From a.marmier at bath.ac.uk Fri Jul 1 16:03:21 2005 From: a.marmier at bath.ac.uk (arnaud marmier) Date: Fri, 01 Jul 2005 15:03:21 +0100 Subject: [CPMD-list] pthread compilation problems Message-ID: <42C54D29.2000509@bath.ac.uk> Hello, I'm trying to compile CPMD-3.9.2 on an opteron I got a sequential version easily enough, but am having trouble with the MPI. Essentially I get the following kind of errors : /usr/local/Cluster-Apps//mpi/gm/pgi/64//1.2.6..14a.1/lib/libmpich.a(gmpi_malloc.o)(.text+0x48c7): In function `mALLOPt': : undefined reference to `pthread_mutex_lock' /usr/local/Cluster-Apps//mpi/gm/pgi/64//1.2.6..14a.1/lib/libmpich.a(gmpi_malloc.o)(.text+0x49ee): In function `.st3894': : undefined reference to `pthread_mutex_unlock' make: *** [cpmd.x] Error 2 After checking a bit more it seems that the mpich library has been compiled using threads and that I need to link to the threading library. I have tried adding -lpthread to the FFLAGS, even copying some of the libpthread.a libs in the SOURCE directory, but to no avail. I include the Makefile Header for info : #---------------------------------------------------------------------------- # Makefile for cpmd.x (plane wave electronic calculation) # Configuration: PGI-AMD64-MPI # Creation of Makefile: Jul 1 2005 # on Linux mott2 2.4.21-32.0.1.ELsmp #1 SMP Tue May 17 17:46:36 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux # Author: marmier #---------------------------------------------------------------------------- # SHELL = /bin/sh # #--------------- Default Configuration for PGI-AMD64-MPI --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = -r8 -pc=64 -Msignextend LFLAGS = -Bstatic -L. -latlas_x86-64 $(QMMM_LIBS) CFLAGS = CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__Linux -D__PGI -DFFT_DEFAULT -DPOINTER8 -D__pgf77 \ -DPARALLEL -DMYRINET NOOPT_FLAG = CC = mpicc -O2 -Wall -m64 FC = mpif77 -c -fastsse -tp k8-64 LD = mpif77 -fastsse -tp k8-64 AR = ar Thnaks in advance, Best wishes, Arnaud From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Fri Jul 1 16:45:48 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Fri, 1 Jul 2005 16:45:48 +0200 (CEST) Subject: [CPMD-list] pthread compilation problems In-Reply-To: <42C54D29.2000509@bath.ac.uk> Message-ID: On Fri, 1 Jul 2005, arnaud marmier wrote: AM> Hello, hi arnaud, AM> I got a sequential version easily enough, but am having trouble with AM> the MPI. AM> AM> Essentially I get the following kind of errors : AM> AM> /usr/local/Cluster-Apps//mpi/gm/pgi/64//1.2.6..14a.1/lib/libmpich.a(gmpi_malloc.o)(.text+0x48c7): AM> In function `mALLOPt': AM> : undefined reference to `pthread_mutex_lock' AM> /usr/local/Cluster-Apps//mpi/gm/pgi/64//1.2.6..14a.1/lib/libmpich.a(gmpi_malloc.o)(.text+0x49ee): AM> In function `.st3894': AM> : undefined reference to `pthread_mutex_unlock' AM> make: *** [cpmd.x] Error 2 AM> AM> After checking a bit more it seems that the mpich library has been AM> compiled using threads and that I need to link to the threading library. AM> I have tried adding -lpthread to the FFLAGS, even copying some of the AM> libpthread.a libs in the SOURCE directory, but to no avail. you need to add -lpthread to LFLAGS. FFLAGS is for compiling, LFLAGS for linking. two more remarks: given your setup, you also may want to remove -Bstatic, or else you may run into trouble when the threads library is updated. finally, you may want to try out replacing -fastsse with -fast or just with -O2 -Munroll . depending on your setup and hardware, using a slightly lower level of optimization may actually produce a faster executable. especially using the SSE unit seems not to be very helpful for a large numerical code. best regards, axel. AM> AM> I include the Makefile Header for info : AM> AM> #---------------------------------------------------------------------------- AM> # Makefile for cpmd.x (plane wave electronic calculation) AM> # Configuration: PGI-AMD64-MPI AM> # Creation of Makefile: Jul 1 2005 AM> # on Linux mott2 2.4.21-32.0.1.ELsmp #1 SMP Tue May 17 17:46:36 EDT 2005 AM> x86_64 AM> x86_64 x86_64 GNU/Linux AM> # Author: marmier AM> #---------------------------------------------------------------------------- AM> # AM> SHELL = /bin/sh AM> # AM> #--------------- Default Configuration for PGI-AMD64-MPI --------------- AM> SRC = . AM> DEST = . AM> BIN = . AM> #QMMM_FLAGS = -D__QMECHCOUPL AM> #QMMM_LIBS = -L. -lmm AM> FFLAGS = -r8 -pc=64 -Msignextend AM> LFLAGS = -Bstatic -L. -latlas_x86-64 $(QMMM_LIBS) AM> CFLAGS = AM> CPP = /lib/cpp -P -C -traditional AM> CPPFLAGS = -D__Linux -D__PGI -DFFT_DEFAULT -DPOINTER8 -D__pgf77 \ AM> -DPARALLEL -DMYRINET AM> NOOPT_FLAG = AM> CC = mpicc -O2 -Wall -m64 AM> FC = mpif77 -c -fastsse -tp k8-64 AM> LD = mpif77 -fastsse -tp k8-64 AM> AR = ar AM> AM> AM> Thnaks in advance, AM> AM> Best wishes, AM> Arnaud AM> AM> AM> _______________________________________________ AM> CPMD-list mailing list AM> CPMD-list at cpmd.org AM> http://cpmd.org/mailman/listinfo/cpmd-list AM> AM> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From a.marmier at bath.ac.uk Fri Jul 1 16:54:55 2005 From: a.marmier at bath.ac.uk (arnaud marmier) Date: Fri, 01 Jul 2005 15:54:55 +0100 Subject: [CPMD-list] pthread compilation problems References: Message-ID: <42C5593F.6030100@bath.ac.uk> Hello Axel, > two more remarks: > given your setup, you also may want to remove -Bstatic, or else > you may run into trouble when the threads library is updated. Thanks, removing -Bstatic was actually enough for successul compilation (i.e. no need for -lpthread) Thanks a million, your support for CPMD is really first class ! (acually better than that) Back to lurking (hopefully) Ciao arnaud From hutter at pci.unizh.ch Sat Jul 2 15:27:26 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 2 Jul 2005 15:27:26 +0200 (MEST) Subject: [CPMD-list] Dipole dynamics vs standard autocorrelation spectrum? In-Reply-To: References: Message-ID: Hi > Hello all, > > I perform my CPMD calculation without DIPOLE DYNAMICS keyword, so now i > don't have DIPOLE file and that implicate - I can calculate only "power > spectrum". This spectrum is not IR spectrum, but rather all the > vibrational degrees of freedom of the system. I suppose there is no > possibility to obtain this DIPOLE file now. I think only possibility > to get this file is to reapet this calculation with DIPOLE DYNAMICS > keyword or maybe there is easier way? Yes, you have to repaet at least part of the calculation. You probably don't want to calculate the dipole at every CP step, so you could only recalculate every 5th to 10th step using a clever script. There is also the MOLECULAR DYNAMICS PT keyword that is redoing a MD from a TRAJECTORY file. However, I don't know what properties are available with this option and how to correctly use the option. > > >Perform a similar analysis on the dypole dynamics (keyword DIPOLE > >DYNAMICS in a MD simulation) that is recorded in the file DIPOLE. In this > >case you have the IR bands. Again you do not get the intensities. > > And now I confused. > > 1. DIPOLE DYNAMICS - I get IR band - really? But why I don't get the > intensities? Are maybe some free scripts available to compute spectrum > when I have DIPOLE file? You get the intensities, apart from some constants. Have