_POINTS (automatic) 2 2 2 0 0 0 From giannozz at nest.sns.it Fri May 7 12:35:10 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Fri, 7 May 2004 12:35:10 +0200 Subject: [Pw_forum] Problems starting pw.x In-Reply-To: <200405072210.49237.a.soon@auckland.ac.nz> References: <200405072210.49237.a.soon@auckland.ac.nz> Message-ID: <200405071235.10737.giannozz@nest.sns.it> On Friday 07 May 2004 12:10, Aloysius Soon Sze Lok wrote: > As a newbie with this code, I have attached my raw input file. > There must be something quite wrong with it as I could not get > it started. and the output? error message? what happens? on which machine? did you follow the instructions in the manual? -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From eyvaz_isaev at yahoo.com Fri May 7 12:51:41 2004 From: eyvaz_isaev at yahoo.com (Eyvaz Isaev) Date: Fri, 7 May 2004 03:51:41 -0700 (PDT) Subject: [Pw_forum] Problems starting pw.x In-Reply-To: <200405072210.49237.a.soon@auckland.ac.nz> Message-ID: <20040507105141.71818.qmail@web60306.mail.yahoo.com> Hi, Please have a look at the attached file. I have corrected your input file. At least, it looks more likely. But there are some comments. Are you sure that your system is not metallic (i.e. there is a band gap)? Ecutwfc=25 (Ry) seems to be very low, because you are going to use TM (i.e. norm-conserving) type of pseudopoyential. Regards, Eyvaz. --- Aloysius Soon Sze Lok wrote: > Dear all, > As a newbie with this code, I have attached my raw > input file. There must be > something quite wrong with it as I could not get it > started. Advice needed. > > Many thanks. > > > > -- > Aloysius Soon SL > Postgraduate Student > Structural & Computational Chemistry > Chemistry Department > The University Of Auckland, New Zealand > Phone: +64 9 373 7599 ext 88291 > Email: a.soon at auckland.ac.nz > > &control > calculation='scf' > restart_mode='from_scratch' > title='uranium_dioxide_3_layered_slab' > pseudo_dir='/DATA/programs/PWscf/pseudo' > prefix='uo2' > tstress = .true. > tprnfor = .true. > disk_io='high' > / > &system > ibrav=14 > a=3.8666, b=3.8666, c=18.9424, cosab=0.0, > cosac=0.0, cosbc=-0.5 > nat= 9 > ntyp= 2 > ecutwfc = 25.0 > ecutrho = 150.0 > occupations='fixed' > / > &electrons > diagonalization='cg' > diago_cg_maxiter=100 > conv_thr = 1.0e-6 > mixing_beta = 0.7 > / > ATOMIC_SPECIES > U 238.0 U_MT_PBE.UPF > O 16.00 O_US_PBE.UPF > ATOMIC_POSITIONS angstrom > U 0.0000 0.0000 0.0000 > U 0.0000 2.2324 15.7853 > U 1.9333 1.1162 3.1571 > O 0.0000 0.0000 16.5746 > O 0.0000 2.2324 3.9476 > O 0.0000 2.2324 18.1525 > O 0.0000 0.0000 2.3678 > O 1.9333 1.1162 14.9948 > O 1.9333 1.1162 0.7899 > K_POINTS (automatic) > 2 2 2 0 0 0 > > > __________________________________ Do you Yahoo!? Win a $20,000 Career Makeover at Yahoo! HotJobs http://hotjobs.sweepstakes.yahoo.com/careermakeover -------------- next part -------------- A non-text attachment was scrubbed... Name: txt Type: application/octet-stream Size: 917 bytes Desc: txt Url : /pipermail/attachments/20040507/372b1b7e/attachment.obj From giannozz at nest.sns.it Fri May 7 13:07:51 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Fri, 7 May 2004 13:07:51 +0200 Subject: [Pw_forum] Compilation error In-Reply-To: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BE@atil.valrho.cea.fr> References: <601BA98F9E8C7645B7A0052ADBFCC714066EB7BE@atil.valrho.cea.fr> Message-ID: <200405071307.51772.giannozz@nest.sns.it> On Wednesday 05 May 2004 10:01, BOUYER Fr?d?ric 153746 wrote: > [...] I still have the same problem during the compilation. > It stops always at the same