From ZRLSRB at ch.ibm.com Tue Jul 1 18:40:16 2003 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Tue, 1 Jul 2003 18:40:16 +0200 Subject: [CPMD-list] Help on a simple question! Message-ID: Dear Zhou, the short answer is: 219474.6314 cm^(-1)/Hartree. Since this answer is not much more useful than 42, here a few hints: - Vibrational analysis in CPMD returns the frequencies in cm^(-1) as well. - You'll find typical frequencies in many books. Usually, they are listed in cm^(-1). - If you need to convert between units, have a look here: http://physics.nist.gov/cuu/Constants/index.html For this conversion, you would have looked up the Hartree energy, Planck's constant, and the speed of light in vacuum. A note about the link to NIST's database of fundamental constants: it contains the updated values of the latest values CODATA agreed on. The last update was 1998. It's a good idea to look up the values there, since even newer textbooks (and some articles, too) tend to stick for some unknown reasons to the 1979 recommended values. Some of the numbers changed quite considerably since then. Best regards, Salomon From wyin at mail.unomaha.edu Tue Jul 1 23:58:08 2003 From: wyin at mail.unomaha.edu (Weiguo Yin) Date: Tue, 1 Jul 2003 16:58:08 -0500 Subject: [CPMD-list] TH Message-ID: Dear all, >as far as i remember the only advice for this problem on an SGI so >far was to compile for 32 bit, i.e. get rid of -DPOINTER8, >use -n32 instead of -64, and recompile everything. > >axel kohlmeyer. It works for cpmd-test/std/Si64. Thanks to Axel kohlmeyer. My SGI (CPU: MIPS R12000 + FPU: MIPS R12010) did not crash when SPLINE POINTS were set 500,1000,3000,5000, and 10000. Besides, neither using v3.5.4 nor compilation without -r8 has significant impact on the speed of this simple test. **************************** ** FURTHER QUESTION ** **************************** Why are "THE NEW SIZE OF THE PROGRAM" and "MEMORY" obtained in the test on SGI much larger than those given in cpmd-test/std/Si64/out-1 and out-2? Please refer to the following data: -------------------------- cpmd-test/std/Si64/inp-1 ------------------------ [My result, 2 processors, -n32 -r8] *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 76611 kBYTES *** [PEAK NUMBER 83] PEAK MEMORY 4806098 = 38.4 MBytes [cpmd-test/std/Si64/out-1, 32 processors] *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 8080 KBytes *** [PEAK NUMBER 82] PEAK MEMORY 683044 = 5.5 MBytes -------------------------cpmd-test/std/Si64/inp-2--------------------------- [My result, 2 processors -n32 -r8] *** GMOPTS| THE NEW SIZE OF THE PROGRAM IS 67846 kBYTES *** *** PHFAC| THE NEW SIZE OF THE PROGRAM IS 70139 kBYTES *** [PEAK NUMBER 95] PEAK MEMORY 4185128 = 33.5 MBytes [cpmd-test/std/Si64/out-2, 32 processors] *** GMOPTS| THE NEW SIZE OF THE PROGRAM IS 5324 KBytes *** *** PHFAC| THE NEW SIZE OF THE PROGRAM IS 5524 KBytes *** [PEAK NUMBER 95] PEAK MEMORY 629261 = 5.0 MBytes It appears that CPMD runs on the SGI ORIGIN platform are inefficient. Could somebody tell me how to tune CPMD? Thanks in advance. Weiguo Yin From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Wed Jul 2 05:28:19 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Wed, 02 Jul 2003 05:28:19 +0200 Subject: [CPMD-list] Re: TH In-Reply-To: Message-ID: <200307020328.h623SJD14272@yello.theochem.ruhr-uni-bochum.de> On Tue, 1 Jul 2003 16:58:08 -0500 "Weiguo Yin" wrote: > > Dear all, > > >as far as i remember the only advice for this problem on an SGI so > >far was to compile for 32 bit, i.e. get rid of -DPOINTER8, > >use -n32 instead of -64, and recompile everything. > > > > It works for cpmd-test/std/Si64. Thanks to Axel kohlmeyer. My SGI (CPU: > MIPS R12000 + FPU: MIPS R12010) did not crash when SPLINE POINTS were set hi, my interest was in varying 'SPLINE POINTS' _without_ recompiling. > 500,1000,3000,5000, and 10000. Besides, neither using v3.5.4 nor > compilation without -r8 has significant impact on the speed of this simple > test. that was on a linux pc, which most likely allocates memory differently, so you won't necessarily see an effect. i suspect a silent memory corruption by crossing boundaries of dynamically allocted memory blocks. > > **************************** > ** FURTHER QUESTION ** > **************************** > Why are "THE NEW SIZE OF THE PROGRAM" and "MEMORY" obtained in the test on > SGI much larger than those given in cpmd-test/std/Si64/out-1 and out-2? > Please refer to the following data: because the example job was run on a parallel machine with 32 cpus. cpmd uses a distributed memory parallelization. have a closer at the PARAPARAPARAPARAPARA section. it tells you how the job is distributed over the nodes. > > -------------------------- cpmd-test/std/Si64/inp-1 > ------------------------ > [My result, 2 processors, -n32 -r8] > *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 76611 kBYTES *** > [PEAK NUMBER 83] PEAK MEMORY 4806098 = 38.4 MBytes > > [cpmd-test/std/Si64/out-1, 32 processors] > *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 8080 KBytes *** > [PEAK NUMBER 82] PEAK MEMORY 683044 = 5.5 MBytes > > -------------------------cpmd-test/std/Si64/inp-2--------------------------- > [My result, 2 processors -n32 -r8] > *** GMOPTS| THE NEW SIZE OF THE PROGRAM IS 67846 kBYTES *** > *** PHFAC| THE NEW SIZE OF THE PROGRAM IS 70139 kBYTES *** > [PEAK NUMBER 95] PEAK MEMORY 4185128 = 33.5 MBytes > > [cpmd-test/std/Si64/out-2, 32 processors] > *** GMOPTS| THE NEW SIZE OF THE PROGRAM IS 5324 KBytes *** > *** PHFAC| THE NEW SIZE OF THE PROGRAM IS 5524 KBytes *** > [PEAK NUMBER 95] PEAK MEMORY 629261 = 5.0 MBytes > > It appears that CPMD runs on the SGI ORIGIN platform are inefficient. Could > somebody tell me how to tune CPMD? Thanks in advance. you cannot talk about efficieny with such small runs (look at the cutoff!). for bigger jobs memory usage seems pretty invariant between platforms. this is to be expected, since most of the memory is needed for real*8 data. regards, axel kohlmeyer. > > Weiguo Yin > > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From bala at jncasr.ac.in Wed Jul 2 06:37:45 2003 From: bala at jncasr.ac.in (Dr. Balasubramanian Sundaram) Date: Wed, 02 Jul 2003 10:07:45 +0530 Subject: [CPMD-list] USPP Message-ID: <3F026199.C704E67D@jncasr.ac.in> Hi All, We are trying to learn the ropes on the construction and usage of PPs. Specifically, we are interested in constructing the Vanderbilt USPP to study CO_{2}, with the BLYP exchange/correlation functional. Can we mix the PPs of the elements? i.e., Can we use ultrasoft PPs (with BLYP) for oxygen, and C_MT_GIA_BLYP (that came with the CPMD PPLIB) for carbon? I realize that this question has some doubts on the USPP code as well, but I do not know of any mailing list for that. Hope this cross-posting in the cpmd-list is ok. Thanks! Bala From Ari.P.Seitsonen at iki.fi Wed Jul 2 13:27:25 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Wed, 2 Jul 2003 13:27:25 +0200 Subject: [CPMD-list] USPP In-Reply-To: <3F026199.C704E67D@jncasr.ac.in> (bala@jncasr.ac.in) References: <3F026199.C704E67D@jncasr.ac.in> Message-ID: <200307021127.h62BRPa32107@magadino.cscs.ch> Dear Bala, > We are trying to learn the ropes on the construction and usage of > PPs. Specifically, we are interested in constructing the Vanderbilt > USPP to study CO_{2}, with the BLYP exchange/correlation functional. > > Can we mix the PPs of the elements? i.e., Can we use ultrasoft > PPs (with BLYP) for oxygen, and C_MT_GIA_BLYP (that came with the CPMD > PPLIB) for carbon? > > I realize that this question has some doubts on the USPP code as well, > but I do not know of any mailing list for that. Hope this cross-posting > in the cpmd-list is ok. One can freely mix the different kinds of pseudo potentials. Greetings from cool Zurich, apsi -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From bala at jncasr.ac.in Thu Jul 3 13:06:19 2003 From: bala at jncasr.ac.in (Dr. Balasubramanian Sundaram) Date: Thu, 03 Jul 2003 16:36:19 +0530 Subject: [CPMD-list] USPP: H_{2}O References: <3F026199.C704E67D@jncasr.ac.in> <200307021127.h62BRPa32107@magadino.cscs.ch> Message-ID: <3F040E2B.A3FA5984@jncasr.ac.in> Dear All, We are able to optimize geometry for water monomer as well as water dimer system, using: a) LDA based functionals that came with Vanderbilt's USPP code b) BLYP based functionals that came with the CPMD PP Library. We wanted to gain experience in using USPP with BLYP. We generated (we beleive so!) the USPP's for oxygen and hydrogen using BLYP exchange/correlations, and used these PPs with CPMD to optimize the geometry of one water molecule. The initial configuration of the molecule is very close to the expected final configuration. We have tried several combinations of optimization schemes, but we are unable to converge the wavefunction to anything better than 7.0E-4 (GEMAX). We tried even CONVERGENCE RELAX {5} option. I noticed that Dr. Christian Tuma too had mentioned this in a mail of 30 Oct. 2002 (Item # 7). I have attached the input file below. I would appreciate if you could point out any mistakes that we are committing. The other possibility could be that there is something wrong in the way we are constructing the USPP with BLYP. Many thanks! Best, Bala ----------------------------------- &CPMD OPTIMIZE GEOMETRY COMBINED MAXSTEP 5000 STORE 20 TIMESTEP 5.0 CONVERGENCE GEOMETRY 1.0D-4 CONVERGENCE RELAX 5 ISOLATED MOLECULE INITIALIZE WAVEFUNCTION RANDOM STRUCTURE BONDS ANGLES &END &SYSTEM ANGSTROM CELL 10.0 1.0 1.0 0.0 0.0 0.0 SYMMETRY 1 CUTOFF 25.0 &END &DFT FUNCTIONAL BLYP &END &ATOMS *001-H-gbl--sb.uspp BINARY LMAX=S 2 -0.7731 0.0000 0.0000 1 0 0.7731 0.0000 0.0000 1 0 *008-O-gbl--sb.uspp BINARY LMAX=P 1 0.0000 0.5953 0.0000 8 0 &END ------------------------------------------------------------------------ From ct at chemie.hu-berlin.de Thu Jul 3 16:35:24 2003 From: ct at chemie.hu-berlin.de (Christian Tuma) Date: Thu, 3 Jul 2003 16:35:24 +0200 Subject: [CPMD-list] USPP: H_{2}O In-Reply-To: <3F040E2B.A3FA5984@jncasr.ac.in> References: <3F026199.C704E67D@jncasr.ac.in> <200307021127.h62BRPa32107@magadino.cscs.ch> <3F040E2B.A3FA5984@jncasr.ac.in> Message-ID: <20030703143524.GC81396@uranus.chemie.hu-berlin.de> Hello, I observed the problem of poor convergence of the wavefunction in BLYP calculations on isolated systems as well. Prof Hutter told me that this may be due to Beckes gradient correction that involves the calculation of \nabla \rho / \rho^(4/3). If both nominator and denominator go close to zero the whole expression gets vague and maybe heavily perturbed by numerical noise which is always present in plane wave calculations. That's why you could try to increase the parameter GC_CUTOFF as recommended in the manual - especially when using USPPs. Good luck, Christian. On Thu, Jul 03, 2003 at 04:36:19PM +0530, Dr. Balasubramanian Sundaram wrote: ... > We have tried several combinations of optimization schemes, but we are > unable to converge the wavefunction to anything better than 7.0E-4 > (GEMAX). > We tried even CONVERGENCE RELAX {5} option. ... > &DFT > FUNCTIONAL BLYP > &END ... > -- Christian Tuma Humboldt-Universitaet Berlin ct at chemie.hu-berlin.de Arbeitsgruppe Quantenchemie (Prof. Sauer) phone: +49-30-20937140 Brook-Taylor-Str. 2, 12489 Berlin, GERMANY fax: +49-30-20937136 http://www.chemie.hu-berlin.de/ag_sauer From Henning.Heiberg-Andersen at fi.uib.no Fri Jul 4 10:32:01 2003 From: Henning.Heiberg-Andersen at fi.uib.no (Henning Heiberg-Andersen) Date: Fri, 4 Jul 2003 10:32:01 +0200 (METDST) Subject: [CPMD-list] Recommendable visualization software? Message-ID: Hi, I am looking for software for visualizing both static and dynamic CPMD results, and would be grateful for your recommendation(s). Best regards, Henning Heiberg-Andersen From vb at chemie.hu-berlin.de Fri Jul 4 13:11:18 2003 From: vb at chemie.hu-berlin.de (Veronika Brazdova) Date: Fri, 4 Jul 2003 12:11:18 +0100 Subject: [CPMD-list] Recommendable visualization software? In-Reply-To: References: Message-ID: <20030704111118.GB14306@neptun.chemie.hu-berlin.de> Hi Henning, I use Molekel (http://www.cscs.ch/molekel) for visualising results of vibrational analysis. It can't handle periodic boundary conditions, so that it will display only atoms which are explicitly specified in the input file, but for vibrational analysis this does not matter. It is a free software. It can also display changes in structure (e.g. from an MD simulation), I think. WebLab Viewer Pro 3.7 used to be distributed for free by Accelrys, and it is a good tool for visualising crystals (it can read in a .car file, but not a .xtl file). I am not sure whether it is still available for download. I put online a couple of scripts which I wrote for processing CPMD output. You can download them from http://atrey.karlin.mff.cuni.cz/~verunka/CPMD. The script cpmd2mkl.c is for CPMD -> Molekel conversion, cpmd2car for creating .car files, etc. The file scripts.pdf contains documentation. I know there is a nice tool called gOpenMol. If anyone has a script which would produce a gOpenMol-readable file, I'd be glad to have it. Best regards Veronika Brazdova On Fri, Jul 04, 2003 at 10:32:01AM +0200, Henning Heiberg-Andersen wrote: > > Hi, > > I am looking for software for visualizing both static and dynamic > CPMD results, and would be grateful for your recommendation(s). > > Best regards, > > Henning Heiberg-Andersen > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list -- _______________________________________________________________________________ Veronika Brazdova Humboldt University, Institute of Chemistry, Brook-Taylor-Strasse 2, 