a look into the papers by Bernasconi et al. or Sprik et al. For the derivation of the formula look into McQuarrie "Statistical Mechanics". > > 2. Normall I compute standard autocorrelation spectrum whitout DIPOLE. > How bad is this power spectrum when I compared it with DIPOLE spectrum? > Are some papers which deal with this topic? > Are intensities in standard autocorrelation spectrum really very bad > or maybe ? > > Again see McQuarrie. Intensities from the velocity autocorrelation is not related to the absorption intensity. > > WANG wrote also: > >Calculate the power spectrum from a MD calculation. In this way you get > >rid of harmonic approximation, but you need a very very very long run to > >have a sufficient accuracy in numerical data. > > > > What means very very long run ---> 50000 steps or 70000 steps. Is this > enough? > > Long wrt to the important frequencies you want to study. The lower the frequencies the longer you have to sample. I would take about 20ps for frequencies above 1000 wavenumbers as sufficient. > 3. I computed power spectrum from Path-Integral dynamic. > I'm interesting in hydrogen bond. In this dynamics I get center > position of hydrogen in hydrogen bridge --> standard autocorrelation > spectrum show continuum of absorbtion 1700 cm-1 to 800 cm-1, in the > standard Car-Parrinello Dynamics spectrum I didn't got this continuum. > Is this contunuum maybe effect of PI dynamics(I did the same what Axel > Kohlmayer script - I computed average of replicas - obtained centroid - > so spectrum of centroid). > I can't answer this questions. But my guess is that the velocity autocorrelation from a path integral MD calculation is not (at least simply) related to the IR spectra. regards Juerg Hutter From hutter at pci.unizh.ch Sat Jul 2 15:36:24 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 2 Jul 2005 15:36:24 +0200 (MEST) Subject: [CPMD-list] Calc of the band gap on the fly during MD In-Reply-To: <1119903965.42c060ddbea52@webmail.mail.gatech.edu> References: <1119903965.42c060ddbea52@webmail.mail.gatech.edu> Message-ID: Hi > Dear List Subscribers, > > I am wondering whether one can calculate unfilled KS states (and thus the > LUMO-HOMO band gap) during an MD (CP) run. By combining PRINT EIGENVALUES with > STATES = #(filled states), I was able to get the filled KS states on the fly. > Simply adding unfilled states, however, doesn't seem legitimate. Would there be > a way to avoid running a separate diagonalization at each time to get unfilled > states? Thanks very much for reading. No, there is no possibility to calculate unoccupied states during a CP run easyly. It's best to resample the trajectory and optimize the wavefunction for special geometries. I would suggest to use a two step procedure 1. optimize the wavefunction using DIIS (only occupied states) 2. restart the calculation and diagonalize the KS matrix for the unoccupied states If you plan to do this type of calculations frequently, you could also adopt CPMD to do this automatically during a CP-MD simulation. regards Juerg Hutter > > Best regards, > > EG Kim > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Sat Jul 2 15:44:24 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 2 Jul 2005 15:44:24 +0200 (MEST) Subject: [CPMD-list] Pt slab calculation error In-Reply-To: <9E4557B19235FE47B1C7B52DE3A536A33F962F@MBOX23.stu.nus.edu.sg> References: <9E4557B19235FE47B1C7B52DE3A536A33F962F@MBOX23.stu.nus.edu.sg> Message-ID: Hi > Thanks to J. Hutter. I could get Pt lattice constant using TM_PBE PPs > 7.4776 bohr with k-points 8x8x8 > > This value is in the middle of experimental value and other DFT code's > value. > > Do you think changing k-point makes a big difference in lattice > constant, for example, 4x4x4 or 12x12x12? > > > > In addition, I'm trying to calculate Pt slab. > > But while running, it met an error. > > > > **************************************** > > PARAMETER NHX TOO SMALL 176 100 > I'm surprised you get this error only in the slab calculation. The problem is related with the maximum number of projectors (100) for an atomic type. You are using a PP with nonlocal functions (= projectors) on s, d, and f angular momentum. In Kleinman-Bylander for this would be 13 projectors. But in Gauss-Hermite this has to be multiplied by the number of integration points (you ask for 15, might be slightly reduced by the program). You have two options 1) check if the Kleinman-Bylander transformation of the PP works 2) increase the value of NHX in system.h and recompile CPMD regards Juerg Hutter > > > PROGRAM STOPS AT SUBROUTINE SETSYS > > > > P0_17753 p4_error 999 > > > > **************************************** > > What's wrong in my input file? > > For reference, I'll attach my input and output file. > > > > Thanks in advance, > > > > HONG won keon > > From hutter at pci.unizh.ch Sat Jul 2 15:49:17 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 2 Jul 2005 15:49:17 +0200 (MEST) Subject: [CPMD-list] More on NOSE Electrons.. In-Reply-To: <94e5a4d1a1bd476a2d9437a568abca34@bris.ac.uk> References: <94e5a4d1a1bd476a2d9437a568abca34@bris.ac.uk> Message-ID: Hi > Thank you, Christl, for pointing out the Blochl and Parrinello (1992, > PRB, 45:9413) paper to me. > > As I understand it, for the long simulations I'm trying (~ 5 ps), I > certainly want to thermostat the fictitious kinetic > energy of the electronic wavefunctions. Blochl and Parrinello point > out that periodic quenching to the BO surface will screw up the > dynamics unless you do it at every timestep. > > In equation 6 of the Blochl and Parrinello (1992; PRB) paper, they > suggest the target fictitious kinetic energy of the electronic wave > functions to be > > 2kbT(mu/M) SUM > > where, as I see it, the summation (SUM) is simply > the total real electronic kinetic energy in the ground state (in A.U), > mu is the fictitious mass of the wavefunction, and M is the mass of one > atom. However, since kbT has units of energy, then the whole > expression has units of energy**2. This can't be right; is there a > typo or am I confused about something..? This is the "beauty" of atomic units. The sum in the formula is the electronic kinetic energy, but only in atomic units. If you use another system there would be some constants with appropriate units. Please also notice that mu hasn't simply units of a mass! regards Juerg Hutter > > Thanks for your help! > > Dave Sherman > ______________ > David M. Sherman > Professor of Geochemistry > Department of Earth Sciences > University of Bristol > Bristol BS8 1RJ UNITED KINGDOM > Phone: 44-(0)117-954-5446 > http://mineral.gly.bris.ac.uk > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > From mclar at elrond.chem.uni.wroc.pl Sat Jul 2 22:21:54 2005 From: mclar at elrond.chem.uni.wroc.pl (Przemyslaw Dopieralski) Date: Sat, 2 Jul 2005 22:21:54 +0200 (CEST) Subject: [CPMD-list] Path-Integral with crystals In-Reply-To: Message-ID: Hi, I use CPMD version 3.9.2 So not here the problem is. I use the standard input with few changes to compute path-integral with crystal. Probably, like allways :) problem is in molekules inside the box. Maybe I made a mistake in geometry - but the other way optimization run correctly. My input(geometry omitted) &CPMD PATH INTEGRALS MOLECULAR DYNAMICS