point with the same errors. follow Guido's advice properly, and the error will disappear. The compilation stops in q2r.x, in directory pwtools/ . You have to modify the Makefile there. If you don't need the codes contained in pwtools/, you don't have to do anything Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From giannozz at nest.sns.it Fri May 7 14:38:36 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Fri, 7 May 2004 14:38:36 +0200 Subject: [Pw_forum] Convergence on vc-relax In-Reply-To: <408CEDC2.000003.19815@tide.yandex.ru> References: <408CEDC2.000003.19815@tide.yandex.ru> Message-ID: <200405071438.36004.giannozz@nest.sns.it> On Monday 26 April 2004 13:08, Sergei Lisenkov wrote: > I'm performing the variable-cell relaxation (vc-relax) after usual > relaxation [...] What is threshold of vc-relax ? I think that there is no threshold (i.e. the calculation goes on indefinitely) for variable-cell structural optimization. For sure, it was implemented this way and I don't think anybody has changed this behaviour Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From dchrobak at us.edu.pl Mon May 10 11:25:01 2004 From: dchrobak at us.edu.pl (Dariusz Chrobak) Date: Mon, 10 May 2004 11:25:01 +0200 Subject: [Pw_forum] error message during phonon calc. Message-ID: <409F4A6D.9040204@us.edu.pl> Dear PWSCF users, I'm trying to calculate the phonon frequenties at some q-points for Ni FCC. The following information appears at the start of the ph.x run: nbndx = 15 nbnd = 15 natomwfc = 6 npwx = 211 nelec = 10.00 nkb = 18 ngl = 110 warning: negative or imaginary core charge -0.000017 0.000000 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% from phq_setup : error # -1 occupation numbers probably wrong %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% phonons of ni at Gamma Can I ignore this message? My input files are as bellow: ni.in : &control calculation='scf' prefix='ni', pseudo_dir = '/home/pwscf/pseudo//' / &system ibrav=2, celldm(1) =6.66, nat=1, ntyp=1, nspin = 2, starting_magnetization(1)=1.0, ecutwfc =30.0, ecutrho = 288.0, nbnd=15, occupations='tetrahedra' / &electrons conv_thr=1.0D-6, mixing_beta = 0.7 / ATOMIC_SPECIES Ni 58.69 NiUS.RRKJ3.UPF ATOMIC_POSITIONS Ni 0.0 0.0 0.0 K_POINTS {automatic} 8 8 8 0 0 0 ni.phG.in : phonons of ni at Gamma &inputph tr2_ph=1.0d-14, prefix='ni', amass(1)=58.69, fildyn='ni.dynG', / 0.0 0.0 0.0 Best regards Dariusz Chrobak From eyvaz_isaev at yahoo.com Mon May 10 12:15:18 2004 From: eyvaz_isaev at yahoo.com (Eyvaz Isaev) Date: Mon, 10 May 2004 03:15:18 -0700 (PDT) Subject: [Pw_forum] error message during phonon calc. In-Reply-To: <409F4A6D.9040204@us.edu.pl> Message-ID: <20040510101518.76581.qmail@web60310.mail.yahoo.com> Dear Dariusz, I can suggest that it is due to "tetrahedra". I did the same calculations and there was no problem using "smearing". In this case you should use more fine k-points mesh. Regards, Eyvaz. --- Dariusz Chrobak wrote: > Dear PWSCF users, > I'm trying to calculate the phonon frequenties at > some q-points for Ni FCC. > > The following information appears at the start of > the ph.x run: > > > nbndx = 15 nbnd = 15 natomwfc = > 6 npwx = 211 > nelec = 10.00 nkb = 18 ngl = > 110 > warning: negative or imaginary core charge > -0.000017 0.000000 > > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > from phq_setup : error # -1 > occupation numbers probably wrong > > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > phonons of ni at > Gamma > > > Can I ignore this message? > > > My input files are as bellow: > > > ni.in : > > &control > calculation='scf' > prefix='ni', > pseudo_dir = '/home/pwscf/pseudo//' > / > &system > ibrav=2, celldm(1) =6.66, nat=1, ntyp=1, > nspin = 2, starting_magnetization(1)=1.0, > ecutwfc =30.0, ecutrho = 288.0, nbnd=15, > occupations='tetrahedra' > / > &electrons > conv_thr=1.0D-6, > mixing_beta = 0.7 > / > ATOMIC_SPECIES > Ni 58.69 NiUS.RRKJ3.UPF > ATOMIC_POSITIONS > Ni 0.0 0.0 0.0 > K_POINTS {automatic} > 8 8 8 0 0 0 > > ni.phG.in : > > phonons of ni at Gamma > &inputph > tr2_ph=1.0d-14, > prefix='ni', > amass(1)=58.69, > fildyn='ni.dynG', > / > 0.0 0.0 0.0 > > > Best regards > Dariusz Chrobak > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum __________________________________ Do you Yahoo!? Win a $20,000 Career Makeover at Yahoo! HotJobs http://hotjobs.sweepstakes.yahoo.com/careermakeover From giannozz at nest.sns.it Mon May 10 12:31:48 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Mon, 10 May 2004 12:31:48 +0200 Subject: [Pw_forum] error message during phonon calc. In-Reply-To: <20040510101518.76581.qmail@web60310.mail.yahoo.com> References: <20040510101518.76581.qmail@web60310.mail.yahoo.com> Message-ID: <200405101231.48374.giannozz@nest.sns.it> On Monday 10 May 2004 12:15, Eyvaz Isaev wrote: > I can suggest that it is due to "tetrahedra" definitely: the phonon code for metals is implemented only for the gaussian broadening scheme. It should have stopped with an error message. Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy From dchrobak at us.edu.pl Mon May 10 16:46:57 2004 From: dchrobak at us.edu.pl (Dariusz Chrobak) Date: Mon, 10 May 2004 16:46:57 +0200 Subject: [Pw_forum] error message during phonon calc. In-Reply-To: <200405101231.48374.giannozz@nest.sns.it> References: <20040510101518.76581.qmail@web60310.mail.yahoo.com> <200405101231.48374.giannozz@nest.sns.it> Message-ID: <409F95E1.3050703@us.edu.pl> Paolo Giannozzi wrote: >On Monday 10 May 2004 12:15, Eyvaz Isaev wrote: > > > >>I can suggest that it is due to "tetrahedra" >> >> > >definitely: the phonon code for metals is implemented only for >the gaussian broadening scheme. It should have stopped with >an error message. > >Paolo > > > Thank you, D.Ch. From breezejd at lsbu.ac.uk Tue May 11 15:38:50 2004 From: breezejd at lsbu.ac.uk (Jonathan Breeze) Date: Tue, 11 May 2004 14:38:50 +0100 Subject: [Pw_forum] dynamical matrices Message-ID: <200405111438.50771.breezejd@lsbu.ac.uk> Dear Pwscf Forum. Would I be correct in thinking that the dynamical matrices produced by ph.x are not mass weighted ? i.e. D_{ij}(q) = \sum_R C_{ij}e^{-q.R} and not \tilde{D}_{ij}(q) = \frac{1]{\sqrt{M_i M_j}} \sum_R C_{ij}e^{-q.R} -- Jonathan Breeze Research Fellow Centre for Physical Electronics and Materials London South Bank University 103 Borough Road London SE1 0AA Tel: +44(0)20 7815 7582 Fax: +44(0)20 7815 7599 From sahu at matter3.ph.utexas.edu Tue May 11 16:46:02 2004 From: sahu at matter3.ph.utexas.edu (Dr. B.R.Sahu) Date: Tue, 11 May 2004 09:46:02 -0500 (CDT) Subject: [Pw_forum] vc-relax with atomic position fixed. Message-ID: Dear pwscf users, Can one use vc-relax option to optimize the lattice paramters for a fixed volume(for example optimizing "a" and "c" paramters for a given fixed volume) with the atomic position fixed ( b'cause they are already at the high symmetry positions and therefore donot need any relaxation)? I understand that 'relax' option is used only for realxing the atomic positions . regards sahu From iori.federico.8012 at unimo.it Tue May 11 17:10:39 2004 From: iori.federico.8012 at unimo.it (Iori Federico) Date: Tue, 11 May 2004 17:10:39 +0200 Subject: [Pw_forum] (no subject) Message-ID: <1084288239.40a0ecef48672@webmail.unimore.it> ciao.sto scrivendo la mia tesi di laurea e poiche uso il codice PWSCF volevo descrivere "brevemente" gli algoritmi e le procedure su cui si basa il codice stesso.Non ci sarebbe a tal proposito qualche tesi o pubblicazione gia fatta in cui questi argomenti sono trattati in maniera chiara e schematica?? grazie! federico iori ------------------------------------------------- This mail sent through IMP: http://horde.org/imp/ From silviu at Princeton.EDU Wed May 12 04:38:49 2004 From: silviu at Princeton.EDU (Silviu Zilberman) Date: Tue, 11 May 2004 22:38:49 -0400 Subject: [Pw_forum] BFGS Message-ID: <40A18E39.70505@Princeton.EDU> Hi, Is there a way to feed an initial guess of the Hessian in the BFGS structural relaxation, instead of the (presumably) default unity matrix? Thanks, Silviu. -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Zilberman Silviu, PhD 213 Frick Laboratory, Department of Chemistry Princeton University Princeton, NJ 08544 phone: 609-258-1834 fax: 609-258-6746 silviu at Princeton.EDU %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -------------- next part -------------- A non-text attachment was scrubbed... Name: silviu.vcf Type: text/x-vcard Size: 272 bytes Desc: not available Url : /pipermail/attachments/20040511/354a07c7/attachment.vcf From degironc at sissa.it Wed May 12 09:23:29 2004 From: degironc at sissa.it (Stefano de Gironcoli) Date: Wed, 12 May 2004 09:23:29 +0200 Subject: [Pw_forum] vc-relax with atomic position fixed. References: Message-ID: <40A1D0F1.8040704@sissa.it> > > > Can one use vc-relax option to optimize the lattice >paramters for a fixed volume(for example optimizing "a" and "c" paramters >for a given fixed volume) with the atomic position fixed ( b'cause they >are already at the high symmetry positions and therefore donot need any >relaxation)? > > > with the variable cell relaxation option you can look for combined relaxation of atomic positions and cell at a fixed pressure (not volume... the two things are of course related) If the positions are already correct (for symmetry reasons) then the internal forces acting on them should vanish and they will not participate in the relaxation. stefano From g.ballabio at cineca.it Wed May 12 09:42:31 2004 From: g.ballabio at cineca.it (Gerardo Ballabio) Date: Wed, 12 May 2004 09:42:31 +0200 (MEST) Subject: [Pw_forum] vc-relax with atomic position fixed. In-Reply-To: <40A1D0F1.8040704@sissa.it> References: <40A1D0F1.8040704@sissa.it> Message-ID: <1084347675.25166.4.camel@pc-ballabio> On Wed, 2004-05-12 at 09:23, Stefano de Gironcoli wrote: > with the variable cell relaxation option you can look for combined > relaxation of atomic positions and cell at a fixed pressure (not > volume... the two things are of course related) > > If the positions are already correct (for symmetry reasons) then the > internal forces acting on them should vanish and they will not > participate in the relaxation. You may specify a constraint on the atomic structure, which in this case is "all atoms are kept fixed". However I'm not sure that this is possible with variable cell relaxation -- in particular, I wonder whether atoms would be kept fixed in lattice coordinates (that would be correct) or in absolute coordinates (that would be wrong). Gerardo From sbraccia at sissa.it Wed May 12 09:49:49 2004 From: sbraccia at sissa.it (sbraccia carlo) Date: Wed, 12 May 2004 09:49:49 +0200 Subject: [Pw_forum] BFGS In-Reply-To: <40A18E39.70505@Princeton.EDU> References: <40A18E39.70505@Princeton.EDU> Message-ID: <200405120949.49536.sbraccia@sissa.it> On Wednesday 12 May 2004 04:38 am, Silviu Zilberman wrote: > Hi, > > Is there a way to feed an initial guess of the Hessian in the BFGS > structural relaxation, instead of the (presumably) default unity matrix? > > Thanks, Silviu. Yes, it is possible: whenever you start a new relaxation the code looks for a file prefix.hess and, if present, the inverse hessian written on that file is used as initial guess (an information message is also printed). If you can generate a good guess for the inverse hessian of your system you have only to write it on a file: WRITE( unit, * ) SHAPE( inverse_hessian ) WRITE( unit, * ) inverse_hessian where inverse_hessian is a 2-dim array. Put this file in the outdir and change the name to prefix.hess (prefix is the keyword set in &CONTROL). Anyway, you can find all that you need in the file Modules/bfgs_mdule.f90, SUBROUTINES read_bfgs_file() and terminate_bfgs(). carlo From silviu at Princeton.EDU Wed May 12 16:05:58 2004 From: silviu at Princeton.EDU (Silviu Zilberman) Date: Wed, 12 May 2004 10:05:58 -0400 Subject: [Pw_forum] BFGS In-Reply-To: <200405120949.49536.sbraccia@sissa.it> References: <40A18E39.70505@Princeton.EDU> <200405120949.49536.sbraccia@sissa.it> Message-ID: <40A22F46.9060200@Princeton.EDU> sbraccia carlo wrote: Hi Carlo, Thanks for the detailed reply, I'll try it out. Is the hessian in the bfgs implementation given in mass-weighted coordinates? Thanks, Silviu. >On Wednesday 12 May 2004 04:38 am, Silviu Zilberman wrote: > > >>Hi, >> >>Is there a way to feed an initial guess of the Hessian in the BFGS >>structural relaxation, instead of the (presumably) default unity matrix? >> >>Thanks, Silviu. >> >> > >Yes, it is possible: whenever you start a new relaxation the code looks for a >file prefix.hess and, if present, the inverse hessian written on that file is >used as initial guess (an information message is also printed). If you can >generate a good guess for the inverse hessian of your system you have only to >write it on a file: > >WRITE( unit, * ) SHAPE( inverse_hessian ) >WRITE( unit, * ) inverse_hessian > >where inverse_hessian is a 2-dim array. Put this file in the outdir and change >the name to prefix.hess (prefix is the keyword set in &CONTROL). >Anyway, you can find all that you need in the file Modules/bfgs_mdule.f90, >SUBROUTINES read_bfgs_file() and terminate_bfgs(). > >carlo >_______________________________________________ >Pw_forum mailing list >Pw_forum at pwscf.org >http://www.democritos.it/mailman/listinfo/pw_forum > > -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Zilberman Silviu 213 Frick Laboratory, Department of Chemistry Princeton University Princeton, NJ 08544, USA phone: 609-258-1834 fax: 609-258-6746 silviu at Princeton.EDU %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% From sbraccia at sissa.it Wed May 12 16:08:55 2004 From: sbraccia at sissa.it (sbraccia carlo) Date: Wed, 12 May 2004 16:08:55 +0200 Subject: [Pw_forum] BFGS In-Reply-To: <40A22F46.9060200@Princeton.EDU> References: <40A18E39.70505@Princeton.EDU> <200405120949.49536.sbraccia@sissa.it> <40A22F46.9060200@Princeton.EDU> Message-ID: <200405121608.55309.sbraccia@sissa.it> Dear Silviu, > Is the hessian in the bfgs implementation given in mass-weighted > coordinates? no, it is simply the inverse of the second-derivatives matrix. carlo From ps at ned.sims.nrc.ca Wed May 12 13:55:56 2004 From: ps at ned.sims.nrc.ca (Serguei Patchkovskii) Date: Wed, 12 May 2004 07:55:56 -0400 (EDT) Subject: [Pw_forum] Maximing amount of memory