12489, Berlin vb at chemie.hu-berlin.de http://atrey.karlin.mff.cuni.cz/~verunka phone: +49 30 2093 7140 fax: +49 03 2093 7136 _______________________________________________________________________________ From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Fri Jul 4 14:11:29 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Fri, 04 Jul 2003 14:11:29 +0200 Subject: [CPMD-list] Recommendable visualization software? In-Reply-To: Message-ID: <200307041211.h64CBUj11485@yello.theochem.ruhr-uni-bochum.de> On Fri, 4 Jul 2003 10:32:01 +0200 (METDST) Henning Heiberg-Andersen wrote: > > Hi, > > I am looking for software for visualizing both static and dynamic > CPMD results, and would be grateful for your recommendation(s). hi henning, there are many ways to visualize the cpmd output. programs that are used for that purpose here in bochum include: Molden: gOpenMol: Molekel: VMD: generally, you want to convert your trajectories into xyz movies so that you can read them easily with those programs. you can use the TRAJECTORY XYZ flag to get such a file directly or use the attached traj2xyz.pl perl script to generate a xyz movie from a GEOMETRY.xyz and a TRAJECTORY file. for geometry optimizations, there is a similar XYZ flag, to get a GEO_OPT.xyz file. for densities, electrostatic potentials and wannier functions you can use the cpmd2cube utility to create files in the gaussian cube format, which can be imported into the programs named above. for frequency calculations you can use the GAUSS suboption to VIBRATIONAL ANALYSIS which will create the file VIB1.log (and optionally VIB2.log) which look like part of a gaussian frequency calculation and can be read by all programs, that read (and understand) gaussian log files, e.g. Molden and Molekel. hope this helps, axel kohlmeyer. > > Best regards, > > Henning Heiberg-Andersen > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/octet-stream Size: 5842 bytes Desc: traj2xyz.pl Url : http://cpmd.org/pipermail/cpmd-list/attachments/20030704/f255fb03/attachment.obj From audrius.alkauskas at unibas.ch Sat Jul 5 15:31:06 2003 From: audrius.alkauskas at unibas.ch (Audrius Alkauskas) Date: Sat, 05 Jul 2003 15:31:06 +0200 Subject: [CPMD-list] Parallelization on k-points References: <200307051001.MAA33390@cpmd.org> Message-ID: <3F06D31A.4010201@unibas.ch> Hello We have a new cluster here in Basel, and I began running big jobs of Ag(111) surface. I noticed that each k-point is done in serial. Would it make a difference to do each of, let's say, 8 different k-points on a different node? Is it difficult to implement? Would there be any time economy or not? Audrius Alkauskas From Ari.P.Seitsonen at iki.fi Sat Jul 5 20:44:15 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Sat, 5 Jul 2003 20:44:15 +0200 Subject: [CPMD-list] Parallelization on k-points In-Reply-To: <3F06D31A.4010201@unibas.ch> (message from Audrius Alkauskas on Sat, 05 Jul 2003 15:31:06 +0200) References: <200307051001.MAA33390@cpmd.org> <3F06D31A.4010201@unibas.ch> Message-ID: <200307051844.h65IiFK19697@magadino.cscs.ch> Dear Audrius, > We have a new cluster here in Basel, and I began running big jobs of > Ag(111) surface. I noticed that each k-point is done in serial. Would it > make a difference to do each of, let's say, 8 different k-points on a > different node? Is it difficult to implement? Would there be any time > economy or not? The k point parallellisation is/would be indeed _very_ efficient as the k points are very independent. It wouldn't be too much of work, I believe (I did it once in a couple of days, yet on an smaller code, once one knows the code it probably wouldn't take even that long. One important issue is of course I/O, whether to write the restart files in parallel, only on a single node etc). Exhausted Greetings from again-sunny Zurich, apsi -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From hutter at pci.unizh.ch Sun Jul 6 18:51:03 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sun, 6 Jul 2003 18:51:03 +0200 (MEST) Subject: [CPMD-list] Parallelization on k-points In-Reply-To: <200307051844.h65IiFK19697@magadino.cscs.ch> References: <200307051001.MAA33390@cpmd.org> <3F06D31A.4010201@unibas.ch> <200307051844.h65IiFK19697@magadino.cscs.ch> Message-ID: Hi in general a parallelisation over k-points is not too difficult. However, in CPMD this would require some work. Especially the FFT parallelisation would have to be generalised. Using Fortran90 this could be done using modular programming and then a combined k-Point/G-vector parallelisation would be possible. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Sat, 5 Jul 2003 Ari.P.Seitsonen at iki.fi wrote: > > Dear Audrius, > > > We have a new cluster here in Basel, and I began running big jobs of > > Ag(111) surface. I noticed that each k-point is done in serial. Would it > > make a difference to do each of, let's say, 8 different k-points on a > > different node? Is it difficult to implement? Would there be any time > > economy or not? > > The k point parallellisation is/would be indeed _very_ efficient as > the k points are very independent. It wouldn't be too much of work, I > believe (I did it once in a couple of days, yet on an smaller code, > once one knows the code it probably wouldn't take even that long. One > important issue is of course I/O, whether to write the restart files > in parallel, only on a single node etc). > > Exhausted Greetings from again-sunny Zurich, > > apsi > > -- > -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- > Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ > Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 > Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) > Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich > Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From davide.provasi at mi.infn.it Mon Jul 7 12:06:18 2003 From: davide.provasi at mi.infn.it (Davide Provasi) Date: 07 Jul 2003 12:06:18 +0200 Subject: [CPMD-list] Performance on beowulf cluster Message-ID: <1057572378.6721.6.camel@caronte.mi.infn.it> I am trying to optimize the preformance of CPMD (version 3.7.1) on a beowulf cluster. The code is compiled with IFC Compiler (Version 7.1, Build 20030521Z),Libraries: ATLAS 3.4.1, LAPACK 3.0+updates, optimized for Pentium 4 My makefile contains the following flags: > FFLAGS = -xiMK -pc64 -mp1 -ip -unroll -cm -w90 -w95 > -I/home/libs/mpich-1.2.5/include -safe_cray_ptr > LFLAGS = -xiMK -Vaxlib -L/home/libs/ATLAS/Linux_P4/lib -lcblas -lf77blas > -latlas -L/home/libs/LAPACK/libs -llapackP4 > -L/home/libs/ATLAS/Linux_P4/lib -lcblas -lf77blas -latlas > -L/home/libs/mpich-1.2.5/lib -lmpich -safe_cray_ptr > CFLAGS = > CPP = /lib/cpp -P -C -traditional > CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC -DPARALLEL > -DMP_LIBRARY=__MPI > CC = gcc -O3 -Wall > FC = ifc -c -tpp7 -O3 > #-safe_cray_ptr > LD = ifc -tpp7 -O3 We have noticed a large Elapsed time in the FFTCOM, and were wondering whether we could improve this. **************************************************************** * * * TIMING * * * **************************************************************** SUBROUTINE CALLS CPU TIME ELAPSED TIME FFT-G/S 19040 291.46 292.76 S_INVFFT 4736 253.78 253.85 RHOOFR 54 103.04 102.86 S_FWFFT 1280 69.84 69.65 RGS 54 64.93 64.94 VPSI 114 43.68 44.02 EICALC 55 41.43 41.39 FFTCOM 6512 32.04 1756.83 FWFFT 275 25.34 25.28 GCENER 55 22.26 22.27 OVLAP 19 21.67 21.72 RNLSM1 150 21.02 21.11 INVFFT 221 18.21 18.26 FNONLOC 114 12.58 12.59 ROTATE 18 11.37 11.37 PCGRAD 18 10.21 10.54 GRADEN 55 7.76 7.94 PHASE 496 6.77 6.94 OVLAP2 96 5.00 5.82 XCENER 55 4.80 4.74 FORCES 18 3.30 3.37 ---------------------------------------------------------------- TOTAL TIME 1070.49 2798.24 ---------------------------------------------------------------- thank you in advance, Davide ----------------------------- Davide Provasi University of Milano, Physics Department via Celoria 16, 20133 Milano Phone +39 02 5031 7 717 ----------------------------- From ct at chemie.hu-berlin.de Mon Jul 7 12:47:08 2003 From: ct at chemie.hu-berlin.de (Christian Tuma) Date: Mon, 7 Jul 2003 12:47:08 +0200 Subject: [CPMD-list] Performance on beowulf cluster In-Reply-To: <1057572378.6721.6.camel@caronte.mi.infn.it> References: <1057572378.6721.6.camel@caronte.mi.infn.it> Message-ID: <20030707104708.GB86767@uranus.chemie.hu-berlin.de> Hello, your problem may be hardware related. What kind of network between the nodes is used? For CPMD fast communication between the nodes is essential, in our group we observed the same problem when using e.g. Fast Ethernet which was totally insufficient. Even if you use a much better network hardware the chipset on the motherboard where the network is connected to has to show the corresponding performance features. You should have a look at all this. Finally, you may switch from MPICH to LAM/MPI to get a little better, but, however, the substantial hardware problem cannot be fixed this way. Good luck, Christian. -- Christian Tuma Humboldt-Universitaet Berlin ct at chemie.hu-berlin.de Arbeitsgruppe Quantenchemie (Prof. Sauer) phone: +49-30-20937140 Brook-Taylor-Str. 2, 12489 Berlin, GERMANY fax: +49-30-20937136 http://www.chemie.hu-berlin.de/ag_sauer From chimaju at magnum.barc.ernet.in Tue Jul 8 05:09:41 2003 From: chimaju at magnum.barc.ernet.in (Chiranjib Majumder) Date: Tue, 8 Jul 2003 08:39:41 +0530 (IST) Subject: [CPMD-list] Bulk cohesive energy In-Reply-To: <200210231447.g9NElqN08571@magadino.cscs.ch> Message-ID: Dear Ari I shall appreciate to know how one should calculated the cohesive energy of a bulk metal using CPMD code. Thanking in anticipation of your cooperation. Chiranjib Majumder ***************************************************************** * Dr. Chiranjib Majumder * * Scientific Officer * * Chemistry Division * * Bhabha Atomic Research Centre, * * Trombay, Mumbai-400 085,INDIA * * Off:091-22-5592081 * * Res:91-22-4555531 * * E-Mail :chimaju at magnum.barc.ernet.in * * * ***************************************************************** From ekasciut at unity.ncsu.edu Tue Jul 8 22:26:15 2003 From: ekasciut at unity.ncsu.edu (Eliana Asciutto) Date: Tue, 08 Jul 2003 16:26:15 -0400 Subject: [CPMD-list] water parameters Message-ID: <3F0B28E7.7060401@unity.ncsu.edu> Hi everyone, Does anybody know how to generate water parameters in "psf" and "pro or param" formats? Thanks, Eliana Asciutto ekasciut at unity.ncsu.edu From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Tue Jul 8 23:15:46 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Tue, 08 Jul 2003 23:15:46 +0200 Subject: [CPMD-list] water parameters In-Reply-To: <3F0B28E7.7060401@unity.ncsu.edu> Message-ID: <200307082115.h68LFke22141@yello.theochem.ruhr-uni-bochum.de> On Tue, 08 Jul 2003 16:26:15 -0400 Eliana Asciutto wrote: > Hi everyone, > > Does anybody know how to generate water parameters in "psf" and "pro or > param" formats? > Thanks, hi eliana, could you please elaborate a little bit. unless i miss my guess, files with these names are used in classical molecular dynamics or molecular modelling programs like charmm, amber etc. there is a wealth of documentation on these programs and this mailing list would be not the right place to ask questions about these programs (they probably have their own mailing list). ... or do you want to use cpmd to create a new water potential (as if there are not already far to many of them)? cheers, axel kohlmeyer. > > Eliana Asciutto > ekasciut at unity.ncsu.edu > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From M.A.Zwijnenburg at tnw.tudelft.nl Fri Jul 11 10:54:56 2003 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Fri, 11 Jul 2003 10:54:56 +0200 Subject: [CPMD-list] NVT Message-ID: <3F0E977E.8828.4FB4925@localhost> Hi, I'm doing a NVT md run on a small cluster (see for inputfile below) and have some questions. 1) In my md run EHAM is not constant but rather varies from the third decimal on. However, since this run employs an NVT ensemble rather then an NVE I wonder if EHAM should still be conserved and thus stay constant? 