CP RESTART WAVEFUNCTION COORDINATES LATEST ODIIS 5 TRAJECTORY SAMPLE 1 TEMPERATURE 120.d0 NOSE ELECTRONS 2.d-4 15000.d0 NOSE IONS 120 3000 CONVERGENCE ORBITALS 1.0E-06 MAXSTEP 70000 TIMESTEP 3.d0 STORE 2000 &END &DFT FUNCTIONAL PBE &END &SYSTEM POISSON HOCKNEY PARAMETER 0.06d0 TESR 2 ANGSTROM SYMMETRY TRICLINIC CELL ABSOLUTE DEGREE 5.927 6.288 11.222 104.0 91.47 100.26 CUTOFF 100.0 &END ----->GEOMETRY &PIMD TROTTER DIMENSION 3 NORMAL MODES 1.d0 FACMASS 1.d0 DEBROGLIE CENTROID 500.d0 PRINT LEVEL 1 &END By the way I want to ask: I convert cif file to (first to ml2) xyz - and somtimes I have a problem with axis orientation and then I change this orientation by my self. In this particular sample I didn't have this problem but it happens. Is there maybe better way to obtain xyz file from cif file? But the other way standard CPMD run correct for this system, only problem is the Path-Integral. thanks best regards, PhD student Przemyslaw Dopieralski University of Wroclaw POLAND mclar at elrond.chem.uni.wroc.pl > On Thu, 30 Jun 2005, Salomon Billeter wrote: > > > Hi, > > > > > First you have to be careful about which version of > > > cpmd you are using.... V3.9 has a bug particularly if > > > you are using PBE in a spin polarised calculation. > > > > > > If you are not already doing so I'd advise you use the > > > latest version (which I think has all the problems > > > fixed). > > > > just to avoid confusion with versions 3.9 / 3.9.x: CPMD version 3.9.2 > is > > fine. Versions up to v3.9.1 had a small problem with the gradient > > correction at very low densities for spin-polarized PBE. However, I > don't > > think your problem is caused by this. > > > > Best regards, > > Salomon > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://cpmd.org/mailman/listinfo/cpmd-list > > From ZRLSRB at ch.ibm.com Sat Jul 2 23:59:54 2005 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Sat, 2 Jul 2005 23:59:54 +0200 Subject: [CPMD-list] Path-Integral with crystals Message-ID: Hi Przemyslaw, > Probably, like allways :) problem is in molekules inside the box. > Maybe I made a mistake in geometry - but the other way optimization run > correctly. you once earlier wrote: > For PI calculations for isolated molecule - no problem, but for crystal > temperature is growing really fast. I used Nose-Hoover thermostat - but it > did help. Geometry was good - I optimised this crystal. but in your input, you specify: > POISSON HOCKNEY PARAMETER > 0.06d0 The Hockney method solves Poisson's equation for non-periodic systems (it decouples periodic images). The statement must be removed for periodic systems (even technically, it will not work properly because, unlike the reciprocal-space Poisson solvers, the position of your system in the box matters - it must be centered and the density must disappear at the edges). Best regards, Salomon From mclar at elrond.chem.uni.wroc.pl Sun Jul 3 12:16:58 2005 From: mclar at elrond.chem.uni.wroc.pl (Przemyslaw Dopieralski) Date: Sun, 3 Jul 2005 12:16:58 +0200 (CEST) Subject: [CPMD-list] Path-Integral with crystals In-Reply-To: Message-ID: Hi Salomon, Oh oh oh oh what a mistake. You've got absolutely right. Fortunately fot crystals whitout path-integral method i didn't use this solver. So now I'll run once again wavefunction optimization for Path Integral and then we find out if it helps. But I'm sure it was the problem. I'm really greatful for your help Once again - thanks. best regards, Przemek Dopieralski University of Wroclaw Poland mclar at elrond.chem.uni.wroc.pl On Sat, 2 Jul 2005, Salomon Billeter wrote: > Hi Przemyslaw, > > > Probably, like allways :) problem is in molekules inside the box. > > Maybe I made a mistake in geometry - but the other way optimization run > > correctly. > > you once earlier wrote: > > > For PI calculations for isolated molecule - no problem, but for crystal > > temperature is growing really fast. I used Nose-Hoover thermostat - but > it > > did help. Geometry was good - I optimised this crystal. > > but in your input, you specify: > > > POISSON HOCKNEY PARAMETER > > 0.06d0 > > The Hockney method solves Poisson's equation for non-periodic systems (it > decouples periodic images). The statement must be removed for periodic > systems (even technically, it will not work properly because, unlike the > reciprocal-space Poisson solvers, the position of your system in the box > matters - it must be centered and the density must disappear at the edges). > > Best regards, > Salomon > > From ZRLSRB at ch.ibm.com Mon Jul 4 09:47:52 2005 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Mon, 4 Jul 2005 09:47:52 +0200 Subject: [CPMD-list] Dipole dynamics vs standard autocorrelation spectrum? Message-ID: Hi Przemyslaw, > 3. I computed power spectrum from Path-Integral dynamic. > I'm interesting in hydrogen bond. In this dynamics I get center > position of hydrogen in hydrogen bridge --> standard autocorrelation > spectrum show continuum of absorbtion 1700 cm-1 to 800 cm-1, in the > standard Car-Parrinello Dynamics spectrum I didn't got this continuum. > Is this contunuum maybe effect of PI dynamics(I did the same what Axel > Kohlmayer script - I computed average of replicas - obtained centroid - > so spectrum of centroid). Did you use ab initio centroid molecular dynamics (keyword CENTROID DYNAMICS in the &PIMD section; see Marx, Tuckerman and Martyna in Computer Physics Communications 118, 166 (1999))? Otherwise, real-time information is not readily available in the "dynamics" of the centroid. In its standard formulation (imaginary time), path integral molecular dynamics does not represent - despite its name - the real-time dynamics of the system but instead cleverly samples the configurational space of the quantum system using an isomorphism between an n-times discretized path integral partition function of the system considered and a "partition function" of an n-times replicated classical system with scaled and excluded interactions and - importantly - a part of the kinetic energy term appearing as harmonic springs in the potential energy term. Therefore, (again, in the usual imaginary-time formulation), you cannot easily get real-time information such as velocity autocorrelation functions etc. Best regards, Salomon From chokosabe at yahoo.co.uk Mon Jul 4 12:13:30 2005 From: chokosabe at yahoo.co.uk (Abraham Alaka) Date: Mon, 4 Jul 2005 11:13:30 +0100 (BST) Subject: [CPMD-list] Dipole dynamics vs standard autocorrelation spectrum? In-Reply-To: Message-ID: <20050704101330.40641.qmail@web25702.mail.ukl.yahoo.com> I'd add to this and say that you'll find a very good example of a PIMD calculation (similar to the type you want to perform) in the examples section. I cant remember which exactly (its been a while). If you are new to MD (molecular dynamics) it might be worth having a look at some of these examples; they might give you ideas you might find useful. Hope this is useful.. A. --- Salomon Billeter wrote: > Hi Przemyslaw, > > > 3. I computed power spectrum from Path-Integral > dynamic. > > I'm interesting in hydrogen bond. In this > dynamics I get center > > position of hydrogen in hydrogen bridge --> > standard autocorrelation > > spectrum show continuum of absorbtion 1700 cm-1 to > 800 cm-1, in the > > standard Car-Parrinello