available under Linux/x86 In-Reply-To: <200312022208.56162.giannozz@nest.sns.it> Message-ID: Hi folks, Since the last time the question on maximizing the amount of RAM available for dynamic allocation on Linux/x86 systems came up, I had a chance to revisit the issue. As it turns out, recent Linux kernels provide a very convenient way of fiddling with some of the kernel parameters, which required a kernel rebuild before. As of 2.4.21 kernels, TASK_UNMAPPED_BASE can be adjusted on a per-process basis. So, now programs which need to maximize the amount of ALLOCATE'able memory can get it, even with stock kernels. The needed adjustment (to /proc/self/mapped_base) can only be done by euid=0 process. To get it in a normal user process, the it has to be invoked with a small wrapper (attached), which MUST BE SET-UID ROOT. Here is what I get on a stock Suse 9.0 kernel: ps at e05-01:~/snippets> ./bigmem BSS is around 807DD58 ( 128.492 Mb) stack is around BFFFF4B8 ( -1024.003 Mb) Max array size = 2036.000 mbyte Last array was at 409BD040 ( 1033.738 Mb) Max second array size = 0.000 mbyte Last second array was at 409BD040 ( 1033.738 Mb) ps at e05-01:~/snippets> maxram ./bigmem BSS is around 807DD58 ( 128.492 Mb) stack is around BFFFF4D8 ( -1024.003 Mb) Max array size = 2044.000 mbyte Last array was at 109BC040 ( 265.734 Mb) Max second array size = 760.000 mbyte Last second array was at 905BD040 ( -1786.262 Mb) ps at e05-01:~/snippets> uname -a Linux e05-01 2.4.21-199-smp4G #1 SMP ... Enjoy! Serguei --- Dr. Serguei Patchkovskii Tel: +1-(613)-990-0945 Fax: +1-(613)-947-2838 E-mail: Serguei.Patchkovskii at nrc.ca Coordinator of Modelling Software Theory and Computation Group Steacie Institute for Molecular Sciences National Research Council Canada Room 2011, 100 Sussex Drive Ottawa, Ontario K1A 0R6 Canada -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: maxram.c Url: /pipermail/attachments/20040512/4163cf42/attachment.txt From marzari at MIT.EDU Thu May 13 06:24:53 2004 From: marzari at MIT.EDU (Nicola Marzari) Date: Thu, 13 May 2004 00:24:53 -0400 Subject: [Pw_forum] Maximing amount of memory available under Linux/x86 In-Reply-To: References: Message-ID: <40A2F895.6030703@mit.edu> Dear Serguei, many thanks for your comments ! We'll experiment with this. One poor man's solution that we have found out relies on just increasing the number of processes on a single CPU. Let me explain - our first test involves running on 4 CPUs, each of them with 4 GB of physical ram, and suffering from the Linux limit of 2 GB per process. For our test system (a large molecule) we can run on those 4 CPUs with mpirun -np 4 only if we keep the cutoff downt to 15 Ry; a larger cutoff means that the processes hit the 2 GB allocation ceiling. But if we run on those same 4 CPUs with mpirun -np 8, we can go up to a 20 Ry cutoff, and with mpirun -np 12 we can go up to 25 Ry. Performance doesn't seem to degrade much - here are some other tests on a bulk system with ~100 atoms, run on one single CPU, with a physical RAM (DDR3200) of 2GB, and using mpirun -np X, with X=1,2,3,4 . In the ideal case, all jobs should take the same time. Wall clock time reported on an empty machine is: np size per process time(5 iter) (10 iter) (10-5) 1 1721 MB 3084s 5191s 2107s 2 1007 MB 3251s 5580s 2329s 2 1007 MB 3347s 5591s 2247s 3 782 MB 4020s 6463s 2443s 3 782 MB 3556s 5942s 2386s 3 782 MB 3812s 6159s 2347s 4 649 MB 3-6000s 9-17000s 5-10000s Using mpirun -np 4 the computer starts swapping a lot, and performance greatly decreases. Otherwise, it looks OK; the difference in wall time