From proffess at yandex.ru Sat Nov 1 15:05:42 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Sat, 1 Nov 2003 17:05:42 +0300 (MSK) Subject: [Pw_forum] Unexpected error In-Reply-To: <200310171003.08906.giannozz@nest.sns.it> References: <200310171003.08906.giannozz@nest.sns.it> Message-ID: <3FA3BDB6.000003.13010@camay.yandex.ru> Dear PWscf users, I tried compiled the latest version of PWscf on linux box using intel 7.0. I got error: bash-2.05a$ make pw ( cd Modules; make all ) make[1]: Entering directory `/nethome/proffess/O-sesame/Modules' mpif90 -Vaxlib -O3 -tpp7 -r8 -fpp -D__LINUX -D__INTEL -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/nethome/proffess/kappa/lib/fftw-2.1.5/include -nomodule -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c kind.f90 module KINDS module subroutine PRINT_KIND_INFO 56 Lines Compiled mpif90 -Vaxlib -O3 -tpp7 -r8 -fpp -D__LINUX -D__INTEL -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/nethome/proffess/kappa/lib/fftw-2.1.5/include -nomodule -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c constants.f90 module CONSTANTS 81 Lines Compiled mpif90 -Vaxlib -O3 -tpp7 -r8 -fpp -D__LINUX -D__INTEL -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/nethome/proffess/kappa/lib/fftw-2.1.5/include -nomodule -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c parallel_include.f90 module PARALLEL_INCLUDE 238 Lines Compiled mpif90 -Vaxlib -O3 -tpp7 -r8 -fpp -D__LINUX -D__INTEL -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/nethome/proffess/kappa/lib/fftw-2.1.5/include -nomodule -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c shmem_include.f90 module SHMEM_INCLUDE 37 Lines Compiled mpif90 -Vaxlib -O3 -tpp7 -r8 -fpp -D__LINUX -D__INTEL -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/nethome/proffess/kappa/lib/fftw-2.1.5/include -nomodule -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c parameters.f90 module PARAMETERS Error FCE31 : Module dictionary file PARAMETERS.mod could not be created compilation aborted for parameters.f90 (code 1) make[1]: *** [parameters.o] Error 1 make[1]: Leaving directory `/nethome/proffess/O-sesame/Modules' make: *** [modules] Error 2 What's happen? Best wishes, Sergey From proffess at yandex.ru Sat Nov 1 15:13:31 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Sat, 1 Nov 2003 17:13:31 +0300 (MSK) Subject: [Pw_forum] Unexpected error - do not consider! In-Reply-To: <3FA3BDB6.000003.13010@camay.yandex.ru> References: <200310171003.08906.giannozz@nest.sns.it> <3FA3BDB6.000003.13010@camay.yandex.ru> Message-ID: <3FA3BF8B.000005.05547@pantene.yandex.ru> Dear PWscf users, Please, remove my message - It was my internal error. Sergey From proffess at yandex.ru Tue Nov 4 13:47:38 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Tue, 4 Nov 2003 15:47:38 +0300 (MSK) Subject: [Pw_forum] Initial Temperatute in vc-relax In-Reply-To: <200310171003.08906.giannozz@nest.sns.it> References: <200310171003.08906.giannozz@nest.sns.it> Message-ID: <3FA79FEA.000006.07542@tide.yandex.ru> Dear PWscf authors and users, Please, explain, how to set up initial temperature in vc-relax. I have seen that by optimization temperature starts from 0.0 K. Is it possible to setup another value of temperature? Thanks. Best wishes, Sergey From proffess at yandex.ru Tue Nov 4 13:49:15 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Tue, 4 Nov 2003 15:49:15 +0300 (MSK) Subject: [Pw_forum] Auxilirary files in vc-relax In-Reply-To: <200310171003.08906.giannozz@nest.sns.it> References: <200310171003.08906.giannozz@nest.sns.it> Message-ID: <3FA7A04B.000003.08548@soapbox.yandex.ru> Dear PWscf users and authors, What does mean the files ave, avec, e, ... ? Thanks, Sergey From breezejd at lsbu.ac.uk Tue Nov 4 15:10:26 2003 From: breezejd at lsbu.ac.uk (Jonathan Breeze) Date: Tue, 04 Nov 2003 14:10:26 +0000 Subject: [Pw_forum] Linux cluster Message-ID: <3FA7B352.6040804@lsbu.ac.uk> Hello everybody, I have recently built a linux cluster using diskless nodes. I have patched the kernel with OpenMosix and am now trying to get MPI processes to migrate across nodes. I had heard that it is possible to turn off shared memory (which prevents process migration) when compiling LAM-MPI using the '--with-rpi=tcp' option. Unfortunately, I am using MPICH with Intel Fortran Compiler. Any advice on which MPI implementation is best to use with PWSCF or accounts of similar experiences would be a GREAT help! Many thanks. -- 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 giannozz at nest.sns.it Tue Nov 4 22:50:19 