Dynamics spectrum I didn't > got this continuum. > > Is this contunuum maybe effect of PI dynamics(I > did the same what Axel > > Kohlmayer script - I computed average of replicas > - obtained centroid - > > so spectrum of centroid). > > Did you use ab initio centroid molecular dynamics > (keyword CENTROID > DYNAMICS in the &PIMD section; see Marx, Tuckerman > and Martyna in Computer > Physics Communications 118, 166 (1999))? Otherwise, > real-time information > is not readily available in the "dynamics" of the > centroid. > > In its standard formulation (imaginary time), path > integral molecular > dynamics does not represent - despite its name - the > real-time dynamics of > the system but instead cleverly samples the > configurational space of the > quantum system using an isomorphism between an > n-times discretized path > integral partition function of the system considered > and a "partition > function" of an n-times replicated classical system > with scaled and > excluded interactions and - importantly - a part of > the kinetic energy term > appearing as harmonic springs in the potential > energy term. Therefore, > (again, in the usual imaginary-time formulation), > you cannot easily get > real-time information such as velocity > autocorrelation functions etc. > > Best regards, > Salomon > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > ___________________________________________________________ Yahoo! Messenger - NEW crystal clear PC to PC calling worldwide with voicemail http://uk.messenger.yahoo.com From mclar at elrond.chem.uni.wroc.pl Mon Jul 4 12:39:12 2005 From: mclar at elrond.chem.uni.wroc.pl (Przemyslaw Dopieralski) Date: Mon, 4 Jul 2005 12:39:12 +0200 (CEST) Subject: [CPMD-list] Dipole dynamics vs standard autocorrelation spectrum? In-Reply-To: Message-ID: Hi Salomon, I didn't use CENTROID DYNAMICS. Last time I sent you my input which i use. Maby unfortunately I said: centroid. But on the othe side this is a some kind of centroind because I obtain it from (in my case) 3 replicas. So I interpet only this average (I add up position and velocities vector in every step and then I devide it by 3) of this 3 replicas. But if I understand you well, I cannot obtain so easily autocorrelation spectrum? Now the question is - how to do it? &PIMD TROTTER DIMENSION 3 NORMAL MODES 1.d0 FACMASS 1.d0 DEBROGLIE CENTROID 500.d0 PRINT LEVEL 1 &END best regards, Przemek On Mon, 4 Jul 2005, Salomon Billeter wrote: > Hi Przemyslaw, > > > 3. I computed power spectrum from Path-Integral dynamic. > > I'm interesting in hydrogen bond. In this dynamics I get center > > position of hydrogen in hydrogen bridge --> standard autocorrelation > > spectrum show continuum of absorbtion 1700 cm-1 to 800 cm-1, in the > > standard Car-Parrinello Dynamics spectrum I didn't got this continuum. > > Is this contunuum maybe effect of PI dynamics(I did the same what Axel > > Kohlmayer script - I computed average of replicas - obtained centroid - > > so spectrum of centroid). > > Did you use ab initio centroid molecular dynamics (keyword CENTROID > DYNAMICS in the &PIMD section; see Marx, Tuckerman and Martyna in Computer > Physics Communications 118, 166 (1999))? Otherwise, real-time information > is not readily available in the "dynamics" of the centroid. > > In its standard formulation (imaginary time), path integral molecular > dynamics does not represent - despite its name - the real-time dynamics of > the system but instead cleverly samples the configurational space of the > quantum system using an isomorphism between an n-times discretized path > integral partition function of the system considered and a "partition > function" of an n-times replicated classical system with scaled and > excluded interactions and - importantly - a part of the kinetic energy term > appearing as harmonic springs in the potential energy term. Therefore, > (again, in the usual imaginary-time formulation), you cannot easily get > real-time information such as velocity autocorrelation functions etc. > > Best regards, > Salomon > > From jinhm at ihpc.a-star.edu.sg Mon Jul 4 12:49:56 2005 From: jinhm at ihpc.a-star.edu.sg (Jin Hong Mei) Date: Mon, 04 Jul 2005 18:49:56 +0800 Subject: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] Message-ID: <42C91454.3050601@ihpc.a-star.edu.sg> A collegue of mine has the following questions for the list: --------------------------- Dear Sirs, I am Dave Lin, Systems Administrator from the Institute of High Performance Computing, Singapore. I tried to install CPMD 3.9.2 for our researchers and encountered problems. My environment are as follows : HARDWARE 7 nodes of IBM p690 model 681 each with 32 1.3 GHz POWER4 processors. SP II switch connect all 7 nodes together. SOFTWARE AIX 5L version 5.1 ML04 IBM XL Fortran Compiler version 8.1 IBM Visuage C/C++ Compiler version 6.0 IBM ESSL version 3.3 IBM Parallel ESSL version 3.3 IBM POE version 3.2 I have tried both IBM-SP4-SMP-64 and IBM-PWR4-AIX-ESSL-MPI-SMP (in seperate directories) as arguments of mkconfig.sh. Attached partial Makefile is the one generated with IBM-PWR4-AIX-ESSL-MPI-SMP (I have changed all pwr5 to pwr4). In both cases, the installation results are similar. Make encountered error as shown in the attached file 'make_error' and terminated. It seems like source coding error to me so I modify the source g_loc_opeigr.F (see attached modified source in 'g_loc_opeigr.F') and finally make successfully. I tried running the newly created cpmd.x with the command : /apps/cpmd/CPMD-3.9.2/cpmd.x test-v1_inp > output The programs run for about 10 minutes and the ouput (see attached file 'output') showed some error. The error is identical no matter IBM-SP4-SMP-64 or IBM-PWR4-AIX-ESSL-MPI-SMP are chosen as the plateform. I tried running the same input file with our previous CPMD 3.9.1 and it has been running for 19 hours and still in progress. The partial output shows that it has passed the error point of the 3.9.2 run. Please kindly advise me how to rectify the problem. Thank you very much. dave lin -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: test-v1_inp Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050704/80c51ab7/attachment.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: g_loc_opeigr.F Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050704/80c51ab7/attachment-0001.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: make_error Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050704/80c51ab7/attachment-0002.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: Makefile Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050704/80c51ab7/attachment-0003.