2) The systems seems to cool down over time. I set the thermostat temperature at 1000K but the electron temperature cools down, after a small overshoot, to allready 600K after 200 steps. I guess this means my thermostat is not working properly? Cheers, Martijn Inputfile: ! Si4O8 cluster MD ! &CPMD MOLECULAR DYNAMICS STRUCTURE BONDS ANGLES DIHEDRALS CONVERGENCE 1E-6 5E-6 1. ISOLATED MOLECULE HESSIAN UNIT RESTART WAVEFUNCTION GEOMETRY LATEST NOSE IONS 1000 800 TIMESTEP 5.0 EMASS 700 TRAJECTORY XYZ &END &SYSTEM SYMMETRY 0 ANGSTROM CELL 15 1.0 1.0 0 0 0 ANGSTROM CUTOFF 30.00 POINT GROUP AUTO &END &ATOMS *Si_ps.uspp BINARY new.f LMAX=P 4 0.000000000000 0.000000000000 0.000000000000 0.471809914327 0.000000000000 3.182682023275 3.653225710458 -0.034526749450 2.711038719844 3.181409275132 -0.034559479116 -0.471644591443 *o_ps.uspp BINARY new.f LMAX=P 8 0.000000000000 0.000000000000 1.626292446278 2.098080331254 -0.016336484977 3.186832736955 3.653206618118 -0.034138170947 1.084746035991 1.555131813001 -0.018896584142 -0.475808582925 4.871737465360 -0.049224011432 3.612939458029 -0.432765838343 0.012873808042 4.399223302846 -1.218511827443 0.015066749685 -0.901895052361 4.086009030180 -0.046817197390 -1.688173543327 &END &DFT FUNCTIONAL PW91 GC-CUTOFF 5E-5 &END ---------------------------------------------------------------------- --- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Mon Jul 14 11:13:56 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Mon, 14 Jul 2003 11:13:56 +0200 Subject: [CPMD-list] benchmark data for intel itanium and amd opteron Message-ID: <200307140913.h6E9Du820949@yello.theochem.ruhr-uni-bochum.de> hello everybody! i am currently evaluating several 64-bit platforms to find one that will provide a good price-porformace ratio when running CPMD (for the type of problems we are running here in bochum). especially i am looking for benchmark data for running CPMD on intel itanium and amd opteron machines (single cpu, SMP, or distributed network does not matter). so if anyone can help me out, please send me your data or your experiences. please send the responses to me personally. if people are interested, i can post a summary here later. thanks in advance, axel kohlmeyer. p.s.: if anyone would be willing to run (or let me run) a few of my benchmarks this would be even better. -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From carlospi at estudiantes.fisica.unal.edu.co Mon Jul 14 23:09:16 2003 From: carlospi at estudiantes.fisica.unal.edu.co (Pinilla Castellanos Carlos) Date: Mon, 14 Jul 2003 16:09:16 -0500 (COT) Subject: [CPMD-list] Question Message-ID: In the Kohn-Sham energies, I have the following output file ........................................... Kohn-Sham energies 273 -2.3933041NC 2.000 274 -2.2976590NC 2.000 275 -2.1840810NC 2.000 276 -2.1469507NC 2.000 277 -1.9491218NC 2.000 278 -1.7301004NC 2.000 279 -1.3792735NC 2.000 280 -1.3078462NC 2.000 281 -1.1192683NC 2.000 282 -1.0102267NC 2.000 283 -0.6190014NC 2.000 284 -0.4036707NC 2.000 285 -0.3121118NC 2.000 286 -0.1649033NC 2.000 287 0.0153771NC 2.000 288 0.1742432NC 2.000 289 0.2419758NC 0.000 CHEMICAL POTENTIAL = 0.1742423776 E What is NC into the values of energy. The best wishes cpmd-list and thanks for your help Carlos From hutter at pci.unizh.ch Tue Jul 15 08:54:05 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 15 Jul 2003 08:54:05 +0200 (MEST) Subject: [CPMD-list] NVT In-Reply-To: <3F0E977E.8828.4FB4925@localhost> References: <3F0E977E.8828.4FB4925@localhost> Message-ID: Hi On Fri, 11 Jul 2003, Martijn Zwijnenburg wrote: > Hi, > > I'm doing a NVT md run on a small cluster (see for inputfile below) > and have some questions. > > 1) In my md run EHAM is not constant but rather varies from the third > decimal on. However, since this run employs an NVT ensemble rather > then an NVE I wonder if EHAM should still be conserved and thus stay > constant? EHAM is still a constant of motion. If it's not constant you have to change the integration procedure. If the system without the thermostat is integrated without problems, then you have to change the parameters for the NOSE thermostats. Parameters 5 and 6 (Suzuki integrator and decomposition parameter). > > 2) The systems seems to cool down over time. I set the thermostat > temperature at 1000K but the electron temperature cools down, after a > small overshoot, to allready 600K after 200 steps. I guess this means > my thermostat is not working properly? > I don't get this. You have a thermostat at the ions only but you are talking about electron temperature. If you have a bad drift in EHAM the temperature of the ions and the kinetic energy of the electrons can vary a lot. However, the ionic temperature should be on average at the preset value. Juerg > Cheers, > > Martijn > > Inputfile: > > ! Si4O8 cluster MD > ! > &CPMD > MOLECULAR DYNAMICS > STRUCTURE BONDS ANGLES DIHEDRALS > CONVERGENCE > 1E-6 5E-6 1. > ISOLATED MOLECULE > HESSIAN UNIT > RESTART WAVEFUNCTION GEOMETRY LATEST > NOSE IONS > 1000 800 > TIMESTEP > 5.0 > EMASS > 700 > TRAJECTORY XYZ > &END > &SYSTEM > SYMMETRY > 0 > ANGSTROM > CELL > 15 1.0 1.0 0 0 0 > ANGSTROM > CUTOFF > 30.00 > POINT GROUP > AUTO > &END > &ATOMS > *Si_ps.uspp BINARY new.f > LMAX=P > 4 > 0.000000000000 0.000000000000 0.000000000000 > 0.471809914327 0.000000000000 3.182682023275 > 3.653225710458 -0.034526749450 2.711038719844 > 3.181409275132 -0.034559479116 -0.471644591443 > > *o_ps.uspp BINARY new.f > LMAX=P > 8 > 0.000000000000 0.000000000000 1.626292446278 > 2.098080331254 -0.016336484977 3.186832736955 > 3.653206618118 -0.034138170947 1.084746035991 > 1.555131813001 -0.018896584142 -0.475808582925 > 4.871737465360 -0.049224011432 3.612939458029 > -0.432765838343 0.012873808042 4.399223302846 > -1.218511827443 0.015066749685 -0.901895052361 > 4.086009030180 -0.046817197390 -1.688173543327 > > &END > &DFT > FUNCTIONAL PW91 > GC-CUTOFF > 5E-5 > &END > ---------------------------------------------------------------------- > --- > Martijn Zwijnenburg > Lab. of Applied Organic Chemistry and Catalysis > Delft University of Technology > Julianalaan 136 > 2628 BL Delft > The Netherlands > Tel: 0031-(0)152782691 > Fax: 0031-(0)152784700 > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > web page: http://come.to/tock > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Tue Jul 15 08:55:04 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 15 Jul 2003 08:55:04 +0200 (MEST) Subject: [CPMD-list] Question In-Reply-To: References: Message-ID: Hi NC after the Kohn-Sham eigenvalue stands for NOT CONVERGED. You should restart the calculation to fully converge these values. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Mon, 14 Jul 2003, Pinilla Castellanos Carlos wrote: > > In the Kohn-Sham energies, I have the following output file > > ........................................... > Kohn-Sham energies > 273 -2.3933041NC 2.000 274 -2.2976590NC 2.000 > 275 -2.1840810NC 2.000 276 -2.1469507NC 2.000 > 277 -1.9491218NC 2.000 278 -1.7301004NC 2.000 > 279 -1.3792735NC 2.000 280 -1.3078462NC 2.000 > 281 -1.1192683NC 2.000 282 -1.0102267NC 2.000 > 283 -0.6190014NC 2.000 284 -0.4036707NC 2.000 > 285 -0.3121118NC 2.000 286 -0.1649033NC 2.000 > 287 0.0153771NC 2.000 288 0.1742432NC 2.000 > 289 0.2419758NC 0.000 > CHEMICAL POTENTIAL = 0.1742423776 E > > What is NC into the values of energy. > > The best wishes cpmd-list and thanks for your help > > Carlos > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From M.A.Zwijnenburg at tnw.tudelft.nl Tue Jul 15 16:35:36 2003 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Tue, 15 Jul 2003 16:35:36 +0200 Subject: [CPMD-list] p4_error: : 14 Message-ID: <3F142D57.11649.119B81E@localhost> Hi, When I run the inputfile given below on our linux-cluster (cpmd 3.7.1 / compiled with PGI compiler/ MPICH) the MD job always crashes after the 205th step (independent of the precise nodes I'm running / or the number of nodes I'm running on). The error message seems to be precisely repeatable, so I checked if a file nears a limit (say 2GB) but everything is much much smaller. The precise error message (for 3 procs) is: 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 0.180 70.43 p1_15605: p4_error: : 14 p2_25632: p4_error: : 14 bm_list_27416: (20873.130096) wakeup_slave: unable to interrupt slave 0 pid 27415 Broken pipe Broken pipe and in case of the 2 proc job: 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 0.180 108.08 p1_23514: p4_error: : 14 Broken pipe The funny thing is that we never had such problems with CPMD optimization jobs, which ran for weeks, nor with any other program on the cluster (gamess-uk, dlpolly). It appears that MPI is the problem but does anybody have a solution? Gr. Martijn Inputfile: ! Si4O8 cluster MD ! &CPMD MOLECULAR DYNAMICS STRUCTURE BONDS ANGLES DIHEDRALS SPLINE RANGE 5.00 SPLINE POINTS 2500 CONVERGENCE 1E-6 5E-6 1. ISOLATED MOLECULE HESSIAN UNIT RESTART WAVEFUNCTION GEOMETRY LATEST TEMPERATURE 1000 TEMPCONTROL IONS 1000 20 TIMESTEP 5.0 EMASS 700 TRAJECTORY XYZ &END &SYSTEM SYMMETRY 0 ANGSTROM CELL 15 1.0 1.0 0 0 0 ANGSTROM CUTOFF 30.00 sTATES 36 POINT GROUP AUTO &END &ATOMS *Si_ps.uspp BINARY new.f LMAX=P 4 0.000000000000 0.000000000000 0.000000000000 0.471809914327 0.000000000000 3.182682023275 3.653225710458 -0.034526749450 2.711038719844 3.181409275132 -0.034559479116 -0.471644591443 *o_ps.uspp BINARY new.f LMAX=P 8 0.000000000000 0.000000000000 1.626292446278 2.098080331254 -0.016336484977 3.186832736955 3.653206618118 -0.034138170947 1.084746035991 1.555131813001 -0.018896584142 -0.475808582925 4.871737465360 -0.049224011432 3.612939458029 -0.432765838343 0.012873808042 4.399223302846 -1.218511827443 0.015066749685 -0.901895052361 4.086009030180 -0.046817197390 -1.688173543327 &END &DFT FUNCTIONAL PW91 GC-CUTOFF 5E-5 &END ---------------------------------------------------------------------- --- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From adrainzhou at yahoo.com.cn Tue Jul 15 17:11:05 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Tue, 15 Jul 2003 23:11:05 +0800 (CST) Subject: [CPMD-list] MD problem! Message-ID: <20030715151105.40902.qmail@web15302.mail.bjs.yahoo.com> Dear all, I met the following problem when performing MD calculation! 367 0.00112 16.6 -294.14844 -294.14379 -294.14267 0.009 26.34 368 0.00114 16.9 -294.14855 -294.14380 -294.14267 0.009 25.94 369 0.00115 17.3 -294.14866 -294.14382 -294.14267 0.009 24.05 370 0.00116 17.6 -294.14877 -294.14383 -294.14267 0.008 25.49 371 0.00118 18.0 -294.14889 -294.14384 -294.14267 0.008 26.80 372 0.00119 18.4 -294.14901 -294.14386 -294.14267 0.008 27.22 373 0.00121 18.8 -294.14913 -294.14387 -294.14267 0.008 25.35 374 0.00123 19.2 -294.14926 -294.14389 -294.14266 0.008 25.89 375 0.00124 19.5 -294.14938 -294.14390 -294.14266 0.008 27.72 376 0.00126 19.9 -294.14951 -294.14392 -294.14265 0.008 25.37 377 0.00130 20.3 -294.14964 -294.14394 -294.14264 0.008 25.12 378 0.00135 20.7 -294.14973 -294.14391 -294.14257 0.008 24.31 379 0.04902 21.0 -294.14982 -294.14392 -294.09491 0.008 26.99 380***************** -294.06582 3216785.58973************** 0.008 25.40 PROGRAM STOPS IN SUBROUTINE DSPEVY| FAILED TO DIAGONALIZE [PROC= 5] In the 379th steps, the electronic temperature becomes very high and the EHAM is not constant. So the timestep may be too long! But I perfom almost the same calculation with the same timestep and EMASS at the same system, the program runs okay! Moreover, the electron can follow the movement of Ion. Can anyone tell me what's wrong with it? Many thanks in advance! Zhou PROGRAM STOPS IN SUBROUTINE DSPEVY| FAILED TO DIAGONALIZE [PROC= 7] PROGRAM STOPS IN SUBROUTINE DSPEVY| FAILED TO DIAGONALIZE [PROC= 4] --------------------------------- Do You Yahoo!? ?????????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030715/e1e9978a/attachment.html From adrainzhou at yahoo.com.cn Tue Jul 15 17:22:35 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Tue, 15 Jul 2003 23:22:35 +0800 (CST) Subject: [CPMD-list] On PCG Minimize! Message-ID: <20030715152235.63074.qmail@web15311.mail.bjs.yahoo.com> Dear all, It seems there are some problems when using PCG Minimize to perform a multisteps wavefunction optimization. It is as follows, LINE SEARCH : LAMBDA=.103 ENERGY = -294.14310 228 3.006E-04 3.043E-06 -294.142940 -4.685E-06 43.33 LINE SEARCH : LAMBDA=.113E-01 ENERGY = -294.14314 229 5.253E-04 4.688E-06 -294.142920 2.063E-05 40.45 LINE SEARCH : LAMBDA=.116E-02 ENERGY = -294.14278 230 7.913E-04 9.229E-06 -294.142801 1.182E-04 40.08 LINE SEARCH : LAMBDA=.217E-03 ENERGY = -294.14273 231 8.708E-04 1.001E-05 -294.142741 6.047E-05 39.11 LINE SEARCH : LAMBDA=.407E-04 ENERGY = -294.14272 232 8.895E-04 1.086E-05 -294.142725 1.576E-05 40.98 LINE SEARCH : LAMBDA=.764E-05 ENERGY = -294.14272 233 8.932E-04 1.086E-05 -294.142722 3.149E-06 42.24 LINE SEARCH : LAMBDA=.143E-05 ENERGY = -294.14272 234 8.939E-04 1.086E-05 -294.142721 5.747E-07 39.37 ~ LAMBDA is less than 0.01, the convergence never meet! When I just stop at the 120th step, and then restart again. The convergence reached very fast! It only take another 20 steps. It is so strange! Can anyone tell me what's wrong with it? Many thanks in advance! Zhou --------------------------------- Do You Yahoo!? ?????????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030715/cc9136ea/attachment.html From adrainzhou at yahoo.com.cn Tue Jul 15 19:01:05 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Wed, 16 Jul 2003 01:01:05 +0800 (CST) Subject: [CPMD-list] MD problem! Message-ID: <20030715170105.64856.qmail@web15310.mail.bjs.yahoo.com> Dear all, Following the MD problems I put forward before, I choose another small step and perfrom MD calculation (no thermostat). The problem still met. 339 0.00120 30.8 -294.15250 -294.14388 -294.14268 0.011 20.55 340 0.00120 30.4 -294.15241 -294.14388 -294.14268 0.011 22.52 341 0.00120 30.1 -294.15232 -294.14388 -294.14268 0.011 20.17 342 0.00121 29.7 -294.15222 -294.14389 -294.14268 0.011 21.72 343 0.00121 29.4 -294.15212 -294.14389 -294.14268 0.011 21.73 344 0.00121 29.0 -294.15202 -294.14389 -294.14268 0.010 21.10 345 0.00122 28.6 -294.15192 -294.14390 -294.14268 0.010 21.93 346 0.00122 28.2 -294.15181 -294.14390 -294.14268 0.010 19.99 347 0.00123 27.8 -294.15171 -294.14390 -294.14268 0.010 19.55 348 0.00123 27.4 -294.15160 -294.14391 -294.14268 0.010 20.57 349 0.00124 27.0 -294.15149 -294.14391 -294.14267 0.010 20.38 350 0.00125 26.6 -294.15138 -294.14392 -294.14267 0.010 21.35 RESTART INFORMATION WRITTEN ON FILE ./RESTART.1 351 0.00125 26.2 -294.15126 -294.14392 -294.14267 0.010 20.84 352 0.00126 25.8 -294.15115 -294.14392 -294.14266 0.010 20.53 353 0.00195 25.3 -294.15103 -294.14393 -294.14198 0.010 20.71 354138.10491 49.1 -294.14967 -294.13592 -156.03100 0.010 23.11 The fictitious electronic kinetic energy become very stange at the 354th step. The timestep is very small! Can anybody tell me what is wrong with it? Many thanks! Zhou --------------------------------- Do You Yahoo!? ?????????