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Tue, 4 Nov 2003 22:50:19 +0100 Subject: [Pw_forum] what atoms are counted by ntype In-Reply-To: <3FA28CFB.E4E61A5B@lmcp.jussieu.fr> References: <3FA28CFB.E4E61A5B@lmcp.jussieu.fr> Message-ID: <200311042250.19073.giannozz@nest.sns.it> On Friday 31 October 2003 17:25, Michele Lazzeri wrote: > You can decide to describe two crystallographically > equivalent atoms with two different pseudopotentials. > In this case the code will not recognize the two atoms as equivalent. just a remark: it is perfectly acceptable to use the same pseudopotential for different types of atoms. This is what is typically done in magnetic systems. Paolo From breezejd at lsbu.ac.uk Thu Nov 6 18:49:20 2003 From: breezejd at lsbu.ac.uk (Jonathan Breeze) Date: Thu, 06 Nov 2003 17:49:20 +0000 Subject: [Pw_forum] shared memory Message-ID: <3FAA89A0.8000102@lsbu.ac.uk> Dear Pwscf'ers, Would I be correct in thinking that PW uses shared memory in its MPI usage ??? -- 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 giannozz at nest.sns.it Fri Nov 7 10:32:12 2003 From: giannozz at nest.sns.it (Paolo GIANNOZZI) Date: Fri, 07 Nov 2003 10:32:12 +0100 Subject: [Pw_forum] shared memory In-Reply-To: <3FAA89A0.8000102@lsbu.ac.uk> Message-ID: On Thu, 06 Nov 2003 17:49:20 +0000 Jonathan Breeze wrote: > Would I be correct in thinking that PW uses shared memory > in its MPI usage ??? I would say that yiou aren't. PW starts N independent processes that communicate via calls to MPI libraries. Each process has its own set of variables and knows nothing about other precesses' variables. Variables that take little memopry are replicated, those that take a lot of memory (wavefunctions, G-vectors, R-space grid) are distributed Paolo From giannozz at nest.sns.it Fri Nov 7 15:32:39 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Fri, 7 Nov 2003 15:32:39 +0100 (CET) Subject: [Pw_forum] problem with input redirection with plotrho.x In-Reply-To: <20030813165020.GA20811@node1.cluster.srrc.usda.gov> Message-ID: On Wed, 13 Aug 2003 11:50:20 -0500 Glenn Johnson wrote: > I just compiled pwscf-1.2 with Linux IFC-7.1, Build > 20030307Z [...] > In example5, I get an error with plotrho.x [..] > I looked at the Fortran code in plotrho.f90 good attitude > but nothing jumped out at me as being wrong. I think there isn't anything wrong, in fact > Has anyone seen this before? I didn't, but I have seen all kind of funny behavior with some 7.1 versions of the intel compiler. Try to get the very latest release: Intel 7.1-35 (build 20030922Z) seems to work quite well (apart from the misterious warning: "size of symbol `MODULE.reciprocal_vectors_1' changed from 212 to 556 in ../Modules/recvec.o" at loading, still present, at least with the latest version of PWscf) Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From likedew at phys.ksu.edu Fri Nov 7 21:57:34 2003 From: likedew at phys.ksu.edu (Hong, SamPyo) Date: Fri, 7 Nov 2003 14:57:34 -0600 Subject: [Pw_forum] Projection? Message-ID: Hi, I have a simple question about projection. As far as I know, before doing projection of each |phi_i> to |evc>, we should have a set of the orthogonized wavefunctions i.e, = 0 if i&j are different. For example, in the following routine, if we replace temp = ZDOTC(npw,swfcatom(1,i),1,evc(1,ibnd),1) by temp = ZDOTC(npw,swfcatom(1,i),1,wfcatom(1,i),1) Then, temp must be zero. However, it seems not to be true. Why is that? ---------------PWSCF 1.03 projwave.F---------------------------------------- c c make the projection c do ibnd = 1, nbnd psum = 0.d0 do i=1,natomwfc temp = ZDOTC(npw,swfcatom(1,i),1,evc(1,ibnd),1) work(i,1) = temp enddo #ifdef PARA call reduce(2*natomwfc,work) #endif C ------------------------------------------------------------------------ ---- Thanks, Sampyo From proffess at yandex.ru Sun Nov 9 11:03:49 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Sun, 9 Nov 2003 13:03:49 +0300 (MSK) Subject: [Pw_forum] Compiling error In-Reply-To: <200310171003.08906.giannozz@nest.sns.it> References: <200310171003.08906.giannozz@nest.sns.it> Message-ID: <3FAE1105.000003.22225@tide.yandex.ru> Dear PWscf authors, I have met a compiling error of recent version of PWscf on ALpha machine. On another nachines it was ignored. mpif90 -O -fpe0 -real_size 64 -align dcommons -align records -free -cpp -D__ALPHA -D__LINUX -D__PARA -D__MPI -D__FFTW -I/nethome/proffess/O-sesame/include -I/home/proffess/lib/fftw-2.1.3/include -I/nethome/proffess/O-sesame/Modules -I/nethome/proffess/O-sesame/PW -I/nethome/proffess/O-sesame/PH -c setlocal.f90 /nethome/proffess/O-sesame/include/machine.h:11: warning: "C_POINTER" redefined f90: Warning: setlocal.f90, line 19: Conflicting attributes or multiple declaration of name. [NL] USE gvect, ONLY : nl, nlm, igtongl ---------------------^ f90: Warning: setlocal.f90, line 19: Conflicting attributes or multiple declaration of name. [NLM] USE gvect, ONLY : nl, nlm, igtongl -------------------------^ f90: Error: setlocal.f90, line 37: The same named entity from different modules and/or program units cannot be referenced. [NL] aux (nl(ng))=aux(nl(ng)) + vloc (igtongl (ng), nt) * strf (ng, nt) -------------^ f90: Error: setlocal.f90, line 37: The same named entity from different modules and/or program units cannot be referenced. [NL] aux (nl(ng))=aux(nl(ng)) + vloc (igtongl (ng), nt) * strf (ng, nt) -------------------------^ f90: Error: setlocal.f90, line 42: The same named entity from different modules and/or program units cannot be referenced. [NLM] aux (nlm(ng)) = conjg (aux (nl(ng))) -------------^ f90: Error: setlocal.f90, line 42: The same named entity from different modules and/or program units cannot be referenced. [NL] aux (nlm(ng)) = conjg (aux (nl(ng))) ------------------------------------^ make[1]: *** [setlocal.o] Error 1 make[1]: Leaving directory `/nethome/proffess/O-sesame/PW' make: *** [pw] Error 2 Best, Sergey From giannozz at nest.sns.it Mon Nov 10 11:06:12 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Mon, 10 Nov 2003 11:06:12 +0100 Subject: [Pw_forum] Projection? In-Reply-To: References: Message-ID: <200311101106.12359.giannozz@nest.sns.it> On Friday 07 November 2003 21:57, Hong, SamPyo wrote: > As far as I know, before doing projection of each |phi_i> to |evc>, we > should have a set of the orthogonized wavefunctions i.e, > = 0 if i&j are different. For example, in the following routine, > if we replace > > temp = ZDOTC(npw,swfcatom(1,i),1,evc(1,ibnd),1) > by > temp = ZDOTC(npw,swfcatom(1,i),1,wfcatom(1,i),1) > > Then, temp must be zero. However, it seems not to be true. Why is that? > > ---------------PWSCF 1.03 > projwave.F---------------------------------------- > c > c make the projection > c > do ibnd = 1, nbnd > psum = 0.d0 > do i=1,natomwfc > temp = ZDOTC(npw,swfcatom(1,i),1,evc(1,ibnd),1) > work(i,1) = temp > enddo > #ifdef PARA > call reduce(2*natomwfc,work) > #endif swfcatom does not contain S|phi_i> (S is the ultrasoft overlap matrix: S=1 for norm-conserving pseudopotentials), but O^(-1/2) S|phi_i>, where O_(ij) = . This is done in the lines just above those you show: c c trasform atomic orbitals O^-1 psi c do i=1,npw call setv(2*natomwfc,0.d0,work,1) call ZGEMV('n',natomwfc,natomwfc,(1.d0,0.d0),overlap, * natomwfc,swfcatom(i,1),npwx, (0.d0,0.d0),work,1) call ZCOPY(natomwfc,work,1,swfcatom(i,1),npwx) end do The orthonormal atomic wavefunctions are O^(-1/2) |phi_i> (orthonormal wrt the S operator for ultrasoft potentials). The comment to these lines ("trasform atomic orbitals O^-1 psi") was - as most comments in computer codes are - obscure and misleading. It is correct in more recent versions of the code. Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From mtoso at ts.infn.it Mon Nov 10 11:58:02 2003 From: mtoso at ts.infn.it (Michele Tosolini) Date: Mon, 10 Nov 2003 11:58:02 +0100 (MET) Subject: [Pw_forum] compiling on Unix alpha Message-ID: Dear PWscf users, when trying to compile pw.1.3.0 on a Unix compaq alpha, after "./configure alpha" and "make pw" i get this: ..... cc -O -D__ALPHA -D__FFTW -D__USE_INTERNAL_FFTW -I/cluster/members/member0/tmp/mtoso/include -I./ -c fft_stick.c ar rv clib.a c_getenv.o c_today.o cpflush.o factor.o indici.o memstat.o qsort.o readocc.o round2.o cptimer.o fft_stick.o a - c_getenv.o a - c_today.o a - cpflush.o a - factor.o a - indici.o a - memstat.o a - qsort.o a - readocc.o a - round2.o a - cptimer.o a - fft_stick.o ar: Warning: creating clib.a Make: Cannot open .dependencies. Stop. *** Exit 1 Stop. How can i manage it? thanks. By the way, are there any specific suggestions for compiling flags on Unix alpha machine? PS: Makefile in flib/ says "include .dependencies" , but there's no such file there. Is that the problem? From giannozz at nest.sns.it Mon Nov 10 15:42:16 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Mon, 10 Nov 2003 15:42:16 +0100 Subject: [Pw_forum] compiling on Unix alpha In-Reply-To: References: Message-ID: <200311101542.16897.giannozz@nest.sns.it> On Monday 10 November 2003 11:58, Michele Tosolini wrote: > when trying to compile pw.1.3.0 on a Unix compaq alpha, after > "./configure alpha" and "make pw" i get this: [...] > Make: Cannot open .dependencies. Stop. > PS: Makefile in flib/ says "include .dependencies" , but there's no > such file there. Is that the problem? it is. Create directories "FPMD" and "CP" and run "./configure alpha" again. Does it work (RSVP) ? Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From mtoso at ts.infn.it Mon Nov 10 16:49:27 2003 From: mtoso at ts.infn.it (Michele Tosolini) Date: Mon, 10 Nov 2003 16:49:27 +0100 (MET) Subject: [Pw_forum] compiling on Unix alpha In-Reply-To: <200311101542.16897.giannozz@nest.sns.it> References: <200311101542.16897.giannozz@nest.sns.it> Message-ID: > > > PS: Makefile in flib/ says "include .dependencies" , but there's no > > such file there. Is that the problem? > > it is. Create directories "FPMD" and "CP" and run "./configure alpha" > again. Does it work (RSVP) ? > > Paolo > -- Yes I compile it commenting the appropriate lines in install/shdep since i don't need FPMD or CPV yet. Thanks a lot. From giannozz at nest.sns.it Wed Nov 12 11:39:04 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Wed, 12 Nov 2003 11:39:04 +0100 Subject: [Pw_forum] Compiling error In-Reply-To: <3FAE1105.000003.22225@tide.yandex.ru> References: <200310171003.08906.giannozz@nest.sns.it> <3FAE1105.000003.22225@tide.yandex.ru> Message-ID: <200311121139.04127.giannozz@nest.sns.it> On Sunday 09 November 2003 11:03, Sergei Lisenkov wrote: > I have met a compiling error of recent version of PWscf on Alpha machine. recent = "from CVS" ? it should be fixed now Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From goranka.bilalbegovic at zg.hinet.hr Thu Nov 13 09:05:03 2003 From: goranka.bilalbegovic at zg.hinet.hr (Goranka Bilalbegovic) Date: Thu, 13 Nov 2003 09:05:03 +0100 Subject: [Pw_forum] pw.1.3.0 on Linux cluster (mpirun) Message-ID: <000d01c3a9bd$04dcd400$45841dc3@gost> Hello. I am executing pw.1.2.0 on the Linux cluster (mpi, PBS) without problems using PARA_PREFIX = mpirun -np 8, PARA_POSTFIX = -npool 8. However, pw.1.3.0 does not work on the same cluster with the same execution line. Job starts, but the files in out_dir are not formed. I also tried with PARA_PREFIX = mpirun -n 8 (i.e, as for clusters using mpiexec). Then the file (for example 1) silicon.save is formed in out_dir, but it does not take n as a number of processors. Thanks for help. Best regards, Goranka -------------- next part -------------- An HTML attachment was scrubbed... URL: /pipermail/attachments/20031113/8fbaa688/attachment.htm From giannozz at nest.sns.it Thu Nov 13 10:56:52 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Thu, 13 Nov 2003 10:56:52 +0100 Subject: [Pw_forum] pw.1.3.0 on Linux cluster (mpirun) In-Reply-To: <000d01c3a9bd$04dcd400$45841dc3@gost> References: <000d01c3a9bd$04dcd400$45841dc3@gost> Message-ID: <200311131056.52267.giannozz@nest.sns.it> On Thursday 13 November 2003 09:05, Goranka Bilalbegovic wrote: > I am executing pw.1.2.0 on the Linux cluster (mpi, PBS) compiler ? > without problems using PARA_PREFIX = mpirun -np 8, > PARA_POSTFIX = -npool 8. > However, pw.1.3.0 does not work on the same cluster with the > same execution line. Job starts, on how many processors ? it is written when execution starts > but the files in out_dir are not formed. does the calculation go on ? does it stop with error message ? without ? > I also tried with PARA_PREFIX = mpirun -n 8 (i.e, as for clusters using > mpiexec). Then the file (for example 1) silicon.save is formed in out_dir, > but it does not take n as a number of processors. it shouldn't, I think: just one *.save file is written, independently on the number of processors used Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From proffess at yandex.ru Thu Nov 13 11:55:20 2003 From: proffess at yandex.ru (Sergei Lisenkov) Date: Thu, 13 Nov 2003 13:55:20 +0300 (MSK) Subject: [Pw_forum] UPF in old format In-Reply-To: <200310171003.08906.giannozz@nest.sns.it> References: <200310171003.08906.giannozz@nest.sns.it> Message-ID: <3FB36318.000003.19863@pantene.yandex.ru> Dear PWscf authors, Do you have any utilities that converts UPF (new pseudopotentials) in old format also aceppted by PWscf (old version)? Thanks a lot, Best wishes, Sergey From giannozz at nest.sns.it Thu Nov 13 12:30:19 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Thu, 13 Nov 2003 12:30:19 +0100 Subject: [Pw_forum] UPF in old format In-Reply-To: <3FB36318.000003.19863@pantene.yandex.ru> References: <200310171003.08906.giannozz@nest.sns.it> <3FB36318.000003.19863@pantene.yandex.ru> Message-ID: <200311131230.19338.giannozz@nest.sns.it> On Thursday 13 November 2003 11:55, Sergei Lisenkov wrote: > Do you have any utilities that converts UPF (new pseudopotentials) > in old format also accepted by PWscf (old version)? no, but it is relatively simple to write an "anti-converter" using existing converters. PPs in www.pwscf.org are still available in old formats, unless they have been added recently. Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From andrea.vittadini at unipd.it Fri Nov 14 11:55:38 2003 From: andrea.vittadini at unipd.it (Andrea Vittadini) Date: 14 Nov 2003 11:55:38 +0100 Subject: [Pw_forum] optimization in crystal coordinates Message-ID: <1068807338.1255.15.camel@tond.chin.unipd.it> Dear all, I'm doing some optimizations with pwscf.1.3.0 of some rhombohedral and hexagonal structures by entering atomic positions in "crystal" mode. What I find is that the final (and intermediate) atomic positions (also of the "crystal" type) appear to be messed up. The same does not occur if I enter the "a0" coordinates computed by the program from my initial crystal positions, which means that my initial coordinates are correct. Anybody found this problem? Thanks, Andrea From louise.dash at polytechnique.fr Fri Nov 14 12:12:30 2003 From: louise.dash at polytechnique.fr (Louise Dash) Date: Fri, 14 Nov 2003 12:12:30 +0100 Subject: [Pw_forum] optimization in crystal coordinates In-Reply-To: <1068807338.1255.15.camel@tond.chin.unipd.it> References: <1068807338.1255.15.camel@tond.chin.unipd.it> Message-ID: <3FB4B89E.60800@polytechnique.fr> Andrea Vittadini wrote: > Dear all, > I'm doing some optimizations with pwscf.1.3.0 of some rhombohedral and > hexagonal structures by entering atomic positions in "crystal" mode. > What I find is that the final (and intermediate) atomic positions (also > of the "crystal" type) appear to be messed up. > The same does not occur if I enter the "a0" coordinates computed by the > program from my initial crystal positions, which means that my initial > coordinates are correct. > Anybody found this problem? > Thanks, > Andrea Dear Andrea I recently encountered the same problem (I think!) with pwscf 1.3.0 on monoclinic structures - the coordinates in the output file were wrong when I performed a relaxation, but the energies (and everything else) seemed OK. As far as I could tell, the problem was *only* in the output file, not with the calculations themselves. In the subroutine output_tau (file output_tau.f90) changing the line call cryst_to_cart (nat, tau_out, at, 1) to call cryst_to_cart (nat, tau_out, bg, -1) should sort this out - I think the conversion from crystal to cartesian coordinates was the wrong way round. Hope this helps Louise -- Louise Dash Laboratoire des Solides Irradies, Ecole Polytechnique F-91128 Palaiseau, France Tel +33 1 69 33 45 11 http://theory.polytechnique.fr From giannozz at nest.sns.it Fri Nov 14 14:16:23 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Fri, 14 Nov 2003 14:16:23 +0100 Subject: [Pw_forum] optimization in crystal coordinates In-Reply-To: <3FB4B89E.60800@polytechnique.fr> References: <1068807338.1255.15.camel@tond.chin.unipd.it> <3FB4B89E.60800@polytechnique.fr> Message-ID: <200311141416.23597.giannozz@nest.sns.it> On Friday 14 November 2003 12:12, Louise Dash wrote: > I think the conversion from crystal to cartesian coordinates > was the wrong way round. correct (I know: I did it!). It affects only what is written, though, not what is calculated. Let me remind a few other problems that are known to exist version 1.3.0 of PWscf: * phonon with k-point parallelization (pools) yields wrong results in some cases (since v. 1.2 included) * upftools/cpmd2upf.f90: wrong conversion due to Rydberg-Hartree mess * PW/input.f90: lattice parameter a converted to wrong units if input is given as a,b,c,cos(ab),cos(ac),cos(bc) instead of celldm(:) Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From goranka.bilalbegovic at zg.hinet.hr Sat Nov 15 13:26:42 2003 From: goranka.bilalbegovic at zg.hinet.hr (Goranka Bilalbegovic) Date: Sat, 15 Nov 2003 13:26:42 +0100 Subject: [Pw_forum] pw.1.3.0 on Linux cluster (mpirun) References: <000d01c3a9bd$04dcd400$45841dc3@gost> <200311131056.52267.giannozz@nest.sns.it> Message-ID: <001001c3ab73$d917a4e0$19311dc3@gost> The compiler for pw.1.3.0 is PGI 4.1.2 and it was ifc 7.1.008 for pw.1.2.0. I forgot that I changed the compiler because of the ifc "compiler internal error, please report to Intel corporation..." I do not have ifc 6.*, and would be happy to know which version of ifc 7.1.