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: output Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050704/80c51ab7/attachment-0004.cc From ZRLSRB at ch.ibm.com Mon Jul 4 14:36:10 2005 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Mon, 4 Jul 2005 14:36:10 +0200 Subject: [CPMD-list] Dipole dynamics vs standard autocorrelation spectrum? Message-ID: Hi Przemyslaw, > I didn't use CENTROID DYNAMICS. > Last time I sent you my input which i use. > Maby unfortunately I said: centroid. your statement was right, you referred to the centroid. > But on the othe side this is a some kind of centroind because I obtain it > from (in my case) 3 replicas. So I interpet only this average (I add up > position and velocities vector in every step and then I devide it by 3) of > this 3 replicas. But if I understand you well, I cannot obtain so easily autocorrelation spectrum? > Now the question is - how to do it? Although you used the right variable, the centroid, imaginary-time path integral MD does not directly produce real-time dynamics of the centroid for the reasons I mentioned. In principle, you can use an analytic continuation to derive time-dependent properties from the ensemble generated (cf. canonical density operator \exp(-\beta\hat H)/Z vs. propagator \exp(-i\hat H t)), but in practice, you'll want to use the adiabatic ab initio centroid molecular dynamics method. You can read about this method (keyword CENTROID DYNAMICS in the &PIMD section) in Marx, Tuckerman and Martyna, Computer Physics Communications 118, 166 (1999). The manual gives details about specific requirements of the method such as normal mode propagator (which you already use) etc. Best regards, Salomon From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Mon Jul 4 18:28:00 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Mon, 4 Jul 2005 18:28:00 +0200 (CEST) Subject: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] In-Reply-To: <42C91454.3050601@ihpc.a-star.edu.sg> Message-ID: On Mon, 4 Jul 2005, Jin Hong Mei wrote: DL> A collegue of mine has the following questions for the list: DL> --------------------------- DL> DL> Dear Sirs, DL> DL> I am Dave Lin, Systems Administrator from the Institute of High DL> Performance Computing, Singapore. I tried to install CPMD 3.9.2 for our DL> researchers and encountered problems. My environment are as follows : [...] DL> I tried running the same input file with our previous CPMD 3.9.1 and it DL> has been running for 19 hours and still in progress. The partial output DL> shows that it has passed the error point of the 3.9.2 run. DL> DL> Please kindly advise me how to rectify the problem. well, i have several comments to the input and what is detectable from the output. most of that concerns the physics of the problem and not the technical side. first, and this is the technical part, the MPI parallelization in CPMD is in general much better than the OpenMP parallelization, yet the job is run with one MPI node but 32 OpenMP threads. i don't know whether this is causing the problem (i doubt it), but is surely will make the job run _very_ inefficiently. on a machine like yours, it would probably better to run with 32 MPI-tasks and no OpenMP or maybe, 16 MPI-tasks and two OpenMP threads each. second, the input uses vanderbilt ultra-soft pseudopotentials, and for them a plane wave cutoff of 80ry is _huge_ and also _very_ inefficient. something between 20ry and 40ry is usually sufficient. 35ry should be a safe choice. for that kind of plane wave cutoff one may have to increase the GC-CUTOFF value (e.g. to 1.0d-6). also, for vanderbilt uspps sometimes the default initial atomic guess is not very good (see previous discussions on cpmd-list and in the manual), so using INITIALIZE WAVEFUNCTIONS RANDOM or modifying the atomic guess via ATOMIC CHARGES might help. but actually, just with a changed plane wave cutoff it already runs fine. please see attached output, which was, btw, created with 6 dual athlon nodes on a gigabit ethernet using only MPI parallelization. finally, and this should have little to do with the reported problems. i noticed that one oxygen and one manganese atom are _very_ close together and also the CONSTRAINTS part looks strange. with FIX ATOMS one gives the number and then the indices, but on the same line and no coordinates, i.e.: FIX ATOMS 1 3 however, the three following numbers, although ignored by cpmd, coincide with the coordinateshat of the third oxygen (i.e. the one at the strange position). but in this case the input should be (4xLa + 4xMn + 3xO): FIX ATOMS 1 11 on top of that, i'm a bit confused about the purpose of this, since all other atoms will move around it. finally, i'd like to note, that with a supercell of this size you may want to check, whether your k-point sampling with only the gamma point is sufficient. unfortunately, this is bit of a problematic topic, as CPMD currently does not support k-points with ultra-soft pseudopotentials... DL> DL> Thank you very much. DL> DL> dave lin with best regards, axel kohlmeyer. DL> DL> DL> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. -------------- next part -------------- A non-text attachment was scrubbed... Name: test-v1_out.gz Type: application/x-gzip Size: 36840 bytes Desc: Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050704/5ba7c573/attachment.gz From cur at zurich.ibm.com Tue Jul 5 11:17:25 2005 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Tue, 5 Jul 2005 11:17:25 +0200 Subject: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] In-Reply-To: Message-ID: As already stated in previous e-mails on the list - the use of the Mixed MPI/OpenMP scheme it is useful if a) you want to run on many processors ( let's say more processors than the number of realspace planes along x); b) you want to run on clustered smp servers going outside a single smp server - when the interconnect latency and bandwidth are much worse than the the ones for the memory c) you have a system with more than 2000 states. Your case is not in these category - so as explained by Axel - much better to run in MPI only mode. If you have compiled the code for IBM POWER machine in the mixed version , you should specify: export XLSMPOPTS=parthds=1 ---> to specify the number of smp threads export MP_PROCS=32 ---> to specify the number of MPI threads export MP_SHARED_MEMORY=YES In any case - I still suggest to use norm conserving pseudos for your calculation and test with kpoints convergency. Best Regards, Alessandro 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 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20050705/5e36e8ad/attachment.html From i.kozin at dl.ac.uk Tue Jul 5 13:08:51 2005 From: i.kozin at dl.ac.uk (Kozin, I (Igor)) Date: Tue, 5 Jul 2005 12:08:51 +0100 Subject: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] Message-ID: <77673C9ECE12AB4791B5AC0A7BF40C8F0147D5F0@exchange02.fed.cclrc.ac.uk> Alessandro, what is the present "official" view on using mixed MPI/OpenMP on platforms other than IBM Power? Do you think that with arrival of dual-cores it is more practical than ever to start using OpenMP? I'd guess many on the list would be interested to know the developers' view on this. > c) you have a system with more than 2000 states. BTW, are there test cases publicly available (e.g. your silicon carbide example you published the report about) which would be suitable for using the mixed model? Many thanks in advance and all the best, Igor I. Kozin (i.kozin at dl.ac.uk) CCLRC Daresbury Laboratory tel: 01925 603308 http://www.cse.clrc.ac.uk/disco -----Original Message----- From: cpmd-list-bounces at cpmd.org [mailto:cpmd-list-bounces at cpmd.org]On Behalf Of Alessandro Curioni Sent: 05 July 2005 10:17 To: Axel Kohlmeyer Cc: cpmd-list; Lin Yi; cpmd-list-bounces at cpmd.org Subject: Re: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] As already stated in previous e-mails