between 10 iterations and 5 iterations increases only by 10-15%. (BTW, all these tests were done with the Democritos CP code, not PWSCF). Best, nicola Serguei Patchkovskii wrote: > Hi folks, > > Since the last time the question on maximizing the amount of > RAM available for dynamic allocation on Linux/x86 systems came > up, I had a chance to revisit the issue. As it turns out, > recent Linux kernels provide a very convenient way of fiddling > with some of the kernel parameters, which required a kernel > rebuild before. As of 2.4.21 kernels, TASK_UNMAPPED_BASE can > be adjusted on a per-process basis. So, now programs which > need to maximize the amount of ALLOCATE'able memory can get > it, even with stock kernels. --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 617.2586534 marzari at mit.edu http://nnn.mit.edu From sahu at matter3.ph.utexas.edu Thu May 13 18:38:05 2004 From: sahu at matter3.ph.utexas.edu (Dr. B.R.Sahu) Date: Thu, 13 May 2004 11:38:05 -0500 (CDT) Subject: [Pw_forum] vc-relax with atomic position fixed. In-Reply-To: <40A1D0F1.8040704@sissa.it> Message-ID: Hi thanx I noticed that use of vc-relax where the lattice paramters and atomic positions are relaxed simultaneously, it takes large number of ionic iterations to converge say 500(ofcourse it depends on the type of atoms, force and energy tolerence, number of k-points, cutoff etc). But is there a way to accelerate the convergence cycle say using the option 'dt=' in the input file. In list of test example with the source code, example 3( for the Al(001) slab relaxation using "relax" dt=120 is used. How does "dt' the time step influences the relaxation in both relax and vc-relax? what value is appropriate? regards sahu > > > with the variable cell relaxation option you can look for combined > relaxation of atomic positions and cell at a fixed pressure (not > volume... the two things are of course related) > > If the positions are already correct (for symmetry reasons) then the > internal forces acting on them should vanish and they will not > participate in the relaxation. > > stefano > From FB153746 at ATIL.CEA.FR Thu May 13 15:19:42 2004 From: FB153746 at ATIL.CEA.FR (=?iso-8859-1?Q?BOUYER_Fr=E9d=E9ric_153746?=) Date: Thu, 13 May 2004 15:19:42 +0200 Subject: [Pw_forum] Wmass keyword Message-ID: <601BA98F9E8C7645B7A0052ADBFCC714066EB7DD@atil.valrho.cea.fr> Dear PWscf users, Could you please tell me what means the wmass keyword (ficticious cell mass for vc simulation, cv-relax), a typical value for this keyword (or how to optimize it), and its unit ? Many thanks in advance. Best regards, Fr?d?ric From cvasse at fy.chalmers.se Thu May 13 18:06:21 2004 From: cvasse at fy.chalmers.se (Vasile Chis) Date: Thu, 13 May 2004 18:06:21 +0200 Subject: [Pw_forum] PBE PP with Van.Code Message-ID: <40A39CFD.4000507@fy.chalmers.se> Dear PWSCF users and developers, I have been struggling with the generation of a pseudopotential using the Vanderbilt code. (First time user) I'm trying to generate a GGA-PBE ultrasoft PP for Cesium. (I know it's not a easy task) My problem in generating the PP is the convergence of the last state when solving the Sch-equation. I've been testing both with 1 valence el. and with 9 valence el. Both giving the same result. Is there anybody how has encontered the same problem and have some advice? Or is there any of you how has the PBE PP for Cesium? Sorry if the formulation of my questions/problem is kind of diffuse. Best regards. /Vasile -- """""""""""""""""""""""""""""""""""" Vasile Chis -*Interface Modelling Group*- """"""""""""""""""""""""""""""""""""" From