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030716/3f2ead34/attachment.html From agapito at engr.sc.edu Tue Jul 15 23:01:10 2003 From: agapito at engr.sc.edu (Luis Agapito) Date: Tue, 15 Jul 2003 17:01:10 -0400 Subject: [CPMD-list] F90 linking problem Message-ID: <000001c34b14$38220a90$6f91fea9@MUNAY> Hello everybody, I'm installing CPMD v3.7.0 on a DEC-ALPHA machine using the MPI method for parallelizing. I got this error in the linking step, Am I missing any library? (maybe related with the lfmpi option for the F90), where could I get more information about that? Thanks nmr_position_p.o hardness_p.o respin_p.o response_p.o restart_p.o pert_kpoint_ p.o dnlpdk_p.o rnl_dk_p.o kpert_potential_p.o up3_p.o friesner_c_p.o matrix_p .o ks_ener_p.o kpert_util_p.o fukui_p.o kdp_diag.o kdp_prep.o kdp_rho.o kdp_stre ss_kin.o kdpoints.o rhoofr_kdp.o mm_detsp.o mm_init.o mm_dim.o mm_qmmm_forcedr .o mm_mdmain.o mm_coninp.o mm_cpmd_add_MM_forces_f77.o mm_cpmd_esp_charges_f77 .o mm_cpmd_ext_pot_f77.o vdwin.o vdw.o sysdepend.o shmemory.o -ldxml -lfmpi -l mpi -lrt -pthread ld: Can't locate file for: -lfmpi f90: Severe: Failed while trying to link. *** Exit 1 Stop. ------------------------------------------------------------------------ ---------- Luis A. Agapito Department of Electrical Engineering University of South Carolina Phone: 803-777-0720 Moletronics Group: http://www.ee.sc.edu/research/moletronics/ E-mail: agapito at engr.sc.edu -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030715/067e15a4/attachment.html From krauthau at ccm.udel.edu Wed Jul 16 18:34:02 2003 From: krauthau at ccm.udel.edu (Carl Krauthauser) Date: Wed, 16 Jul 2003 12:34:02 -0400 Subject: [CPMD-list] Problem linking with 3.7-1 Message-ID: <3F157E7A.7010506@ccm.udel.edu> Dear CPMD users. I am attempting to compile ver. 3.7-1 with the PC-IFC configuration. At the linking stage, I get the following error messages: > cpmd.o: In function `cpmd_': > cpmd.o(.text+0x51): undefined reference to `signal_' > cpmd.o(.text+0x65): undefined reference to `signal_' > envir.o: In function `envir_': > envir.o(.text+0x2f): undefined reference to `getenv_' > envir.o(.text+0x3e): undefined reference to `getcwd_' > envir.o(.text+0x46): undefined reference to `getpid_' > envir.o(.text+0x50): undefined reference to `getuid_' > control.o: In function `control_': > control.o(.text+0x53): undefined reference to `iargc_' > control.o(.text+0xf9): undefined reference to `sleep_' > timec.o: In function `timec_': > timec.o(.text+0x6): undefined reference to `etime_' > util.o: In function `my_flush_': > util.o(.text+0x11): undefined reference to `flush_' > util.o: In function `datum_': > util.o(.text+0x13f1): undefined reference to `fdate_' > util.o: In function `my_getarg_': > util.o(.text+0x1cf3): undefined reference to `getarg_' > util.o: In function `my_getenv_': > util.o(.text+0x1d0d): undefined reference to `getenv_' > util.o: In function `my_getlog_': > util.o(.text+0x1d1f): undefined reference to `getlog_' > recpnew.o: In function `recpnew_': > recpnew.o(.text+0x17c): undefined reference to `iargc_' > recpnew.o: In function `recpeam_': > recpnew.o(.text+0x3d91): undefined reference to `iargc_' > testex.o: In function `testex_': > testex.o(.text+0x50): undefined reference to `system_' > wv30.o: In function `zhwwf_': > wv30.o(.text+0x6ad): undefined reference to `system_' > setbasis.o: In function `read_basis_.D': > setbasis.o(.text+0x1070): undefined reference to `iargc_' > setbasis.o: In function `read_basis_.A': > setbasis.o(.text+0x369b): undefined reference to `iargc_' > egointer.o: In function `interface_wait_.': > egointer.o(.text+0xb26): undefined reference to `sleep_' > egointer.o: In function `interface_read_.': > egointer.o(.text+0xc6d): undefined reference to `iargc_' > sample.o: In function `sample_wait_': > sample.o(.text+0x115): undefined reference to `sleep_' > sample.o: In function `sample_go_': > sample.o(.text+0x946): undefined reference to `system_' > rwswap.o: In function `del_swap_.': > rwswap.o(.text+0x3961): undefined reference to `system_' > /usr/lib/liblapack.so: undefined reference to `e_wsfe' > /usr/lib/liblapack.so: undefined reference to `do_fio' > /usr/lib/liblapack.so: undefined reference to `s_wsfe' > make: *** [cpmd.x] Error 1 It seems to me that I am missing some library, but I am unsure what it might be. Presently, I have the following libraries: LFLAGS = -L/usr/lib -llapack -L/usr1/quantum_chemistry/atlas_libraries/3.5.2/athlon -llapack -lf77blas -lcblas -latlas $(QMMM_LIBS) Anyone have any thoughts as to what library (or libraries) I am missing? Thank you all for your assistance! Carl -- __________________________________________________________ Carl Krauthauser, Ph.D. Research Associate University of Delaware - Center for Composite Materials 215 Composites Manufacturing Sciences Lab Newark, Delaware 19716-3144 Phone:(302) 831-8701 FAX: (302) 831-8525 e-mail: krauthau at ccm.udel.edu From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Wed Jul 16 19:01:14 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Wed, 16 Jul 2003 19:01:14 +0200 Subject: [CPMD-list] Problem linking with 3.7-1 In-Reply-To: <3F157E7A.7010506@ccm.udel.edu> Message-ID: <200307161701.h6GH1EQ13272@yello.theochem.ruhr-uni-bochum.de> On Wed, 16 Jul 2003 12:34:02 -0400 Carl Krauthauser wrote: > Dear CPMD users. > > I am attempting to compile ver. 3.7-1 with the PC-IFC configuration. At > the linking stage, I get the following error messages: > > > cpmd.o: In function `cpmd_': > > cpmd.o(.text+0x51): undefined reference to `signal_' > > cpmd.o(.text+0x65): undefined reference to `signal_' > > envir.o: In function `envir_': > > envir.o(.text+0x2f): undefined reference to `getenv_' > > envir.o(.text+0x3e): undefined reference to `getcwd_' > > envir.o(.text+0x46): undefined reference to `getpid_' > > envir.o(.text+0x50): undefined reference to `getuid_' > > control.o: In function `control_': > > control.o(.text+0x53): undefined reference to `iargc_' > > control.o(.text+0xf9): undefined reference to `sleep_' > > timec.o: In function `timec_': > > timec.o(.text+0x6): undefined reference to `etime_' > > util.o: In function `my_flush_': > > util.o(.text+0x11): undefined reference to `flush_' > > util.o: In function `datum_': > > util.o(.text+0x13f1): undefined reference to `fdate_' > > util.o: In function `my_getarg_': > > util.o(.text+0x1cf3): undefined reference to `getarg_' > > util.o: In function `my_getenv_': > > util.o(.text+0x1d0d): undefined reference to `getenv_' > > util.o: In function `my_getlog_': > > util.o(.text+0x1d1f): undefined reference to `getlog_' > > recpnew.o: In function `recpnew_': > > recpnew.o(.text+0x17c): undefined reference to `iargc_' > > recpnew.o: In function `recpeam_': > > recpnew.o(.text+0x3d91): undefined reference to `iargc_' > > testex.o: In function `testex_': > > testex.o(.text+0x50): undefined reference to `system_' > > wv30.o: In function `zhwwf_': > > wv30.o(.text+0x6ad): undefined reference to `system_' > > setbasis.o: In function `read_basis_.D': > > setbasis.o(.text+0x1070): undefined reference to `iargc_' > > setbasis.o: In function `read_basis_.A': > > setbasis.o(.text+0x369b): undefined reference to `iargc_' > > egointer.o: In function `interface_wait_.': > > egointer.o(.text+0xb26): undefined reference to `sleep_' > > egointer.o: In function `interface_read_.': > > egointer.o(.text+0xc6d): undefined reference to `iargc_' > > sample.o: In function `sample_wait_': > > sample.o(.text+0x115): undefined reference to `sleep_' > > sample.o: In function `sample_go_': > > sample.o(.text+0x946): undefined reference to `system_' > > rwswap.o: In function `del_swap_.': > > rwswap.o(.text+0x3961): undefined reference to `system_' hi! for this you need to add -Vaxlib to LFLAGS. > > /usr/lib/liblapack.so: undefined reference to `e_wsfe' > > /usr/lib/liblapack.so: undefined reference to `do_fio' > > /usr/lib/liblapack.so: undefined reference to `s_wsfe' this lapack library was most likely compiled for a different compiler. you should use a combined atlas/lapack library or the intel math kernel library instead. > > make: *** [cpmd.x] Error 1 > > It seems to me that I am missing some library, but I am unsure what it > might be. Presently, I have the following libraries: > > LFLAGS = -L/usr/lib -llapack > -L/usr1/quantum_chemistry/atlas_libraries/3.5.2/athlon -llapack > -lf77blas -lcblas -latlas $(QMMM_LIBS) > if your machine is an amd athlon, you can download a precompiled, athlon xp optimized lapack/atlas binary at , copy the file to your compilation directory and use LFLAGS = -L. -latlas -Vaxlib $(QMMM_LIBS) hope this helps, axel kohlmeyer. > Anyone have any thoughts as to what library (or libraries) I am missing? > Thank you all for your assistance! > > > Carl > > -- > __________________________________________________________ > Carl Krauthauser, Ph.D. > Research Associate > University of Delaware - Center for Composite Materials > 215 Composites Manufacturing Sciences Lab > Newark, Delaware 19716-3144 > Phone:(302) 831-8701 > FAX: (302) 831-8525 > e-mail: krauthau at ccm.udel.edu > > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Dr. Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Wed Jul 16 19:05:55 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Wed, 16 Jul 2003 19:05:55 +0200 Subject: [CPMD-list] F90 linking problem In-Reply-To: <000001c34b14$38220a90$6f91fea9@MUNAY> Message-ID: <200307161705.h6GH5tX13460@yello.theochem.ruhr-uni-bochum.de> On Tue, 15 Jul 2003 17:01:10 -0400 "Luis Agapito" wrote: > This is a multi-part message in MIME format. > > ------=_NextPart_000_0001_01C34AF2.B111F130 > Content-Type: text/plain; > charset="us-ascii" > Content-Transfer-Encoding: 7bit > > Hello everybody, > I'm installing CPMD v3.7.0 on a DEC-ALPHA machine using the MPI method > for parallelizing. I got this error in the linking step, Am I missing > any library? (maybe related with the lfmpi option for the F90), where > could I get more information about that? > Thanks > > > nmr_position_p.o hardness_p.o respin_p.o response_p.o restart_p.o > pert_kpoint_ > p.o dnlpdk_p.o rnl_dk_p.o kpert_potential_p.o up3_p.o friesner_c_p.o > matrix_p > .o ks_ener_p.o kpert_util_p.o fukui_p.o kdp_diag.o kdp_prep.o kdp_rho.o > kdp_stre > ss_kin.o kdpoints.o rhoofr_kdp.o mm_detsp.o mm_init.o mm_dim.o > mm_qmmm_forcedr > .o mm_mdmain.o mm_coninp.o mm_cpmd_add_MM_forces_f77.o > mm_cpmd_esp_charges_f77 > .o mm_cpmd_ext_pot_f77.o vdwin.o vdw.o sysdepend.o shmemory.o -ldxml > -lfmpi -l > mpi -lrt -pthread > ld: > Can't locate file for: -lfmpi > f90: Severe: Failed while trying to link. > *** Exit 1 > Stop. hi, you probably only have to specify the directory, e.g. by -L/usr/mpi/lib. you can try: locate libfmpi.a or locate libfmpi.a or find /usr -name libfmpi.a -print or ask your local system administrator to find out what the actual directory is. _if_ mpi support _is_ installed, that is. hope this helps, axel kohlmeyer. > > ------------------------------------------------------------------------ > ---------- > Luis A. Agapito > Department of Electrical Engineering > University of South Carolina > Phone: 803-777-0720 > Moletronics Group: http://www.ee.sc.edu/research/moletronics/ > E-mail: agapito at engr.sc.edu > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From carlospi at estudiantes.fisica.unal.edu.co Thu Jul 17 19:17:40 2003 From: carlospi at estudiantes.fisica.unal.edu.co (Pinilla Castellanos Carlos) Date: Thu, 17 Jul 2003 12:17:40 -0500 (COT) Subject: [CPMD-list] Problem linking with 3.7-1 In-Reply-To: <3F157E7A.7010506@ccm.udel.edu> Message-ID: Hi Carl I used the following makefile in PC-IFC for CPMD 3.7.0 --------------- Default Configuration for PC-IFC --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = -c -w -r8 -w90 -w95 -O3 -pc64 -axM -ip -tpp7 LFLAGS = -L/opt/intel/mkl/32/ -lmkl_lapack -lmkl_p4 \ -lguide -lpthread -Vaxlib CFLAGS = -c -O2 -Wall CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC NOOPT_FLAG = CC = gcc FC = ifc LD = ifc AR = Regards Carlos On Wed, 16 Jul 2003, Carl Krauthauser wrote: > Dear CPMD users. > > I am attempting to compile ver. 3.7-1 with the PC-IFC configuration. At > the linking stage, I get the following error messages: > > > cpmd.o: In function `cpmd_': > > cpmd.o(.text+0x51): undefined reference to `signal_' > > cpmd.o(.text+0x65): undefined reference to `signal_' > > envir.o: In function `envir_': > > envir.o(.text+0x2f): undefined reference to `getenv_' > > envir.o(.text+0x3e): undefined reference to `getcwd_' > > envir.o(.text+0x46): undefined reference to `getpid_' > > envir.o(.text+0x50): undefined reference to `getuid_' > > control.o: In function `control_': > > control.o(.text+0x53): undefined reference to `iargc_' > > control.o(.text+0xf9): undefined reference to `sleep_' > > timec.o: In function `timec_': > > timec.o(.text+0x6): undefined reference to `etime_' > > util.o: In function `my_flush_': > > util.o(.text+0x11): undefined reference to `flush_' > > util.o: In function `datum_': > > util.o(.text+0x13f1): undefined reference to `fdate_' > > util.o: In function `my_getarg_': > > util.o(.text+0x1cf3): undefined reference to `getarg_' > > util.o: In function `my_getenv_': > > util.o(.text+0x1d0d): undefined reference to `getenv_' > > util.o: In function `my_getlog_': > > util.o(.text+0x1d1f): undefined reference to `getlog_' > > recpnew.o: In function `recpnew_': > > recpnew.o(.text+0x17c): undefined reference to `iargc_' > > recpnew.o: In function `recpeam_': > > recpnew.o(.text+0x3d91): undefined reference to `iargc_' > > testex.o: In function `testex_': > > testex.o(.text+0x50): undefined reference to `system_' > > wv30.o: In function `zhwwf_': > > wv30.o(.text+0x6ad): undefined reference to `system_' > > setbasis.o: In function `read_basis_.D': > > setbasis.o(.text+0x1070): undefined reference to `iargc_' > > setbasis.o: In function `read_basis_.A': > > setbasis.o(.text+0x369b): undefined reference to `iargc_' > > egointer.o: In function `interface_wait_.': > > egointer.o(.text+0xb26): undefined reference to `sleep_' > > egointer.o: In function `interface_read_.': > > egointer.o(.text+0xc6d): undefined reference to `iargc_' > > sample.o: In function `sample_wait_': > > sample.o(.text+0x115): undefined reference to `sleep_' > > sample.o: In function `sample_go_': > > sample.o(.text+0x946): undefined reference to `system_' > > rwswap.o: In function `del_swap_.': > > rwswap.o(.text+0x3961): undefined reference to `system_' > > /usr/lib/liblapack.so: undefined reference to `e_wsfe' > > /usr/lib/liblapack.so: undefined reference to `do_fio' > > /usr/lib/liblapack.so: undefined reference to `s_wsfe' > > make: *** [cpmd.x] Error 1 > > It seems to me that I am missing some library, but I am unsure what it > might be. Presently, I have the following libraries: > > LFLAGS = -L/usr/lib -llapack > -L/usr1/quantum_chemistry/atlas_libraries/3.5.2/athlon -llapack > -lf77blas -lcblas -latlas $(QMMM_LIBS) > > Anyone have any thoughts as to what library (or libraries) I am missing? > Thank you all for your assistance! > > > Carl > > -- > __________________________________________________________ > Carl Krauthauser, Ph.D. > Research Associate > University of Delaware - Center for Composite Materials > 215 Composites Manufacturing Sciences Lab > Newark, Delaware 19716-3144 > Phone:(302) 831-8701 > FAX: (302) 831-8525 > e-mail: krauthau at ccm.udel.edu > > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Sat Jul 19 13:06:18 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 19 Jul 2003 13:06:18 +0200 (MEST) Subject: [CPMD-list] p4_error: : 14 In-Reply-To: <3F142D57.11649.119B81E@localhost> References: <3F142D57.11649.119B81E@localhost> Message-ID: Hi Martijn I don't know what really causes your problems. However, I found a bug related to the type of job you are running. The variable "SLIMIT" is not initialized. An easy workaround would be to put the line SLIMIT = 0.D0 somewhere in the file control_def.F Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Tue, 15 Jul 2003, Martijn Zwijnenburg wrote: > Hi, > > When I run the inputfile given below on our linux-cluster (cpmd 3.7.1 > / compiled with PGI compiler/ MPICH) the MD job always crashes after > the 205th step (independent of the precise nodes I'm running / or the > number of nodes I'm running on). The error message seems to be > precisely repeatable, so I checked if a file nears a limit (say 2GB) > but everything is much much smaller. The precise error message (for 3 > procs) is: > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > 0.180 70.43 > p1_15605: p4_error: : 14 > p2_25632: p4_error: : 14 > bm_list_27416: (20873.130096) wakeup_slave: unable to interrupt slave > 0 pid 27415 > Broken pipe > Broken pipe > > and in case of the 2 proc job: > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > 0.180 108.08 > p1_23514: p4_error: : 14 > Broken pipe > > The funny thing is that we never had such problems with CPMD > optimization jobs, which ran for weeks, nor with any other program on > the cluster (gamess-uk, dlpolly). > It appears that MPI is the problem but does anybody have a solution? > > Gr. Martijn > > Inputfile: > > ! Si4O8 cluster MD > ! > &CPMD > MOLECULAR DYNAMICS > STRUCTURE BONDS ANGLES DIHEDRALS > SPLINE RANGE > 5.00 > SPLINE POINTS > 2500 > CONVERGENCE > 1E-6 5E-6 1. > ISOLATED MOLECULE > HESSIAN UNIT > RESTART WAVEFUNCTION GEOMETRY LATEST > TEMPERATURE > 1000 > TEMPCONTROL IONS > 1000 20 > TIMESTEP > 5.0 > EMASS > 700 > TRAJECTORY XYZ > &END > &SYSTEM > SYMMETRY > 0 > ANGSTROM > CELL > 15 1.0 1.0 0 0 0 > ANGSTROM > CUTOFF > 30.00 > sTATES > 36 > POINT GROUP > AUTO > &END > &ATOMS > *Si_ps.uspp BINARY new.f > LMAX=P > 4 > 0.000000000000 0.000000000000 0.000000000000 > 0.471809914327 0.000000000000 3.182682023275 > 3.653225710458 -0.034526749450 2.711038719844 > 3.181409275132 -0.034559479116 -0.471644591443 > > *o_ps.uspp BINARY new.f > LMAX=P > 8 > 0.000000000000 0.000000000000 1.626292446278 > 2.098080331254 -0.016336484977 3.186832736955 > 3.653206618118 -0.034138170947 1.084746035991 > 1.555131813001 -0.018896584142 -0.475808582925 > 4.871737465360 -0.049224011432 3.612939458029 > -0.432765838343 0.012873808042 4.399223302846 > -1.218511827443 0.015066749685 -0.901895052361 > 4.086009030180 -0.046817197390 -1.688173543327 > > &END > &DFT > FUNCTIONAL PW91 > GC-CUTOFF > 5E-5 > &END > ---------------------------------------------------------------------- > --- > Martijn Zwijnenburg > Lab. of Applied Organic Chemistry and Catalysis > Delft University of Technology > Julianalaan 136 > 2628 BL Delft > The Netherlands > Tel: 0031-(0)152782691 > Fax: 0031-(0)152784700 > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > web page: http://come.to/tock > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Sat Jul 19 13:16:22 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 19 Jul 2003 13:16:22 +0200 (MEST) Subject: [CPMD-list] MD problem! In-Reply-To: <20030715151105.40902.qmail@web15302.mail.bjs.yahoo.com> References: <20030715151105.40902.qmail@web15302.mail.bjs.yahoo.com> Message-ID: Hi > Dear all, > > I met the following problem when performing MD calculation! > > 367 0.00112 16.6 -294.14844 -294.14379 -294.14267 0.009 26.34 > 368 0.00114 16.9 -294.14855 -294.14380 -294.14267 0.009 25.94 > 369 0.00115 17.3 -294.14866 -294.14382 -294.14267 0.009 24.05 > 370 0.00116 17.6 -294.14877 -294.14383 -294.14267 0.008 25.49 > 371 0.00118 18.0 -294.14889 -294.14384 -294.14267 0.008 26.80 > 372 0.00119 18.4 -294.14901 -294.14386 -294.14267 0.008 27.22 > 373 0.00121 18.8 -294.14913 -294.14387 -294.14267 0.008 25.35 > 374 0.00123 19.2 -294.14926 -294.14389 -294.14266 0.008 25.89 > 375 0.00124 19.5 -294.14938 -294.14390 -294.14266 0.008 27.72 > 376 0.00126 19.9 -294.14951 -294.14392 -294.14265 0.008 25.37 > 377 0.00130 20.3 -294.14964 -294.14394 -294.14264 0.008 25.12 > 378 0.00135 20.7 -294.14973 -294.14391 -294.14257 0.008 24.31 > 379 0.04902 21.0 -294.14982 -294.14392 -294.09491 0.008 26.99 > 380***************** -294.06582 3216785.58973************** 0.008 25.40 > > PROGRAM STOPS IN SUBROUTINE DSPEVY| FAILED TO DIAGONALIZE [PROC= 5] > > In the 379th steps, the electronic temperature becomes very high and the EHAM is not constant. So the timestep may be too long! But I perfom almost the same calculation with the same timestep and EMASS at the same system, the program runs okay! Moreover, the electron can follow the movement of Ion. > You are using a bad combination of EMASS and TIMESTEP for your system. Obviously there is a difference between this and your other system where the settings of the parameters work. Most likely this system has a smaller electronic gap. Reduce EMASS and probably TIMESTEP and see if you can get stable dynamics. Juerg From hutter at pci.unizh.ch Sat Jul 19 13:18:28 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 19 Jul 2003 13:18:28 +0200 (MEST) Subject: [CPMD-list] On PCG Minimize! In-Reply-To: <20030715152235.63074.qmail@web15311.mail.bjs.yahoo.com> References: <20030715152235.63074.qmail@web15311.mail.bjs.yahoo.com> Message-ID: Hi the PCG minimizer is sometimes traped into a situation where the step and the orthogonalization cancel, leading to very small LAMBDA. If this happens only a restart is possible. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Tue, 15 Jul 2003, [gb2312] Adrain Zhou wrote: > Dear all, > > It seems there are some problems when using PCG Minimize to perform a multisteps wavefunction optimization. It is as follows, > > LINE SEARCH : LAMBDA=.103 ENERGY = -294.14310 > 228 3.006E-04 3.043E-06 -294.142940 -4.685E-06 43.33 > LINE SEARCH : LAMBDA=.113E-01 ENERGY = -294.14314 > 229 5.253E-04 4.688E-06 -294.142920 2.063E-05 40.45 > LINE SEARCH : LAMBDA=.116E-02 ENERGY = -294.14278 > 230 7.913E-04 9.229E-06 -294.142801 1.182E-04 40.08 > LINE SEARCH : LAMBDA=.217E-03 ENERGY = -294.14273 > 231 8.708E-04 1.001E-05 -294.142741 6.047E-05 39.11 > LINE SEARCH : LAMBDA=.407E-04 ENERGY = -294.14272 > 232 8.895E-04 1.086E-05 -294.142725 1.576E-05 40.98 > LINE SEARCH : LAMBDA=.764E-05 ENERGY = -294.14272 > 233 8.932E-04 1.086E-05 -294.142722 3.149E-06 42.24 > LINE SEARCH : LAMBDA=.143E-05 ENERGY = -294.14272 > 234 8.939E-04 1.086E-05 -294.142721 5.747E-07 39.37 > ~ > LAMBDA is less than 0.01, the convergence never meet! When I just stop at the 120th step, and then restart again. The convergence reached very fast! It only take another 20 steps. It is so strange! > > Can anyone tell me what's wrong with it? > > Many thanks in advance! > > Zhou > > > > > > --------------------------------- > Do You Yahoo!? > ?????????????????????????????? From M.A.Zwijnenburg at tnw.tudelft.nl Mon Jul 21 10:21:28 2003 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Mon, 21 Jul 2003 10:21:28 +0200 Subject: [CPMD-list] p4_error: : 14 In-Reply-To: References: <3F142D57.11649.119B81E@localhost> Message-ID: <3F1BBEA0.31591.1EA93DC3@localhost> Hi Juerg, Thanx, I will recompile the code ASAP (or as soon the system administrator will let me). In the meantime we've been doing some experiments regarding the error message. First looking through the source code of mpich the p4_error 14 message appears to be linked to a SIGALRM 14 signal. Somewhere, in mpich an alarm() appears to be set, which expires an terminates the proces in question. However, we've not been able as yet to pin- point where (alarm can be set multiple times), although the most of times with an error string which seems to be absent in our particular example. Furthermore, I ran the nh3-md example from last years CPMD tutorial in Lyon which finished without problems. Then I performed md runs on a small silica molecule (employing a 4 rather then 15A box) with "normal" and Vanderbilt pp's and the md settings from the nh3-md example. While the run with normal pp's finished normally after 30000 steps, the Vanderbilt run crashed at precisely the same point (after cycle 205) as the previous md runs on larger clusters. The problem thus appears to be linked in someway to using Vanderbilt pp's. Finally, the same error message also popped up friday in an geometry optimization job of a colleague of mine. The problems therefore doesn't need to be linked to md (exclusively). Hope this helps, Gr. Martijn On 19 Jul 2003 at 13:06, Juerg Hutter wrote: > Hi Martijn > > I don't know what really causes your problems. > However, I found a bug related to the type of job you > are running. The variable "SLIMIT" is not initialized. > An easy workaround would be to put the line > > SLIMIT = 0.D0 > > somewhere in the file control_def.F > > Juerg > > ---------------------------------------------------------- > Juerg Hutter Phone : ++41 1 635 4491 > Physical Chemistry Institute FAX : ++41 1 635 6838 > University of Zurich E-mail: hutter at pci.unizh.ch > Winterthurerstrasse 190 > CH-8057 Zurich, Switzerland > ---------------------------------------------------------- > > > On Tue, 15 Jul 2003, Martijn Zwijnenburg wrote: > > > Hi, > > > > When I run the inputfile given below on our linux-cluster (cpmd 3.7.1 > > / compiled with PGI compiler/ MPICH) the MD job always crashes after > > the 205th step (independent of the precise nodes I'm running / or the > > number of nodes I'm running on). The error message seems to be > > precisely repeatable, so I checked if a file nears a limit (say 2GB) > > but everything is much much smaller. The precise error message (for 3 > > procs) is: > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > 0.180 70.43 > > p1_15605: p4_error: : 14 > > p2_25632: p4_error: : 14 > > bm_list_27416: (20873.130096) wakeup_slave: unable to interrupt slave > > 0 pid 27415 > > Broken pipe > > Broken pipe > > > > and in case of the 2 proc job: > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > 0.180 108.08 > > p1_23514: p4_error: : 14 > > Broken pipe > > > > The funny thing is that we never had such problems with CPMD > > optimization jobs, which ran for weeks, nor with any other program on > > the cluster (gamess-uk, dlpolly). > > It appears that MPI is the problem but does anybody have a solution? > > > > Gr. Martijn > > > > Inputfile: > > > > ! Si4O8 cluster MD > > ! > > &CPMD > > MOLECULAR DYNAMICS > > STRUCTURE BONDS ANGLES DIHEDRALS > > SPLINE RANGE > > 5.00 > > SPLINE POINTS > > 2500 > > CONVERGENCE > > 1E-6 5E-6 1. > > ISOLATED MOLECULE > > HESSIAN UNIT > > RESTART WAVEFUNCTION GEOMETRY LATEST > > TEMPERATURE > > 1000 > > TEMPCONTROL IONS > > 1000 20 > > TIMESTEP > > 5.0 > > EMASS > > 700 > > TRAJECTORY XYZ > > &END > > &SYSTEM > > SYMMETRY > > 0 > > ANGSTROM > > CELL > > 15 1.0 1.0 0 0 0 > > ANGSTROM > > CUTOFF > > 30.00 > > sTATES > > 36 > > POINT GROUP > > AUTO > > &END > > &ATOMS > > *Si_ps.uspp BINARY new.f > > LMAX=P > > 4 > > 0.000000000000 0.000000000000 0.000000000000 > > 0.471809914327 0.000000000000 3.182682023275 > > 3.653225710458 -0.034526749450 2.711038719844 > > 3.181409275132 -0.034559479116 -0.471644591443 > > > > *o_ps.uspp BINARY new.f > > LMAX=P > > 8 > > 0.000000000000 0.000000000000 1.626292446278 > > 2.098080331254 -0.016336484977 3.186832736955 > > 3.653206618118 -0.034138170947 1.084746035991 > > 1.555131813001 -0.018896584142 -0.475808582925 > > 4.871737465360 -0.049224011432 3.612939458029 > > -0.432765838343 0.012873808042 4.399223302846 > > -1.218511827443 0.015066749685 -0.901895052361 > > 4.086009030180 -0.046817197390 -1.688173543327 > > > > &END > > &DFT > > FUNCTIONAL PW91 > > GC-CUTOFF > > 5E-5 > > &END > > ---------------------------------------------------------------------- > > --- > > Martijn