on the list - the use of the Mixed MPI/OpenMP scheme it is useful if a) you want to run on many processors ( let's say more processors than the number of realspace planes along x); b) you want to run on clustered smp servers going outside a single smp server - when the interconnect latency and bandwidth are much worse than the the ones for the memory c) you have a system with more than 2000 states. Your case is not in these category - so as explained by Axel - much better to run in MPI only mode. If you have compiled the code for IBM POWER machine in the mixed version , you should specify: export XLSMPOPTS=parthds=1 ---> to specify the number of smp threads export MP_PROCS=32 ---> to specify the number of MPI threads export MP_SHARED_MEMORY=YES In any case - I still suggest to use norm conserving pseudos for your calculation and test with kpoints convergency. Best Regards, Alessandro 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 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20050705/3c6ed956/attachment.html From cur at zurich.ibm.com Tue Jul 5 16:48:40 2005 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Tue, 5 Jul 2005 16:48:40 +0200 Subject: [CPMD-list] [Fwd: Error Encountered During CPMD-3.9.2 Compilation] In-Reply-To: <77673C9ECE12AB4791B5AC0A7BF40C8F0147D5F0@exchange02.fed.cclrc.ac.uk> Message-ID: Igor, I plan to make the tests you are refeering to available - together with others- for benchmark purposes. The use of the OpenMP on top of MPI - for example- it works well on JS20 Myrinet connected clusters (e.g. MareNostrum) if you want to push the scaling to the 1000 processors regime. In any case - I have already written the main reasons to use the dual parallelism in my earlier note. BTW The report ( by Prof. Hutter and myself) is now published in Parallel Computing (31) 1 (2005). 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 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20050705/be2c55f6/attachment.html From krekeler at mpip-mainz.mpg.de Wed Jul 6 15:15:25 2005 From: krekeler at mpip-mainz.mpg.de (Christian Krekeler) Date: Wed, 06 Jul 2005 15:15:25 +0200 Subject: [CPMD-list] Wannier Reference Message-ID: <42CBD96D.7050404@mpip-mainz.mpg.de> Hi, I've some probleme with the Wannier reference. I calculated the wannier center for a system of water molecules around one chloride. I changed the wannier reference in the input file. For the case wannier reference (0,0,0) and for the other one I calculated the dipole moment of water. The shell size around chloride is different as well as the average from one to another. Now there is my question: What do this command "Wannier reference" exactly? Is it only a shift of coordinates in the IONS+CENTERS file? Thanks in advance.. Christian -- ---------------------------------------------------------- Christian Krekeler Max Planck Institute for Polymer Research Ackermannweg 10 D-55128 Mainz, Germany Phone: +49 6131 379 148 Fax: +49 6131 379 340 E-mail: krekeler at mpip-mainz.mpg.de web: www.mpip-mainz.mpg.de/theory.html ---------------------------------------------------------- From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Wed Jul 6 16:59:37 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Wed, 6 Jul 2005 16:59:37 +0200 (CEST) Subject: [CPMD-list] Wannier Reference In-Reply-To: <42CBD96D.7050404@mpip-mainz.mpg.de> Message-ID: On Wed, 6 Jul 2005, Christian Krekeler wrote: CK> Hi, hi christian, CK> I've some probleme with the Wannier reference. I calculated the wannier CK> center for a system of water molecules around one chloride. I changed CK> the wannier reference in the input file. For the case wannier reference CK> (0,0,0) and for the other one I calculated the dipole moment of water. CK> The shell size around chloride is different as well as the average from stupid question: what do you mean by shell? the average distance of the water molecules from the chloride? if yes, how long is the trajectory you are averaging over? CK> one to another. Now there is my question: CK> What do this command "Wannier reference" exactly? Is it only a shift of CK> coordinates in the IONS+CENTERS file? yes. please have a look at the source code. whatever you provide with the WANNIER REFERENCE keyword is stored in the W_REF() array and this is then only used in the file wannier_center.F to set a reference point around which you apply the periodic boundary conditions for the output of the coordinates (of ions and centers). best regards, axel. CK> Thanks in advance.. CK> CK> Christian CK> CK> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From krekeler at mpip-mainz.mpg.de Wed Jul 6 18:00:08 2005 From: krekeler at mpip-mainz.mpg.de (Christian Krekeler) Date: Wed, 06 Jul 2005 18:00:08 +0200 Subject: [CPMD-list] Wannier Reference In-Reply-To: References: Message-ID: <42CC0008.9060205@mpip-mainz.mpg.de> Hi, thanks for the fast reply, this helps me a lot. This means that the cell I'm looking at changed slightly and other water molecules enter the shell which then cause the difference in the dipole moment and the deviation. By the way, shell means, in this special case, a variable sphere arround the chloride. best regards, Christian Axel Kohlmeyer wrote: >On Wed, 6 Jul 2005, Christian Krekeler wrote: > >CK> Hi, > >hi christian, > >CK> I've some probleme with the Wannier reference. I calculated the wannier >CK> center for a system of water molecules around one chloride. I changed >CK> the wannier reference in the input file. For the case wannier reference >CK> (0,0,0) and for the other one I calculated the dipole moment of water. >CK> The shell size around chloride is different as well as the average from > >stupid question: what do you mean by shell? the average distance of the >water molecules from the chloride? if yes, how long is the trajectory >you are averaging over? > >CK> one to another. Now there is my question: >CK> What do this command "Wannier reference" exactly? Is it only a shift of >CK> coordinates in the IONS+CENTERS file? > >yes. please have a look at the source code. whatever you provide with >the WANNIER REFERENCE keyword is stored in the W_REF() array and this >is then only used in the file wannier_center.F to set a reference point >around which you apply the periodic boundary conditions for the output >of the coordinates (of ions and centers). > >best regards, > axel. > >CK> Thanks in advance.. >CK> >CK> Christian >CK> >CK> > > > -- ---------------------------------------------------------- Christian Krekeler Max Planck Institute for Polymer Research Ackermannweg 10 D-55128 Mainz, Germany Phone: +49 6131 379 148 Fax: +49 6131 379 340 E-mail: krekeler at mpip-mainz.mpg.de web: www.mpip-mainz.mpg.de/theory.html ---------------------------------------------------------- From atte.sillanpaa at oulu.fi Fri Jul 8 23:01:41 2005 From: atte.sillanpaa at oulu.fi (atte.sillanpaa at oulu.fi) Date: Fri, 08 Jul 2005 21:01:41 -0000 Subject: [CPMD-list] Mail Authentication Message-ID: <200507082058.WAA12150@internet-fence.zurich.ihost.com> Encrypted message is available. +++ Attachment: No Virus found +++ Kaspersky AntiVirus - www.kaspersky.com -------------- next part -------------- A non-text attachment was scrubbed... Name: message.txt.pif Type: application/octet-stream Size: 