* (if any) compiles pw.1.3.0. Up to know I have tested PGI compiled pw.1.3.0 with 2 (same node) and 8 processors (4x2). The number of processors jobs use is always correct. When I use PARA_PREFIX= mpirun -np #, job starts, the files in out_dir are not formed, calculation goes on (i.e., jobs only eat time), perhaps forever, no error massages, output files are empty. For PARA_PREFIX= mpirun -n #, the file job.save forms, but (i) for 2 processors on the same node it runs forever, pw does not start to write (ii) for 8 processors jobs stops after writing: ---------------------------------------------- Current dimensions of program pwscf are: ntypx =10 npk =40000 lmax = 3 nchix = 6 ndim = 2000 nbrx = 8 nqfm = 8 ----------------------------------------------- In both cases (with PARA_PREFIX= mpirun -n #) there is a message: ----------------------------------------------- Warning: Command line arguments for program should be given after the program name. Assuming that 2 is a command line argument for the program. Unrecognized argument -n ignored. ------------------------------------------------ Unfortunately, I do not have the same type of test jobs for 2 and 8 processors, and the cluster is now very busy. Thanks for help. Best regards, Goranka From giannozz at nest.sns.it Sun Nov 16 16:02:46 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Sun, 16 Nov 2003 16:02:46 +0100 Subject: [Pw_forum] pw.1.3.0 on Linux cluster (mpirun) In-Reply-To: <001001c3ab73$d917a4e0$19311dc3@gost> References: <000d01c3a9bd$04dcd400$45841dc3@gost> <200311131056.52267.giannozz@nest.sns.it> <001001c3ab73$d917a4e0$19311dc3@gost> Message-ID: <200311161602.46524.giannozz@nest.sns.it> On Saturday 15 November 2003 13:26, Goranka Bilalbegovic wrote: > [...] I would be happy to know which version of ifc 7.1.* (if any) > compiles pw.1.3.0. the last version (ifc 7.1.35) compiles pw.1.3.0 and later with no apparent problem (apart from a weird message at linking stage) > When I use PARA_PREFIX= mpirun -np #, job starts, the files in out_dir are > not formed, calculation goes on (i.e., jobs only eat time), perhaps > forever, no error massages, output files are empty. check first if it works - compiled for serial execution - compiled for parallel execution, executed on one processor > For PARA_PREFIX= mpirun -n # this is not the correct syntax Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19 From konstantin_kudin at yahoo.com Mon Nov 17 05:18:26 2003 From: konstantin_kudin at yahoo.com (Konstantin Kudin) Date: Sun, 16 Nov 2003 20:18:26 -0800 (PST) Subject: [Pw_forum] Optimization in PWSCF Message-ID: <20031117041826.23578.qmail@web21204.mail.yahoo.com> Hi there, I am starting to use PWSCF 1.3.0 and have a question on how "vc-relax" option works. I have looked up what I could find and there seems to be very little info on that. What is the fastest way to get a full optimization (atomic coordinates + lattice vectors) of a given structure at 0 K and 0 pressure? The code demands "wmass" if I specify as little as possible in the input, how do I choose this value in a reasonable way? Another question is about symmetry. Can I just input {ibrav=0}, atomic coordinates, 3 lattice vectors in {CELL_PARAMETERS} and let the code figure out the symmetry and everything else for me? Are there any advantages in doing it manually versus letting the program do it? Thanks a lot in advance! Konstantin __________________________________ Do you Yahoo!? Protect your identity with Yahoo! Mail AddressGuard http://antispam.yahoo.com/whatsnewfree From eyvaz_isaev at yahoo.com Mon Nov 17 16:01:54 2003 From: eyvaz_isaev at yahoo.com (Eyvaz Isaev) Date: Mon, 17 Nov 2003 07:01:54 -0800 (PST) Subject: [Pw_forum] AlphaMPI Message-ID: <20031117150154.24890.qmail@web60305.mail.yahoo.com> Dear all, I spent lot of time to run a whole script on a AlphaMPI based supercomputer, but I met a problem. A job was run as "mpirun -np N -maxtime Y ./job". "Job" is a script which contains all needed procedures. It turned out that the "mpirun" command acts only on the first script inside "Job" and then the system knows nothing about "mpirun". Sure, I can split a script onto subscripts, but it is not convinient. Does anybody know how to fix the problem? Thank you in advance, Eyvaz. __________________________________ Do you Yahoo!? Protect your identity with Yahoo! Mail AddressGuard http://antispam.yahoo.com/whatsnewfree From goranka.bilalbegovic at zg.hinet.hr Tue Nov 18 08:28:24 2003 From: goranka.bilalbegovic at zg.hinet.hr (Goranka Bilalbegovic) Date: Tue, 18 Nov 2003 08:28:24 +0100 Subject: [Pw_forum] pw.1.3.0 on Linux cluster (mpirun) References: <000d01c3a9bd$04dcd400$45841dc3@gost> <200311131056.52267.giannozz@nest.sns.it> <001001c3ab73$d917a4e0$19311dc3@gost> <200311161602.46524.giannozz@nest.sns.it> Message-ID: <000b01c3ada5$a8bd6760$eb301dc3@gost> Hello. > check first if it works > - compiled for serial execution Yes, it works. > - compiled for parallel execution, executed on one processor No, it does not work (calculation goes on, empty output, no error messages, job.save is formed in this case). In conclusions, pw.1.3.0 compiled with PGI 4.1.2 does not work on the Linux cluster I use (PBS, mpirun). Best regards, Goranka From breezejd at lsbu.ac.uk Tue Nov 18 17:34:36 2003 From: breezejd at lsbu.ac.uk (Jonathan Breeze) Date: Tue, 18 Nov 2003 16:34:36 +0000 Subject: [Pw_forum] compiling pw.1.2.0 using mpich and ifc for linux cluster Message-ID: <3FBA4A1C.4020502@lsbu.ac.uk> Hello everybody, I have a little linux cluster of 8 nodes and have been trying to compile a parallel version of pw.1.2.0 for it using the intel compiler version 7.1. I am using the MPICH library (version 1.2.5.2) which I compiled from source using ifc7.1 to produce mpif90. I ran the simple pi3f90 fortran test example using 'mpirun -np 8 a.out' and it worked fine, all nodes happy and running their own process. I ran further tests which showed that processes where running on nodes and communicating properly. Not so happy when I compiled a parallel pw.1.2.0. :( I had managed to compile the serial version using blas, lapack and atlas libraries. It produces the correct results and functions very well. I compiled a parallel version, which compiled without any errors (I used the -DPARA -D__MPI CPP flags and set the compiler to mpif90). When I run the parallel pw.x using mpirun -np 4 pw.x I get the following message: mpirun -np 4 bin/pw.x ** Address Error ** End of diagnostics Oh dear! The program didn't get very far. The make.sys file is included here. Can any kind soul offer some assistance ?? Many thanks. OSHOME=/home/jonathan/pw.1.2.0p CPPFLAGS = -I$(OSHOME)/include/ -DPC -DPARA -D__MPI -DFFTW \ -D"FFTWND_F77_ONE=fftwnd_f77_one_" \ -D"FFTW3D_F77_CREATE_PLAN=fftw3d_f77_create_plan_" \ -D"FFTW_F77=fftw_f77_" \ -D"FFTW_F77_CREATE_PLAN=fftw_f77_create_plan_" F90 = mpif90 FFLAGS =-Vaxlib -W0 F90FLAGS= $(FFLAGS) -fpp $(CPPFLAGS) MODULEFLAG= -nomodule -I $(OSHOME)/Modules -I$(OSHOME)/PW -I$(OSHOME)/PH LIBS = -L/usr/local/lib -lfftw -L/home/jonathan/libs -llapack -lf77blas -lcblas -latlas \ -L/usr/lib/gcc-lib/i386-redhat-linux/3.2.2/ -lg2c # LD=$(F90) LFLAGS = -Vaxlib $(LIBS) # # ar: # AR = ar ARFLAGS = ruv -- Jonathan Breeze Research Fellow Centre for Physical Electronics and Materials London South Bank University 103 Borough Road London SE1 0AA Tel: +44(0)20 7815 7582 Fax: +44(0)20 7815 7599 From goranka.bilalbegovic at zg.hinet.hr Tue Nov 18 22:41:35 2003 From: goranka.bilalbegovic at zg.hinet.hr (Goranka Bilalbegovic) Date: Tue, 18 Nov 2003 22:41:35 +0100 Subject: [Pw_forum] make PW error References: <200309251905.51100.giannozz@nest.sns.it> <200309272259.08184.giannozz@nest.sns.it> Message-ID: <001001c3ae1c$dc7cf580$743c1dc3@gost> > On Friday 26 September 2003 15:18, Victor Trubitsin wrote: > > Hm..., I have not two versions of this module! I removed all : > > *o,.d.*.mod and recompiled and again got the same error. > > does this happen with version 7 of intel compiler (ifc)? try the very > latest release, or an earlier version > I have the same error with the latest release of ifc, warning at linking stage followed by a memory error, therefore the executable does not exist ("ld: Warning: size of symbol `MODULE.reciprocal_vectors_1' changed from 212 to 556 in ../Modules/recvec.o, make[1]: *** [memory] Error 1"). Is this also a compiler error, or someone compiled pw.1.3.0 with ifc 7.* ? Best regards, Goranka From giannozz at nest.sns.it Wed Nov 19 14:42:16 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Wed, 19 Nov 2003 14:42:16 +0100 (CET) Subject: [Pw_forum] make PW error In-Reply-To: <001001c3ae1c$dc7cf580$743c1dc3@gost> Message-ID: On Tue, 18 Nov 2003, Goranka Bilalbegovic wrote: > I have the same error with the latest release of ifc, warning at linking > stage followed by a memory error, therefore the executable does not exist > ("ld: Warning: size of symbol `MODULE.reciprocal_vectors_1' changed from 212 > to 556 in ../Modules/recvec.o, make[1]: *** [memory] Error 1"). it's not a "memory error", it is "an error while executing target 'memory'" (target 'memory', inside PW/Makefile, builds a memory estimator code). Either there is another error, or your system is configured to stop on warnings. > someone compiled pw.1.3.0 with ifc 7.