Zwijnenburg > > Lab. of Applied Organic Chemistry and Catalysis > > Delft University of Technology > > Julianalaan 136 > > 2628 BL Delft > > The Netherlands > > Tel: 0031-(0)152782691 > > Fax: 0031-(0)152784700 > > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > > web page: http://come.to/tock > > > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://www.cpmd.org/mailman/listinfo/cpmd-list > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list ------------------------------------------------------------------------- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From rjsa at mail.shcnc.ac.cn Mon Jul 21 11:11:29 2003 From: rjsa at mail.shcnc.ac.cn (Rongjian Sa) Date: Mon, 21 Jul 2003 17:11:29 +0800 Subject: [CPMD-list] help! Message-ID: <20030721090502.70E412085D1@mail.shcnc.ac.cn> Dear All: I have found one problem when I used the constrained MD method in CPMD. My calculated system contain one ammonium and one benzene, and I attempted to use the distance of nitrogen and benzene as my reaction coordinate. However, I have noticed the version 3.7 or the higher version occur in error message, while the 3.5 version can deal such question. Because my calculation is based on 3.7 version, so I wonder if you can tell me how to modify it or which code should be replace by original 3.5 code? Thank you! Sincerely, Rongjian -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030721/ab139d4d/attachment.html From cur at zurich.ibm.com Mon Jul 21 14:43:38 2003 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Mon, 21 Jul 2003 14:43:38 +0200 Subject: [CPMD-list] p4_error: : 14 In-Reply-To: <3F1BBEA0.31591.1EA93DC3@localhost> Message-ID: The error you are experiencing has probably to do with a bug in a the memory allocation of a timing routine. This has been corrected in the new version 3.7.2 that will be distributed in the next few days. best regards, Alessandro CURIONI, PhD Research Staff Member Computational Biochemistry and Material Science group IBM Research Division - Zurich Research Laboratory Saumerstrasse 4 8003 Rueschlikon - Switzerland e-mail: cur at zurich.ibm.com www: www.zurich.ibm.com Tel: +41-1-7248633 Fax: +41-1-7248958 "Martijn Zwijnenburg" Juerg Hutter Sent by: cc cpmd-list-admin at c cpmd-list at cpmd.org pmd.org Subject Re: [CPMD-list] p4_error: : 14 07/21/2003 10:21 AM Hi Juerg, Thanx, I will recompile the code ASAP (or as soon the system administrator will let me). In the meantime we've been doing some experiments regarding the error message. First looking through the source code of mpich the p4_error 14 message appears to be linked to a SIGALRM 14 signal. Somewhere, in mpich an alarm() appears to be set, which expires an terminates the proces in question. However, we've not been able as yet to pin- point where (alarm can be set multiple times), although the most of times with an error string which seems to be absent in our particular example. Furthermore, I ran the nh3-md example from last years CPMD tutorial in Lyon which finished without problems. Then I performed md runs on a small silica molecule (employing a 4 rather then 15A box) with "normal" and Vanderbilt pp's and the md settings from the nh3-md example. While the run with normal pp's finished normally after 30000 steps, the Vanderbilt run crashed at precisely the same point (after cycle 205) as the previous md runs on larger clusters. The problem thus appears to be linked in someway to using Vanderbilt pp's. Finally, the same error message also popped up friday in an geometry optimization job of a colleague of mine. The problems therefore doesn't need to be linked to md (exclusively). Hope this helps, Gr. Martijn On 19 Jul 2003 at 13:06, Juerg Hutter wrote: > Hi Martijn > > I don't know what really causes your problems. > However, I found a bug related to the type of job you > are running. The variable "SLIMIT" is not initialized. > An easy workaround would be to put the line > > SLIMIT = 0.D0 > > somewhere in the file control_def.F > > Juerg > > ---------------------------------------------------------- > Juerg Hutter Phone : ++41 1 635 4491 > Physical Chemistry Institute FAX : ++41 1 635 6838 > University of Zurich E-mail: hutter at pci.unizh.ch > Winterthurerstrasse 190 > CH-8057 Zurich, Switzerland > ---------------------------------------------------------- > > > On Tue, 15 Jul 2003, Martijn Zwijnenburg wrote: > > > Hi, > > > > When I run the inputfile given below on our linux-cluster (cpmd 3.7.1 > > / compiled with PGI compiler/ MPICH) the MD job always crashes after > > the 205th step (independent of the precise nodes I'm running / or the > > number of nodes I'm running on). The error message seems to be > > precisely repeatable, so I checked if a file nears a limit (say 2GB) > > but everything is much much smaller. The precise error message (for 3 > > procs) is: > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > 0.180 70.43 > > p1_15605: p4_error: : 14 > > p2_25632: p4_error: : 14 > > bm_list_27416: (20873.130096) wakeup_slave: unable to interrupt slave > > 0 pid 27415 > > Broken pipe > > Broken pipe > > > > and in case of the 2 proc job: > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > 0.180 108.08 > > p1_23514: p4_error: : 14 > > Broken pipe > > > > The funny thing is that we never had such problems with CPMD > > optimization jobs, which ran for weeks, nor with any other program on > > the cluster (gamess-uk, dlpolly). > > It appears that MPI is the problem but does anybody have a solution? > > > > Gr. Martijn > > > > Inputfile: > > > > ! Si4O8 cluster MD > > ! > > &CPMD > > MOLECULAR DYNAMICS > > STRUCTURE BONDS ANGLES DIHEDRALS > > SPLINE RANGE > > 5.00 > > SPLINE POINTS > > 2500 > > CONVERGENCE > > 1E-6 5E-6 1. > > ISOLATED MOLECULE > > HESSIAN UNIT > > RESTART WAVEFUNCTION GEOMETRY LATEST > > TEMPERATURE > > 1000 > > TEMPCONTROL IONS > > 1000 20 > > TIMESTEP > > 5.0 > > EMASS > > 700 > > TRAJECTORY XYZ > > &END > > &SYSTEM > > SYMMETRY > > 0 > > ANGSTROM > > CELL > > 15 1.0 1.0 0 0 0 > > ANGSTROM > > CUTOFF > > 30.00 > > sTATES > > 36 > > POINT GROUP > > AUTO > > &END > > &ATOMS > > *Si_ps.uspp BINARY new.f > > LMAX=P > > 4 > > 0.000000000000 0.000000000000 0.000000000000 > > 0.471809914327 0.000000000000 3.182682023275 > > 3.653225710458 -0.034526749450 2.711038719844 > > 3.181409275132 -0.034559479116 -0.471644591443 > > > > *o_ps.uspp BINARY new.f > > LMAX=P > > 8 > > 0.000000000000 0.000000000000 1.626292446278 > > 2.098080331254 -0.016336484977 3.186832736955 > > 3.653206618118 -0.034138170947 1.084746035991 > > 1.555131813001 -0.018896584142 -0.475808582925 > > 4.871737465360 -0.049224011432 3.612939458029 > > -0.432765838343 0.012873808042 4.399223302846 > > -1.218511827443 0.015066749685 -0.901895052361 > > 4.086009030180 -0.046817197390 -1.688173543327 > > > > &END > > &DFT > > FUNCTIONAL PW91 > > GC-CUTOFF > > 5E-5 > > &END > > ---------------------------------------------------------------------- > > --- > > Martijn Zwijnenburg > > Lab. of Applied Organic Chemistry and Catalysis > > Delft University of Technology > > Julianalaan 136 > > 2628 BL Delft > > The Netherlands > > Tel: 0031-(0)152782691 > > Fax: 0031-(0)152784700 > > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > > web page: http://come.to/tock > > > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://www.cpmd.org/mailman/listinfo/cpmd-list > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list ------------------------------------------------------------------------- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock _______________________________________________ CPMD-list mailing list CPMD-list at cpmd.org http://www.cpmd.org/mailman/listinfo/cpmd-list From M.A.Zwijnenburg at tnw.tudelft.nl Mon Jul 21 17:21:34 2003 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Mon, 21 Jul 2003 17:21:34 +0200 Subject: [CPMD-list] p4_error: : 14 In-Reply-To: References: <3F1BBEA0.31591.1EA93DC3@localhost> Message-ID: <3F1C2116.11307.2029DA16@localhost> Hi, We highly anticipating the new version, md with only 205 steps is just no fun................... Since in the meantime we also compiled CPMD with lam, herewith, for historical reasons probably only, it's error message: 204 0.14522 1500.0 -37.48879 -37.48166 -37.33644 0.151 0.14 205 0.13394 1500.0 -37.50746 -37.50033 -37.36639 0.153 0.15 MPI_Recv: message truncated (rank 1, MPI_COMM_WORLD) Rank (1, MPI_COMM_WORLD): Call stack within LAM: Rank (1, MPI_COMM_WORLD): - MPI_Recv() Rank (1, MPI_COMM_WORLD): - MPI_Bcast() Rank (1, MPI_COMM_WORLD): - MPI_Allreduce() Rank (1, MPI_COMM_WORLD): - main() ---------------------------------------------------------------------- ---- --- One of the processes started by mpirun has exited with a nonzero exit code. This typically indicates that the process finished in error. If your process did not finish in error, be sure to include a "return 0" or "exit(0)" in your C code before exiting the application. PID 28006 failed on node n0 with exit status 1. Cheers, Martijn On 21 Jul 2003 at 14:43, Alessandro Curioni wrote: > > > > > The error you are experiencing has probably to do with a bug in a the > memory allocation of a timing routine. > This has been corrected in the new version 3.7.2 that will be distributed > in the next few days. > > > best regards, > > Alessandro CURIONI, PhD > Research Staff Member > Computational Biochemistry and Material Science group > IBM Research Division - Zurich Research Laboratory > Saumerstrasse 4 > 8003 Rueschlikon - Switzerland > e-mail: cur at zurich.ibm.com > www: www.zurich.ibm.com > Tel: +41-1-7248633 > Fax: +41-1-7248958 > > > > > "Martijn > Zwijnenburg" > tnw.tudelft.nl> Juerg Hutter > Sent by: cc > cpmd-list-admin at c cpmd-list at cpmd.org > pmd.org Subject > Re: [CPMD-list] p4_error: : 14 > > 07/21/2003 10:21 > AM > > > > > > > > Hi Juerg, > > Thanx, I will recompile the code ASAP (or as soon the system > administrator will let me). > > In the meantime we've been doing some experiments regarding the error > message. First looking through the source code of mpich the p4_error > 14 message appears to be linked to a SIGALRM 14 signal. Somewhere, > in mpich an alarm() appears to be set, which expires an terminates > the proces in question. However, we've not been able as yet to pin- > point where (alarm can be set multiple times), although the most of > times with an error string which seems to be absent in our particular > example. > > Furthermore, I ran the nh3-md example from last years CPMD tutorial > in Lyon which finished without problems. Then I performed md runs on > a small silica molecule (employing a 4 rather then 15A box) with > "normal" and Vanderbilt pp's and the md settings from the nh3-md > example. While the run with normal pp's finished normally after 30000 > steps, the Vanderbilt run crashed at precisely the same point (after > cycle 205) as the previous md runs on larger clusters. The problem > thus appears to be linked in someway to using Vanderbilt pp's. > > Finally, the same error message also popped up friday in an geometry > optimization job of a colleague of mine. The problems therefore > doesn't need to be linked to md (exclusively). > > Hope this helps, > > Gr. Martijn > > > On 19 Jul 2003 at 13:06, Juerg Hutter wrote: > > > Hi Martijn > > > > I don't know what really causes your problems. > > However, I found a bug related to the type of job you > > are running. The variable "SLIMIT" is not initialized. > > An easy workaround would be to put the line > > > > SLIMIT = 0.D0 > > > > somewhere in the file control_def.F > > > > Juerg > > > > ---------------------------------------------------------- > > Juerg Hutter Phone : ++41 1 635 4491 > > Physical Chemistry Institute FAX : ++41 1 635 6838 > > University of Zurich E-mail: hutter at pci.unizh.ch > > Winterthurerstrasse 190 > > CH-8057 Zurich, Switzerland > > ---------------------------------------------------------- > > > > > > On Tue, 15 Jul 2003, Martijn Zwijnenburg wrote: > > > > > Hi, > > > > > > When I run the inputfile given below on our linux-cluster (cpmd 3.7.1 > > > / compiled with PGI compiler/ MPICH) the MD job always crashes after > > > the 205th step (independent of the precise nodes I'm running / or the > > > number of nodes I'm running on). The error message seems to be > > > precisely repeatable, so I checked if a file nears a limit (say 2GB) > > > but everything is much much smaller. The precise error message (for 3 > > > procs) is: > > > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > > 0.180 70.43 > > > p1_15605: p4_error: : 14 > > > p2_25632: p4_error: : 14 > > > bm_list_27416: (20873.130096) wakeup_slave: unable to interrupt slave > > > 0 pid 27415 > > > Broken pipe > > > Broken pipe > > > > > > and in case of the 2 proc job: > > > > > > 205 0.00852 1007.3 -143.81623 -143.76839 -143.75987 > > > 0.180 108.08 > > > p1_23514: p4_error: : 14 > > > Broken pipe > > > > > > The funny thing is that we never had such problems with CPMD > > > optimization jobs, which ran for weeks, nor with any other program on > > > the cluster (gamess-uk, dlpolly). > > > It appears that MPI is the problem but does anybody have a solution? > > > > > > Gr. Martijn > > > > > > Inputfile: > > > > > > ! Si4O8 cluster MD > > > ! > > > &CPMD > > > MOLECULAR DYNAMICS > > > STRUCTURE BONDS ANGLES DIHEDRALS > > > SPLINE RANGE > > > 5.00 > > > SPLINE POINTS > > > 2500 > > > CONVERGENCE > > > 1E-6 5E-6 1. > > > ISOLATED MOLECULE > > > HESSIAN UNIT > > > RESTART WAVEFUNCTION GEOMETRY LATEST > > > TEMPERATURE > > > 1000 > > > TEMPCONTROL IONS > > > 1000 20 > > > TIMESTEP > > > 5.0 > > > EMASS > > > 700 > > > TRAJECTORY