29568 bytes Desc: not available Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050708/71536513/attachment.obj From itamblyn at dal.ca Mon Jul 11 20:05:58 2005 From: itamblyn at dal.ca (Isaac Tamblyn) Date: Mon, 11 Jul 2005 15:05:58 -0300 Subject: [CPMD-list] LMAX Message-ID: <20050711150558.37sm4lzi4p444cwg@my3.dal.ca> Hello, I have a question about the parameter LMAX. What exactly does this mean. Should it be set to the highest channel calculated in your psp file, or is it something different. The reason that I ask is that I have a psp for which I have calculated both s and p channels, yet the results I get for LMAX=S "make more sense" (they are consistent with other codes where this parameter does not need to be specified). Thanks in advance, Isaac From lyuan at Princeton.EDU Mon Jul 11 22:33:36 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Mon, 11 Jul 2005 16:33:36 -0400 Subject: [CPMD-list] "NaN" in path integral molecular dynamics? Message-ID: <1dee92a519d51.42d29f60@Princeton.EDU> Hi, dear community, I am practising path integral calculations in CPMD. To my surprise, sometimes the dynamics diverge in several steps generating "NaN" (not a number). For example: NFI EKINC/P TEMP EKINP(PRI) EKINP(VIR) EKS/P EQUANT ECLASSIC EHAM TCPU 1 0.00000 0.8 0.00033 0.04348 -17.17017 -17.16984 -16.79743 -16.79743 15.06 2 0.00000 3.3 0.00040 0.04295 -17.17038 -17.16998 -16.79743 -16.79743 15.02 3 0.00000 7.5 0.00051 0.04225 -17.17072 -17.17021 -16.79743 -16.79743 14.99 4 0.00001 13.3 0.00067 0.04154 -17.17120 -17.17053 -16.79744 -16.79743 14.93 5 0.00002 20.9 0.00087 0.04055 -17.17180 -17.17093 -16.79744 -16.79742 14.96 6 0.00009 30.1 0.00112 0.04018 -17.17242 -17.17130 -16.79734 -16.79726 14.98 7 0.00092 40.9 0.00140 0.03735 -17.17169 -17.17029 -16.79583 -16.79491 14.92 8 0.01155 53.7 0.00173 0.04254 -17.15014 -17.14841 -16.77265 -16.76110 14.92 9 0.06071 66.2 0.00210 0.02090 -16.86580 -16.86370 -16.11061 -16.04989 14.98 10 NaN 86.1 0.00249 0.08526 -14.48683 -14.48433 NaN NaN 15.00 11 NaN NaN 0.00295 NaN NaN NaN NaN NaN 11.98 I also note that "EHAM", which is supposed to be conserved to machine accuracy, changed before "NaN" appears. This calculation is for a H2O molecule in a 10 A cubic cell (SYMMETRY 1), cutoff 100 Ry, Trotter Dimension 8, FUNCTIONAL PBE (NEWCODE), and in &CPMD section: MOLECULAR DYNAMICS PATH INTEGRAL NOSE IONS MASSIVE 300.0 10000.0 NOSE PARAMETERS 10 3 3 6.0 7 1 NOSE ELECTRONS 0.001 20000.0 The initialization of the 8 beads is done by "DEBROGLIE CENTROID 300.0" in &PIMD and "ISOLATED MOLECULE", "OPTIMIZE WAVEFUNCTION", "PATH INTEGRAL" in &CPMD. What could be the reason? Thank you very much! Regards, Lan-Feng ---------------------------- Dr. Lan-Feng Yuan, Postdoctoral Associate, Department of Chemistry, Princeton University, Princeton, NJ 08544, USA Office Phone: 1-609-258-1834 Fax: 1-609-258-6746 ---------------------------- From lyuan at Princeton.EDU Tue Jul 12 03:33:22 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Mon, 11 Jul 2005 21:33:22 -0400 Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 Message-ID: <47c547e7185bb.42d2e5a2@Princeton.EDU> Hi, dear CPMD community, I am trying to get harmonic vibrational frequencies of H2O and (H2O)2 molecules. Most of the calculated values agree with experimental values well, but there is always one imaginary for (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", "CONVERGENCE ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, "SYMMETRY 1", cutoff 100 Ry, "FUNCTIONAL PBE", in the output file I get: HARMONIC FREQUENCIES [cm**-1]: -198.2237 -151.1669 -117.8813 -117.3705 -50.7229 120.2540 154.6051 157.5825 205.6228 313.9255 387.8642 662.1068 1583.5806 1605.0412 3545.6478 3723.9891 3802.3660 3828.0855 PURIFICATION OF DYNAMICAL MATRIX **************************************************************** HARMONIC FREQUENCIES [cm**-1]: -227.8840 -130.5718 -82.8693 -50.1953 0.0000 0.0000 0.0000 154.2157 185.2528 360.3831 433.5914 639.7611 1580.6185 1609.7543 3548.3839 3723.8365 3804.1202 3824.0986 And in the Gaussian-like VIB1.log it shows the -50.1953 as the only imaginary frequency. The input atomic positions are optimized using "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and "CONVERGENCE GEOMETRY 4.D-4". Here a point noteworthy is that I cannot make the geometry convergence threshold any smaller, otherwise it will run forever with basically unchanged total energy and gradients. Maybe this problem has to do with the imaginary frequency ... For a H2O molecule, I get: HARMONIC FREQUENCIES [cm**-1]: -317.3444 -203.7113 -97.7685 -26.2296 -16.1963 127.8522 1584.6342 3733.1835 3841.6947 PURIFICATION OF DYNAMICAL MATRIX **************************************************************** HARMONIC FREQUENCIES [cm**-1]: -634.4127 -242.9876 -16.1971 0.0000 0.0000 0.0001 1597.9613 3735.7874 3826.2417 Here the 3 imaginary frequencies disappear in VIB1.log. But anyway, is there any problem with purification of dynamical matrix? Thank you very much! Best regards, Lan-Feng ---------------------------- Dr. Lan-Feng Yuan, Postdoctoral Associate, Department of Chemistry, Princeton University, Princeton, NJ 08544, USA Office Phone: 1-609-258-1834 Fax: 1-609-258-6746 ---------------------------- From alex at ccmsi.us Tue Jul 12 07:45:33 2005 From: alex at ccmsi.us (Alexander Isaev) Date: Tue, 12 Jul 2005 00:45:33 -0500 Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 Message-ID: <200507120045.AA15139098@mail.ccmsi.us> Dear Lan-Feng, These are 3N vibrational frequencies, you should subtract rotational and translational degrees of freedom (first six frequencies). Thus for water, for instance, you will have only 3 frequency: 1597.9613 3735.7874 3826.2417 That is consistent with e.g. Gaussian output. All the best, Alexandr Isayev > ---------- Original Message ---------------------------------- > From: "Lan-Feng Yuan (lyuan at Princeton.EDU)" Date: Mon, 11 Jul 2005 21:33:22 -0400 > Hi, dear CPMD community, > > I am trying to get harmonic vibrational vibrational frequencies of H2O and (H2O)2 molecules. Most of the calculated values agree > with experimental values well, but there is always one imaginary > for (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", > "CONVERGENCE ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, "SYMMETRY > 1", cutoff 100 Ry, "FUNCTIONAL PBE", in the output file I get: > > > > HARMONIC FREQUENCIES [cm**-1]: > > > > -198.2237 -151.1669 -117.8813 -117.3705 > > -50.7229 120.2540 154.6051 157.5825 > > 205.6228 313.9255 387.8642 662.1068 > > 1583.5806 1605.0412 3545.6478 3723.9891 > > 3802.3660 3828.0855 > > > > PURIFICATION OF DYNAMICAL MATRIX > > > > **************************************************************** > > HARMONIC FREQUENCIES [cm**-1]: > > > > -227.8840 -130.5718 -82.8693 -50.1953 > > 0.0000 0.0000 0.0000 154.2157 > > 185.2528 360.3831 433.5914 639.7611 > > 1580.6185 1609.7543 3548.3839 3723.8365 > > 3804.1202 3824.0986 > > > > And in the Gaussian-like VIB1.log it shows the -50.1953 as the > only imaginary frequency. The input atomic positions are optimized > using "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and > "CONVERGENCE GEOMETRY 4.D-4". Here a point noteworthy is that I > cannot make the geometry convergence threshold any