eyvaz_isaev at yahoo.com Fri May 14 11:17:34 2004 From: eyvaz_isaev at yahoo.com (Eyvaz Isaev) Date: Fri, 14 May 2004 02:17:34 -0700 (PDT) Subject: [Pw_forum] PBE PP with Van.Code In-Reply-To: <40A39CFD.4000507@fy.chalmers.se> Message-ID: <20040514091734.53363.qmail@web60303.mail.yahoo.com> Hi, It is not completely clear what happens in your calculations, but I can suggest to decrease mixing coefficient "damp". I had no problem even for 9 el. Bests, Eyvaz. --- Vasile Chis wrote: > Dear PWSCF users and developers, > > I have been struggling with the generation of a > pseudopotential using > the Vanderbilt code. (First time user) > I'm trying to generate a GGA-PBE ultrasoft PP for > Cesium. (I know it's > not a easy task) > My problem in generating the PP is the convergence > of the last state > when solving the Sch-equation. > I've been testing both with 1 valence el. and with 9 > valence el. Both > giving the same result. > > Is there anybody how has encontered the same problem > and have some advice? > Or is there any of you how has the PBE PP for > Cesium? > > Sorry if the formulation of my questions/problem is > kind of diffuse. > > Best regards. > /Vasile > > -- > """""""""""""""""""""""""""""""""""" > Vasile Chis > -*Interface Modelling Group*- > """"""""""""""""""""""""""""""""""""" > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum __________________________________ Do you Yahoo!? SBC Yahoo! - Internet access at a great low price. http://promo.yahoo.com/sbc/ From sahu at matter3.ph.utexas.edu Fri May 14 18:34:25 2004 From: sahu at matter3.ph.utexas.edu (Dr. B.R.Sahu) Date: Fri, 14 May 2004 11:34:25 -0500 (CDT) Subject: [Pw_forum] vc-relax Message-ID: Dear prof. Stefano yesterday I had sent a query regarding vc-relax in reply to your suggestion. If you have any suggestion on that pl let me know. regards sahu From dchrobak at us.edu.pl Mon May 24 10:23:28 2004 From: dchrobak at us.edu.pl (Dariusz Chrobak) Date: Mon, 24 May 2004 10:23:28 +0200 Subject: [Pw_forum] paralel calculations Message-ID: <40B1B100.5040307@us.edu.pl> Dear PWSCF users, I would like to start 4 processes on two (one-processor) computers. I'm using MPICH-1.2.5.2 MPI implementation with the machines.LINUX file: kometk.ifichm.us.edu.pl kometk.ifichm.us.edu.pl zgredek.ifichm.us.edu.p zgredek.ifichm.us.edu.pl The command: mpirun -np4 ../bin/pw.x -npool 2 < nmgs.in > nmgs.out starts my calculation and create the following PI#### file: kometk.ifichm.us.edu.pl 0 ../pw.x kometk.ifichm.us.edu.pl 1 ../pw.x zgredek.ifichm.us.edu.pl 1 ../pw.x zgredek.ifichm.us.edu.pl 1 ../pw.x Could you tell me are the above numbers correct? What mean these numbers? Best regards Darek Chrobak From giannozz at nest.sns.it Mon May 24 10:46:00 2004 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Mon, 24 May 2004 10:46:00 +0200 Subject: [Pw_forum] paralel calculations In-Reply-To: <40B1B100.5040307@us.edu.pl> References: <40B1B100.5040307@us.edu.pl> Message-ID: <200405241046.00442.giannozz@nest.sns.it> On Monday 24 May 2004 10:23, Dariusz Chrobak wrote: > I would like to start 4 processes on two (one-processor) computers. the only reason to do this is to overcome memory limits > I'm using MPICH-1.2.5.2 MPI implementation [...] The command: > > mpirun -np4 ../bin/pw.x -npool 2 < nmgs.in > nmgs.out > > starts my calculation and create the following PI#### file: > [...] What mean these numbers? the PI#### files are created by MPI, not by PWscf Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050-509876, Fax:-563513 Piazza dei Cavalieri 7 I-56126 Pisa, Italy