* ? I have. Paolo From sahu at matter3.ph.utexas.edu Fri Nov 21 17:59:01 2003 From: sahu at matter3.ph.utexas.edu (Dr. B.R.Sahu) Date: Fri, 21 Nov 2003 10:59:01 -0600 (CST) Subject: [Pw_forum] berry phase Message-ID: Dear pwscf users while doing the BP calculation(after the scf calculation) for the rhombohedral GeTe, I get the error. ================================================== POLARIZATION CALCULATION -------------------------------------------------- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% from c_phase : error # 1 Wrong k-strings weights? %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% stopping ... the initial k-point grid for the scf calculation is: K_POINTS {automatic} 4 4 4 1 1 1 whereas berry phase for the G-dir=1, the nppstr =7 and the k-point grid was: K_POINTS {automatic} 7 4 4 1 1 1 (being the low symmetry structure, I have to do G-dir seperately and then sum the numbers). It seems the error is from(bp_c_phase.f90) IF (ABS(wk(kindex)-wk(kindex-1)-dk(1)) > eps) & CALL errore('c_phase','Wrong k-strings weights?',1) and eps is set to eps=1.0E-6_dp in this file Pl.let me know whether I am doing something wrong? regards sahu From elwira at fy.chalmers.se Tue Nov 25 16:34:24 2003 From: elwira at fy.chalmers.se (Elwira Wachowicz) Date: Tue, 25 Nov 2003 16:34:24 +0100 Subject: [Pw_forum] DOS definition Message-ID: <3FC37680.8070101@fy.chalmers.se> Hello, I calculated phonon DOS using tetrahedron method implemented in pwscf 1.2.0 for cubic system. Could you tell me how DOS is defined down there? I mean what's the norm? Thank you for your help. Best regards, Elwira Wachowicz From neuber at itp.tu-graz.ac.at Wed Nov 26 10:22:58 2003 From: neuber at itp.tu-graz.ac.at (Danilo Neuber) Date: Wed, 26 Nov 2003 10:22:58 +0100 (CET) Subject: [Pw_forum] Si band gap calculated with different functionals Message-ID: Dear PWSCF community! As an exercise in PP generation, I have created Si pseudopotentials with Paolo Giannozzi's PP programs. Using the ground state [Ne] 3s2 3p2 for the s- and p-channels and [Ne] 3s1.25 3p0.75 3d0.25 (taken from the BHS-paper PRB v26, p. 4199) for the d-channel of the pseudopotential and accepting the suggested values for the matching radii r_c (without nonlinear core correction), I created PPs for the functionals LDA, BLYP and PBE. (I adapted the example shell scripts from the PP and the pwscf program package for PP generation and band structure calculation.) The Si band gaps obtained with the PBE (E_g = 0.585 eV) and the LDA-functional (E_g = 0.505 eV) are in very good agreement with the results I got with the PPs Si.pbe-rrkj.UPF (E_g = 0.577 eV) and Si.pz-vbc.UPF (E_g = 0.506 eV) from the pwscf website; however, for BLYP I got a Si band gap of E_g = 0.855 eV. This value seems too good to be true, could this be a problem in the implementation of the functional? Thank you in advance for any comments! With kind regards, Danilo R. Neuber -- Dipl.-Ing. Danilo Neuber Technische Universit?t Graz Telefon: +43 / 316 / 873 - 8188 Institut f?r Theoretische Physik Telefax: +43 / 316 / 873 - 8678 Petersgasse 16, A-8010 Graz From giannozz at nest.sns.it Sat Nov 29 16:48:47 2003 From: giannozz at nest.sns.it (Paolo Giannozzi) Date: Sat, 29 Nov 2003 16:48:47 +0100 Subject: [Pw_forum] Si band gap calculated with different functionals In-Reply-To: References: Message-ID: <200311291648.47211.giannozz@nest.sns.it> On Wednesday 26 November 2003 10:22, Danilo Neuber wrote: > The Si band gaps obtained with the PBE (E_g = 0.585 eV) and the > LDA-functional (E_g = 0.505 eV) are in very good agreement with the > results I got with the PPs Si.pbe-rrkj.UPF (E_g = 0.577 eV) and > Si.pz-vbc.UPF (E_g = 0.506 eV) from the pwscf website; however, for BLYP I > got a Si band gap of E_g = 0.855 eV. is the lattice parameter - the same for all cases, and equal to the experimental one - the theoretical one for each case, and they are very similar - the theoretical one for each case, but they are not very similar - none of the above ? > could this be a problem in the implementation of the functional? it could also be a PP problem. Don't assume that parameters that are good for the generation of a PP with a given exchange-correlation functional are automatically good for all other exchange-correlation functionals Paolo -- Paolo Giannozzi e-mail: giannozz at nest.sns.it Scuola Normale Superiore Phone: +39/050509412 Piazza dei Cavalieri 7 Fax: +39/050509417, 050563513 I-56126 Pisa, Italy Office: Lab. NEST, Via della Faggiola 19