XYZ > > > &END > > > &SYSTEM > > > SYMMETRY > > > 0 > > > ANGSTROM > > > CELL > > > 15 1.0 1.0 0 0 0 > > > ANGSTROM > > > CUTOFF > > > 30.00 > > > sTATES > > > 36 > > > POINT GROUP > > > AUTO > > > &END > > > &ATOMS > > > *Si_ps.uspp BINARY new.f > > > LMAX=P > > > 4 > > > 0.000000000000 0.000000000000 0.000000000000 > > > 0.471809914327 0.000000000000 3.182682023275 > > > 3.653225710458 -0.034526749450 2.711038719844 > > > 3.181409275132 -0.034559479116 -0.471644591443 > > > > > > *o_ps.uspp BINARY new.f > > > LMAX=P > > > 8 > > > 0.000000000000 0.000000000000 1.626292446278 > > > 2.098080331254 -0.016336484977 3.186832736955 > > > 3.653206618118 -0.034138170947 1.084746035991 > > > 1.555131813001 -0.018896584142 -0.475808582925 > > > 4.871737465360 -0.049224011432 3.612939458029 > > > -0.432765838343 0.012873808042 4.399223302846 > > > -1.218511827443 0.015066749685 -0.901895052361 > > > 4.086009030180 -0.046817197390 -1.688173543327 > > > > > > &END > > > &DFT > > > FUNCTIONAL PW91 > > > GC-CUTOFF > > > 5E-5 > > > &END > > > ---------------------------------------------------------------------- > > > --- > > > Martijn Zwijnenburg > > > Lab. of Applied Organic Chemistry and Catalysis > > > Delft University of Technology > > > Julianalaan 136 > > > 2628 BL Delft > > > The Netherlands > > > Tel: 0031-(0)152782691 > > > Fax: 0031-(0)152784700 > > > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > > > web page: http://come.to/tock > > > > > > > > > _______________________________________________ > > > CPMD-list mailing list > > > CPMD-list at cpmd.org > > > http://www.cpmd.org/mailman/listinfo/cpmd-list > > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://www.cpmd.org/mailman/listinfo/cpmd-list > > ------------------------------------------------------------------------- > Martijn Zwijnenburg > Lab. of Applied Organic Chemistry and Catalysis > Delft University of Technology > Julianalaan 136 > 2628 BL Delft > The Netherlands > Tel: 0031-(0)152782691 > Fax: 0031-(0)152784700 > e-mail: M.A.Zwijnenburg at tnw.tudelft.nl > web page: http://come.to/tock > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list ------------------------------------------------------------------------- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From cur at zurich.ibm.com Mon Jul 21 17:29:05 2003 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Mon, 21 Jul 2003 17:29:05 +0200 Subject: [CPMD-list] p4_error: : 14 In-Reply-To: <3F1C2116.11307.2029DA16@localhost> Message-ID: To your knowledge, this problem exist only for certain architectures and for VDB pseudos, so less than in 1% of the CPMD runs. Best Regards, Alessandro From y.liu at fz-juelich.de Wed Jul 23 17:49:16 2003 From: y.liu at fz-juelich.de (y.liu at fz-juelich.de) Date: Wed, 23 Jul 2003 11:49:16 -0400 Subject: [CPMD-list] Problem with building CPMD on HP/Linux Itanium-2. Message-ID: Dear everyone, I am building CPMD on HP/Linux platform (Intel Itanium-2 processors installed). First I made a Makefile by using "./Configure LINUX_IA64_INTEL-MPI", then modified library link to HP's MLIB libraries. However, when I run make, the following error message were given, --------------- rm -f cpmd.f /lib/cpp -P -C -traditional -D__alpha -DPOINTER8 -DLAPACK -DFFT_DEFAULT -DAL PHALINUX -DMYRINET -DPARALLEL=parallel -DMP_LIBRARY=__MPI -DLINUX_IA64_INTEL ./cpmd.F ./cpmd.f mpif90 -c ./cpmd.f -o ./cpmd.o /usr/bin/mpif90: eval: illegal option: -f eval: usage: eval [arg ...] make: *** [cpmd.o] Error 2 --------------- I checked the path for mpif90 is correct. What is the problem with "eval"? Should I do other modifications on Makefile? Any comments about this problem will be appreciated very much. The part of my Makefile is attached. Yi #--------------------------------------------------------------------------- - # Makefile for cpmd.x (plane wave electronic calculation) # Configuration: LINUX_IA64_INTEL-MPI # Creation of Makefile: Jul 20 2003 # on Linux ng126 2.4.18-hp1_pnnl23smp #1 SMP Tue Mar 25 22:00:46 MST 2003 ia64 unknown # Author: ng126!sconstas #--------------------------------------------------------------------------- - # SHELL = /bin/sh # #--------------- Default Configuration for LINUX_IA64_INTEL-MPI --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = LFLAGS = -L/opt/mlib/lib/linux -lveclib8 -llapack8 -lguide -lPEPCF90 \ -lm -lpthread $(QMMM_LIBS) #LFLAGS = -L/opt/intel/mkl/lib/64 -lmkl_itp -lmkl_lapack -lmkl_itp \ # -lPEPCF90 -lm -lpthread $(QMMM_LIBS) CFLAGS = -D__alpha CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__alpha -DPOINTER8 -DLAPACK -DFFT_DEFAULT -DALPHALINUX \ -DMYRINET -DPARALLEL=parallel -DMP_LIBRARY=__MPI -DLINUX_IA64_INTEL NOOPT_FLAG = CC = mpicc -c -O FC = mpif90 -c LD = mpif90 AR = /usr/bin/ar -r ... ... --------------- From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Wed Jul 23 18:17:48 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Wed, 23 Jul 2003 18:17:48 +0200 Subject: [CPMD-list] Problem with building CPMD on HP/Linux Itanium-2. In-Reply-To: Message-ID: <200307231617.h6NGHmh24469@yello.theochem.ruhr-uni-bochum.de> On Wed, 23 Jul 2003 11:49:16 -0400 y.liu at fz-juelich.de wrote: > Dear everyone, > > I am building CPMD on HP/Linux platform (Intel Itanium-2 processors installed). > First I made a Makefile by using "./Configure LINUX_IA64_INTEL-MPI", then modified library link t o HP's MLIB libraries. However, when I run make, the following error message were given, > --------------- > rm -f cpmd.f > /lib/cpp -P -C -traditional -D__alpha -DPOINTER8 -DLAPACK -DFFT_DEFAULT -DAL > PHALINUX -DMYRINET -DPARALLEL=parallel -DMP_LIBRARY=__MPI -DLINUX_IA64_INTEL > ./cpmd.F ./cpmd.f > mpif90 -c ./cpmd.f -o ./cpmd.o > /usr/bin/mpif90: eval: illegal option: -f > eval: usage: eval [arg ...] > make: *** [cpmd.o] Error 2 > --------------- > > I checked the path for mpif90 is correct. What is the problem with "eval"? Should I do other modif ications on Makefile? Any comments about this problem will be appreciated very much. The part of my Makefile is attached. hello yi, mpif90 is a wrapper to call the f90 compiler with the appropriate options for mpi compilation. in this case it seems to be a shell script that is not compatible to your linux installation (thus the eval error). you should probably contact whoever installed the mpi library on that machine and make him or her aware of the problem. cheers, axel kohlmeyer. > > Yi > > > #--------------------------------------------------------------------------- > - > # Makefile for cpmd.x (plane wave electronic calculation) > # Configuration: LINUX_IA64_INTEL-MPI > # Creation of Makefile: Jul 20 2003 > # on Linux ng126 2.4.18-hp1_pnnl23smp #1 SMP Tue Mar 25 22:00:46 MST 2003 > ia64 unknown > # Author: ng126!sconstas > #--------------------------------------------------------------------------- > - > # > SHELL = /bin/sh > # > #--------------- Default Configuration for > LINUX_IA64_INTEL-MPI --------------- > SRC = . > DEST = . > BIN = . > #QMMM_FLAGS = -D__QMECHCOUPL > #QMMM_LIBS = -L. -lmm > FFLAGS = > LFLAGS = -L/opt/mlib/lib/linux -lveclib8 -llapack8 -lguide -lPEPCF90 \ > -lm -lpthread $(QMMM_LIBS) > #LFLAGS = -L/opt/intel/mkl/lib/64 -lmkl_itp -lmkl_lapack -lmkl_itp \ > # -lPEPCF90 -lm -lpthread $(QMMM_LIBS) > CFLAGS = -D__alpha > CPP = /lib/cpp -P -C -traditional > CPPFLAGS = -D__alpha -DPOINTER8 -DLAPACK -DFFT_DEFAULT -DALPHALINUX \ > -DMYRINET -DPARALLEL=parallel -DMP_LIBRARY=__MPI -DLINUX_IA64_INTEL > NOOPT_FLAG = > CC = mpicc -c -O > FC = mpif90 -c > LD = mpif90 > AR = /usr/bin/ar -r > ... ... > --------------- > > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From eesantis at unity.ncsu.edu Wed Jul 23 22:59:26 2003 From: eesantis at unity.ncsu.edu (Erik Santiso) Date: Wed, 23 Jul 2003 15:59:26 -500 Subject: [CPMD-list] Reluctant hydrogen In-Reply-To: <20030721090502.70E412085D1@mail.shcnc.ac.cn> Message-ID: <200307231959.h6NJxKPB017485@uni01mr.unity.ncsu.edu> Hi everybody. I've been dealing with a strange problem and I can't seem to find where the error is. I'm trying to do a wavefunction optimization for an isolated hydrogen atom using CPMD. I'm using a Vanderbilt pseudopotential and the HCTH functional, here's my input file: &CPMD OPTIMIZE WAVEFUNCTION ISOLATED MOLECULE MEMORY BIG LSD &END &SYSTEM SYMMETRY 1 CELL 30.0000 1.0 1.0 0.0 0.0 0.0 CUTOFF 35 &END &ATOMS *001-H-ghct--leweb.uspp BINARY LMAX=S 1 10.0000 0.000000 0.0000000 &END &DFT FUNCTIONAL HCTH GC-CUTOFF 5.D-5 &END Now, for some reason the above doesn't converge. If I try a different method (e.g. PCG MINIMIZE) it converges but gives a total energy of circa -0.4, which is not right. Since I programmed the HCTH into Dr. Vanderbilt's code myself, I thought there might be an error in the pseudopotential, but I've checked it and compared it with the one in functionals.f in CPMD and they give identical results. I've tried running the same thing with other pseudopotentials and with a similar input file they converge extremely fast to about -0.5. There's another glitch: if I remove the LSD option from the above input, it converges very fast to an energy of about -0.46. Now the question is: shouldn't it converge to a lower value in the LSD run? I thought there was a problem with the LSD_HCTH routine in CPMD, but I couldn't find any error in it either. I'm now out of ideas on what could be wrong here, so if anyone can give me a hint on what may be failing here I'd appreciate it very much. I'm not attaching the pseudopotential file or the code I used to generate it to avoid cluttering up all your mailboxes, if anybody wants to look into it let me know and I'll send the files ASAP. Thanks a lot! Erik. ----------------------------------------- Instant psychoanalysis - Just add coffee. From WANGYI at engr.sc.edu Thu Jul 24 15:11:50 2003 From: WANGYI at engr.sc.edu (WANG, YIXUAN ) Date: Thu, 24 Jul 2003 09:11:50 -0400 Subject: [CPMD-list] Reluctant hydrogen Message-ID: <99296F29E0A01A40A0503EB30507ECF91E9A46@mail1.engr.sc.edu> Hi, If you include multiplicity ( 2) for the system, how will it work by LSD? Regards, Yixuan Wang -----Original Message----- From: cpmd-list-admin at cpmd.org [mailto:cpmd-list-admin at cpmd.org]On Behalf Of Erik Santiso Sent: Wednesday, July 23, 2003 11:59 AM To: cpmd-list at cpmd.org Subject: [CPMD-list] Reluctant hydrogen Hi everybody. I've been dealing with a strange problem and I can't seem to find where the error is. I'm trying to do a wavefunction optimization for an isolated hydrogen atom using CPMD. I'm using a Vanderbilt pseudopotential and the HCTH functional, here's my input file: &CPMD OPTIMIZE WAVEFUNCTION ISOLATED MOLECULE MEMORY BIG LSD &END &SYSTEM SYMMETRY 1 CELL 30.0000 1.0 1.0 0.0 0.0 0.0 CUTOFF 35 &END &ATOMS *001-H-ghct--leweb.uspp BINARY LMAX=S 1 10.0000 0.000000 0.0000000 &END &DFT FUNCTIONAL HCTH GC-CUTOFF 5.D-5 &END Now, for some reason the above doesn't converge. If I try a different method (e.g. PCG MINIMIZE) it converges but gives a total energy of circa -0.4, which is not right. Since I programmed the HCTH into Dr. Vanderbilt's code myself, I thought there might be an error in the pseudopotential, but I've checked it and compared it with the one in functionals.f in CPMD and they give identical results. I've tried running the same thing with other pseudopotentials and with a similar input file they converge extremely fast to about -0.5. There's another glitch: if I remove the LSD option from the above input, it converges very fast to an energy of about -0.46. Now the question is: shouldn't it converge to a lower value in the LSD run? I thought there was a problem with the LSD_HCTH routine in CPMD, but I couldn't find any error in it either. I'm now out of ideas on what could be wrong here, so if anyone can give me a hint on what may be failing here I'd appreciate it very much. I'm not attaching the pseudopotential file or the code I used to generate it to avoid cluttering up all your mailboxes, if anybody wants to look into it let me know and I'll send the files ASAP. Thanks a lot! Erik. ----------------------------------------- Instant psychoanalysis - Just add coffee. From eesantis at unity.ncsu.edu Thu Jul 24 19:35:07 2003 From: eesantis at unity.ncsu.edu (Erik Santiso) Date: Thu, 24 Jul 2003 12:35:07 -500 Subject: [CPMD-list] Reluctant hydrogen In-Reply-To: <99296F29E0A01A40A0503EB30507ECF91E9A46@mail1.engr.sc.edu> Message-ID: <200307241634.h6OGYop3022648@uni03mr.unity.ncsu.edu> Hi. Thanks for replying! It doesn't change anything, though. I think the program realizes that automatically, in the previous output files it recognized the system as a doublet without me saying it. This problem is driving me crazy because it is such a simple thing and it's not working. I thought that maybe it was somehow related to using OLDCODE, but it doesn't seem to be that either. I did the whole same thing using OLYP (with a Vanderbilt pp generated for this functional) and it works. The problem only happens with HCTH. Thanks! Erik. > Hi, > > If you include multiplicity ( 2) for the system, > how will it work by LSD? > > Regards, > > Yixuan Wang > > -----Original Message----- > From: cpmd-list-admin at cpmd.org [mailto:cpmd-list-admin at cpmd.org]On > Behalf Of Erik Santiso > Sent: Wednesday, July 23, 2003 11:59 AM > To: cpmd-list at cpmd.org > Subject: [CPMD-list] Reluctant hydrogen > > > > Hi everybody. I've been dealing with a strange problem and I can't seem to find where the error is. I'm trying to do a wavefunction optimization for an isolated hydrogen atom using CPMD. I'm using a Vanderbilt pseudopotential and the HCTH functional, here's my input file: > > &CPMD > > OPTIMIZE WAVEFUNCTION > ISOLATED MOLECULE > MEMORY BIG > LSD > > &END > > &SYSTEM > > SYMMETRY > 1 > CELL > 30.0000 1.0 1.0 0.0 0.0 0.0 > CUTOFF > 35 > > &END > > &ATOMS > > *001-H-ghct--leweb.uspp BINARY > LMAX=S > 1 > 10.0000 0.000000 0.0000000 > &END > > &DFT > > FUNCTIONAL HCTH > GC-CUTOFF > 5.D-5 > > &END > > Now, for some reason the above doesn't converge. If I try a different method (e.g. PCG MINIMIZE) it converges but gives a total energy of circa -0.4, which is not right. Since I programmed the HCTH into Dr. Vanderbilt's code myself, I thought there might be an error in the pseudopotential, but I've checked it and compared it with the one in functionals.f in CPMD and they give identical results. I've tried running the same thing with other pseudopotentials and with a similar input file they converge extremely fast to about -0.5. > > There's another glitch: if I remove the LSD option from the above input, it converges very fast to an energy of about -0.46. Now the question is: shouldn't it converge to a lower value in the LSD run? I thought there was a problem with the LSD_HCTH routine in CPMD, but I couldn't find any error in it either. I'm now out of ideas on what could be wrong here, so if anyone can give me a hint on what may be failing here I'd appreciate it very much. I'm not attaching the pseudopotential file or the code I used to generate it to avoid cluttering up all your mailboxes, if anybody wants to look into it let me know and I'll send the files ASAP. Thanks a lot! > > Erik. > > ----------------------------------------- > Instant psychoanalysis - Just add coffee. > > > ----------------------------------------- Instant psychoanalysis - Just add coffee. From LAMAS at engr.sc.edu Thu Jul 24 21:24:13 2003 From: LAMAS at engr.sc.edu (LAMAS, EDUARDO JORGE) Date: Thu, 24 Jul 2003 15:24:13 -0400 Subject: [CPMD-list] Please help with NLCC in Dfhipp and Gianozzi's program Message-ID: <8E1833C9EAA58E4581CDBA9AA5D96497064946@mail1.engr.sc.edu> Dear All: I was plotting the rho from the nonlinear core corrections that comes as an output from Gianozzi's PP generation program and comparing it with the one generated by Dfhipp program. But when plotting both functions I found the same shape and the same decay but with a factor of about 30 in the y access. Could somebody please explain me what that factor is? Thanks in advance, Eduardo -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030724/74cb827e/attachment.html From eesantis at unity.ncsu.edu Fri Jul 25 00:13:08 2003 From: eesantis at unity.ncsu.edu (Erik Santiso) Date: Thu, 24 Jul 2003 17:13:08 -500 Subject: [CPMD-list] Hydrogen =?iso-8859-1?q?solution=3F?= Message-ID: <200307242113.h6OLD8Gd019744@uni07mr.unity.ncsu.edu> Hi everyone! I think I found what's causing the problem with the hydrogen atom run for HCTH/120. There are actually two problems, both related to the SMALL parameter in the functional routines. In the subroutine LSD_HCTH, the value of the SMALL parameter is quite small (1.d-32), and that's what seemed to be causing the unstability with ODIIS. Making it bigger (say 1.d-10) eliminates the unstability, but it still converges to a wrong energy (about -0.4, same that it gave using PCG minimize before). The second problem only happens because the system has just one electron (that's why it was working for me with all the other systems but hydrogen). In this case one of the densities (and its gradient) is always zero. Now, the program assigns the SMALL value to both this density and its gradient and that must be making a spurious contribution to the energy (there is an exchange-correlation contribution from an electron that it's not there). The only solution I see to this is to do something like what's done in the functionals.f file, i.e. assign zero to the xc contribution for a density that is small, *but* it would have to be done for each density separately (i.e., don't count a contribution coming from the density of the absent electron). At least for HCTH this can be done easily, I'll be fixing this and see if it works. Thanks to all who replied! Erik. ----------------------------------------- Instant psychoanalysis - Just add coffee. From bala at jncasr.ac.in Fri Jul 25 09:40:39 2003 From: bala at jncasr.ac.in (Dr. Balasubramanian Sundaram) Date: Fri, 25 Jul 2003 13:10:39 +0530 Subject: [CPMD-list] NaNQ on P690 & on P630 References: <3F026199.C704E67D@jncasr.ac.in> <200307021127.h62BRPa32107@magadino.cscs.ch> <3F040E2B.A3FA5984@jncasr.ac.in> <20030703143524.GC81396@uranus.chemie.hu-berlin.de> Message-ID: <3F20DEF6.4CC44457@jncasr.ac.in> Hi All, I use CPMD-3.7.1. I am trying to run a MD run of one CO_2 molecule on IBM's P690 (and also tried on P630 with identical experience). The job runs fine for Geometry Optimization. When I do a MD run, the job runs fine for a few steps, and then the electron kinetic energy increases to large values and I get a NaNQ. Also note that, I do a "QUENCH BO" before the MD starts, as is required. I used to run this in Parallel with a Makefile generated from the Configure script for option, "IBM-SP4-SMP". In the Makefile, I removed the "-qsmp=omp" option, as my machine does not have OpenMP support. I got a NaNQ for this run. So, I stripped every option (such as -O3, -qtune, -qarch etc..), and also changed the compiler to just "xlf_r" instead of "mpxlf_r", i.e., basically produced a serial code. (also removed the appropriate CPPFLAGS). Yet, I got the NaNQ at the exact step number. Importantly, the input file for the job appears to be fine. I used it to run the CPMD job on a Intel PC/IFC platform, where it runs fine for several thousands of steps. Here are some details of the P630: AIX 5L, v5.1, xlf v8.1, ESSL v3.3, Lapack v3.0, PWR4 CPU. I have tried changing the pseudopotentials also (between USPP and MT_BLYP for oxygen). We compiled lapack ourselves. Is there something specific to be done in compiling lapack on IBMs? (We tried running the code with & without the CCI library also; Both ways produced NaNQs). I would very much appreciate if you can point out any mistakes that I have been doing. It has been a nerve-wracking week. I give below 3 files: (a) Relevant portion of the Makefile, (b) Input file, and (c) Portion of the output. Many Thanks, Best, Bala http://www.jncasr.ac.in/bala -------------------Makefile used on P690------------------------------------- FFLAGS = -qmaxmem=32768 -qtune=pwr4 -qarch=pwr4 LFLAGS = -L/export2/usr/lib -llapack -lesslsmp -llapack \ -bbinder:/usr/lib/bind -bmaxdata:2048000000 -qarch=pwr4 $(QMMM_LIBS) CFLAGS = CPP = /usr/ccs/lib/cpp -P CPPFLAGS = -D__IBM -DLAPACK -DPARALLEL=PARALLEL -DMP_LIBRARY=__MPI -DFFT_ESSL NOOPT_FLAG = CC = cc FC = mpxlf_r -c LD = mpxlf_r AR = /usr/bin/ar ------------------------------------------------------------------------------- INPUT FILE ^^^^^^^^^^ &CPMD MOLECULAR DYNAMICS MAXSTEP 5000 STORE 10 TIMESTEP 5.0 EMASS 400.0 NOSE IONS 300 2000 NOSE IONS 0.0002 5000 QUENCH BO STRUCTURE BONDS ANGLES &END &SYSTEM ANGSTROM CELL 8.0 1.0 1.0 0.0 0.0 0.0 SYMMETRY 0 CUTOFF 70.0 &END &DFT FUNCTIONAL BLYP GC-CUTOFF 1.0D-8 &END &ATOMS *C_MT_GIA_BLYP KLEINMAN-BYLANDER LMAX=P 1 0.0 0.0 0.0 6 0 *O_MT_GIA_BLYP KLEINMAN-BYLANDER LMAX=P 2 -1.188 0.00 0.00 8 0 1.188 0.00 0.00 8 0 &END ------------------------------------------------------------------- PARTIAL OUTPUT SHOWING THE GENERATION OF NANQ DURING MD ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 11 .00000 .1 -37.64491 -37.64490 -37.64490 .000 2.57 12 .00000 .1 -37.64491 -37.64490 -37.64490 .000 2.69 13 .00000 .1 -37.64491 -37.64490 -37.64490 .000 2.41 14 .00000 .1 -37.64490 -37.64489 -37.64489 .000 2.55 15 .00001 .1 -37.64487 -37.64485 -37.64485 .000 2.60 16 .00002 .1 -37.64475 -37.64473 -37.64471 .000 2.61 17 .00006 .1 -37.64431 -37.64429 -37.64423 .000 2.65 18 .00023 .1 -37.64275 -37.64274 -37.64251 .000 2.66 19 .00082 .1 -37.63717 -37.63715 -37.63634 .000 2.45 20 .00295 .1 -37.61712 -37.61710 -37.61415 .000 2.33 RESTART INFORMATION WRITTEN ON FILE ./RESTART.1 21 .01063 .1 -37.54493 -37.54491 -37.53428 .000 2.34 22 .03834 .1 -37.28448 -37.28447 -37.24613 .000 2.37 23 .13800 .0 -36.34269 -36.34267 -36.20467 .000 2.79 24 .49550 .0 -32.90880 -32.90878 -32.41328 .000 3.03 25 1.73934 NaNQ -19.91875 NaNQ NaNQ .000 2.66 26 NaNQ NaNQ NaNQ NaNQ NaNQ NaNQ 2.39 From kallies at zib.de Fri Jul 25 10:30:09 2003 From: kallies at zib.de (Bernd Kallies) Date: 25 Jul 2003 10:30:09 +0200 Subject: [CPMD-list] NaNQ on P690 & on P630 In-Reply-To: <3F20DEF6.4CC44457@jncasr.ac.in> References: <3F026199.C704E67D@jncasr.ac.in> <200307021127.h62BRPa32107@magadino.cscs.ch> <3F040E2B.A3FA5984@jncasr.ac.in> <20030703143524.GC81396@uranus.chemie.hu-berlin.de> <3F20DEF6.4CC44457@jncasr.ac.in> Message-ID: <1059121809.1237.9.camel@kallies.zib.de> Hi, I noticed a similar error, which results from FFT done with IBM's ESSL. You could try to use CPMD's own FFT instead. Do do so, recompile with CPPFLAGS = -D__IBM -DESSL -DPARALLEL=PARALLEL -DMP_LIBRARY=__MPI -DFFT_DEFAULT There is also a certain probability that the xlf8.1 you used is buggy. Try to use xlf7, or install the latest patch, which results in xlf8.1.1. On Fri, 2003-07-25 at 09:40, Dr. Balasubramanian Sundaram wrote: > Hi All, > > I use CPMD-3.7.1. > > I am trying to run a MD run of one CO_2 molecule on IBM's P690 > (and also tried on P630 with identical experience). The job runs > fine for Geometry Optimization. When I do a MD run, the job runs fine > for a few steps, and then the electron kinetic energy increases to > large values and I get a NaNQ. > > Also note that, I do a "QUENCH BO" before the MD starts, as is > required. > > I used to run this in Parallel with a Makefile generated from the > Configure script for option, "IBM-SP4-SMP". In the Makefile, I removed > the "-qsmp=omp" option, as my machine does not have OpenMP support. > I got a NaNQ for this run. > > So, I stripped every option (such as -O3, -qtune, -qarch etc..), and > also changed the compiler to just "xlf_r" instead of "mpxlf_r", i.e., > basically produced a serial code. (also removed the appropriate > CPPFLAGS). > Yet, I got the NaNQ at the exact step number. > > Importantly, the input file for the job appears to be fine. I used it > to run the CPMD job on a Intel PC/IFC platform, where it runs fine for > several thousands of steps. > > Here are some details of the P630: > AIX 5L, v5.1, xlf v8.1, ESSL v3.3, Lapack v3.0, PWR4 CPU. > > I have tried changing the pseudopotentials also (between USPP and > MT_BLYP > for oxygen). > > We compiled lapack ourselves. Is there something specific to be done > in > compiling lapack on IBMs? (We tried running the code with & without the > CCI > library also; Both ways produced NaNQs). > > I would very much appreciate if you can point out any mistakes that > I have > been doing. It has been a nerve-wracking week. > > I give below 3 files: (a) Relevant portion of the Makefile, (b) Input > file, > and (c) Portion of the output. > > Many Thanks, > > Best, > Bala > http://www.jncasr.ac.in/bala > > -------------------Makefile used on > P690------------------------------------- > > FFLAGS = -qmaxmem=32768 -qtune=pwr4 -qarch=pwr4 > LFLAGS = -L/export2/usr/lib -llapack -lesslsmp -llapack \ > -bbinder:/usr/lib/bind -bmaxdata:2048000000 -qarch=pwr4 > $(QMMM_LIBS) > CFLAGS = > CPP = /usr/ccs/lib/cpp -P > CPPFLAGS = -D__IBM -DLAPACK -DPARALLEL=PARALLEL -DMP_LIBRARY=__MPI > -DFFT_ESSL > NOOPT_FLAG = > CC = cc > FC = mpxlf_r -c > LD = mpxlf_r > AR = /usr/bin/ar > ------------------------------------------------------------------------------- > INPUT FILE > ^^^^^^^^^^ > &CPMD > MOLECULAR DYNAMICS > MAXSTEP > 5000 > STORE > 10 > TIMESTEP > 5.0 > EMASS > 400.0 > NOSE IONS > 300 2000 > NOSE IONS > 0.0002 5000 > QUENCH BO > STRUCTURE BONDS ANGLES > &END > > &SYSTEM > ANGSTROM > CELL > 8.0 1.0 1.0 0.0 0.0 0.0 > SYMMETRY > 0 > CUTOFF > 70.0 > &END > > &DFT > FUNCTIONAL BLYP > GC-CUTOFF > 1.0D-8 > &END > > &ATOMS > *C_MT_GIA_BLYP KLEINMAN-BYLANDER > LMAX=P > 1 > 0.0 0.0 0.0 6 0 > *O_MT_GIA_BLYP KLEINMAN-BYLANDER > LMAX=P > 2 > -1.188 0.00 0.00 8 0 > 1.188 0.00 0.00 8 0 > &END > ------------------------------------------------------------------- > > PARTIAL OUTPUT SHOWING THE GENERATION OF NANQ DURING MD > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > 11 .00000 .1 -37.64491 -37.64490 -37.64490 > .000 2.57 > 12 .00000 .1 -37.64491 -37.64490 -37.64490 > .000 2.69 > 13 .00000 .1 -37.64491 -37.64490 -37.64490 > .000 2.41 > 14 .00000 .1 -37.64490 -37.64489 -37.64489 > .000 2.55 > 15 .00001 .1 -37.64487 -37.64485 -37.64485 > .000 2.60 > 16 .00002 .1 -37.64475 -37.64473 -37.64471 > .000 2.61 > 17 .00006 .1 -37.64431 -37.64429 -37.64423 > .000 2.65 > 18 .00023 .1 -37.64275 -37.64274 -37.64251 > .000 2.66 > 19 .00082 .1 -37.63717 -37.63715 -37.63634 > .000 2.45 > 20 .00295 .1 -37.61712 -37.61710 -37.61415 > .000 2.33 > > RESTART INFORMATION WRITTEN ON FILE ./RESTART.1 > 21 .01063 .1 -37.54493 -37.54491 -37.53428 > .000 2.34 > 22 .03834 .1 -37.28448 -37.28447 -37.24613 > .000 2.37 > 23 .13800 .0 -36.34269 -36.34267 -36.20467 > .000 2.79 > 24 .49550 .0 -32.90880 -32.90878 -32.41328 > .000 3.03 > 25 1.73934 NaNQ -19.91875 NaNQ NaNQ > .000 2.66 > 26 NaNQ NaNQ NaNQ NaNQ NaNQ > NaNQ 2.39 > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list -- Dr. Bernd Kallies Konrad-Zuse-Zentrum f?r Informationstechnik Berlin Takustr. 7 14195 Berlin Tel: +49-30-84185-270 Fax: +49-30-84185-311 e-mail: kallies at zib.de From bala at jncasr.ac.in Fri Jul 25 10:41:40 2003 From: bala at jncasr.ac.in (Dr. Balasubramanian Sundaram) Date: Fri, 25 Jul 2003 14:11:40 +0530 Subject: [CPMD-list] NaNQ on P690 & on P630 References: <3F026199.C704E67D@jncasr.ac.in> <200307