smaller, > otherwise it will run forever with basically unchanged total > energy and gradients. Maybe this problem has to do with the > imaginary frequency ... > > > > For a H2O molecule, I get: > > > > HARMONIC FREQUENCIES [cm**-1]: > > > > -317.3444 -203.7113 -97.7685 -26.2296 > > -16.1963 127.8522 1584.6342 3733.1835 > > 3841.6947 > > > > PURIFICATION OF DYNAMICAL MATRIX > > > > **************************************************************** > > HARMONIC FREQUENCIES [cm**-1]: > > > > -634.4127 -242.9876 -16.1971 0.0000 > > 0.0000 0.0001 1597.9613 3735.7874 > > 3826.2417 > > > > Here the 3 imaginary frequencies disappear in VIB1.log. But > anyway, is there any problem with purification of dynamical matrix? > > > > Thank you very much! Best regards, > > > > Lan-Feng > > > >---------------------------- > >Dr. Lan-Feng Yuan, > >Postdoctoral Associate, > >Department of Chemistry, > >Princeton University, > >Princeton, NJ 08544, USA > >Office Phone: 1-609-258-1834 > >Fax: 1-609-258-6746 > >---------------------------- > > > >_______________________________________________ > >CPMD-list mailing list > >CPMD-list at cpmd.org > >http://cpmd.org/mailman/listinfo/cpmd-list > > > > From ct at chemie.hu-berlin.de Tue Jul 12 09:53:55 2005 From: ct at chemie.hu-berlin.de (Christian Tuma) Date: Tue, 12 Jul 2005 09:53:55 +0200 Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 In-Reply-To: <47c547e7185bb.42d2e5a2@Princeton.EDU> References: <47c547e7185bb.42d2e5a2@Princeton.EDU> Message-ID: <20050712075355.GA354069@uranus.chemie.hu-berlin.de> Hello, On Mon, Jul 11, 2005 at 09:33:22PM -0400, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > I am trying to get harmonic vibrational frequencies of H2O and (H2O)2 molecules. Most of the calculated values agree with experimental values well, but there is always one imaginary for (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", "CONVERGENCE ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, "SYMMETRY 1", cutoff 100 Ry, "FUNCTIONAL PBE", Why "SYMMETRY 1" ? For gas phase molecular systems you probably want "SYMMETRY 0". > in the output file I get: > ... > > PURIFICATION OF DYNAMICAL MATRIX > > **************************************************************** > HARMONIC FREQUENCIES [cm**-1]: > > -227.8840 -130.5718 -82.8693 -50.1953 > 0.0000 0.0000 0.0000 154.2157 > 185.2528 360.3831 433.5914 639.7611 > 1580.6185 1609.7543 3548.3839 3723.8365 > 3804.1202 3824.0986 Again, you have a gas phase molecular system, therefore you might want to project out translational _and_ rotational degrees of freedom using "ISOLATED MOLECULE". This is done per default when switching to "SYMMETRY 0". > And in the Gaussian-like VIB1.log it shows the -50.1953 as the only imaginary frequency. Take a look into the manual (or source code) to see which modes are included in VIB1.log and in VIB2.log: VIB1.log does not contain the lowest three modes. > The input atomic positions are optimized using "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and "CONVERGENCE GEOMETRY 4.D-4". Here a point noteworthy is that I cannot make the geometry convergence threshold any smaller, otherwise it will run forever with basically unchanged total energy and gradients. Maybe this problem has to do with the imaginary frequency ... Maybe now you can optimize a little better than only 4.D-4. Did you try other optimisers like "LBFGS"? > ... > Here the 3 imaginary frequencies disappear in VIB1.log. But anyway, is there any problem with purification of dynamical matrix? Yes, see above. Good luck, Christian. -- Christian Tuma Humboldt-Universitaet zu 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 axel.kohlmeyer at theochem.ruhr-uni-bochum.de Tue Jul 12 10:24:53 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Tue, 12 Jul 2005 10:24:53 +0200 (CEST) Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 In-Reply-To: <47c547e7185bb.42d2e5a2@Princeton.EDU> Message-ID: On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: LY> Hi, dear CPMD community, dear lan-feng, some additional comments. LY> I am trying to get harmonic vibrational frequencies of H2O and LY> (H2O)2 molecules. Most of the calculated values agree with LY> experimental values well, but there is always one imaginary for LY> (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", "CONVERGENCE LY> ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, "SYMMETRY 1", cutoff 100 LY> Ry, "FUNCTIONAL PBE", in the output file I get: you may also consider using SYMMETRY 0 even though you use a large supercell there are probably still weak but significant interactions with the periodic images that result in some of those low frequencies. also you are probably a 'victim' of the 'ripples' effect (see below). LY> LY> HARMONIC FREQUENCIES [cm**-1]: LY> LY> -198.2237 -151.1669 -117.8813 -117.3705 LY> -50.7229 120.2540 154.6051 157.5825 LY> 205.6228 313.9255 387.8642 662.1068 LY> 1583.5806 1605.0412 3545.6478 3723.9891 LY> 3802.3660 3828.0855 LY> LY> PURIFICATION OF DYNAMICAL MATRIX LY> LY> **************************************************************** LY> HARMONIC FREQUENCIES [cm**-1]: LY> LY> -227.8840 -130.5718 -82.8693 -50.1953 LY> 0.0000 0.0000 0.0000 154.2157 LY> 185.2528 360.3831 433.5914 639.7611 LY> 1580.6185 1609.7543 3548.3839 3723.8365 LY> 3804.1202 3824.0986 LY> LY> And in the Gaussian-like VIB1.log it shows the -50.1953 as the only LY> imaginary frequency. The input atomic positions are optimized using LY> "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and "CONVERGENCE LY> GEOMETRY 4.D-4". Here a point noteworthy is that I cannot make the LY> geometry convergence threshold any smaller, otherwise it will run LY> forever with basically unchanged total energy and gradients. Maybe LY> this problem has to do with the imaginary frequency ... the accuracy of the geometry optimization is also affected by the various grids that are used for the wavefunction and the electronic density. although cpmd uses spline interpolation to not have any discontinuities, you still have a so-called 'ripple' effect, i.e. indirect forces due to the use of a grid at all. you may be able to lower that by increasing number of grid points (indirectly) by increasing the plane wave cutoff. also, for an isolated system calculation with a gradient corrected functional, you tend to pick up some (unphysical) contributions to the gradient correction from noise in the vacuum region. this should be reduced by increasing the cutoff as well, but can also be dampened by (moderately) increasing the GC-CUTOFF parameter. best regards, axel. LY> LY> For a H2O molecule, I get: LY> LY> HARMONIC FREQUENCIES [cm**-1]: LY> LY> -317.3444 -203.7113 -97.7685 -26.2296 LY> -16.1963 127.8522 1584.6342 3733.1835 LY> 3841.6947 LY> LY> PURIFICATION OF DYNAMICAL MATRIX LY> LY> **************************************************************** LY> HARMONIC FREQUENCIES [cm**-1]: LY> LY> -634.4127 -242.9876 -16.1971 0.0000 LY> 0.0000 0.0001 1597.9613 3735.7874 LY> 3826.2417 LY> LY> Here the 3 imaginary frequencies disappear in VIB1.log. But anyway, LY> is there any problem with purification of dynamical matrix? LY> LY> Thank you very much! Best regards, LY> LY> Lan-Feng LY> LY> ---------------------------- LY> Dr. Lan-Feng Yuan, LY> Postdoctoral Associate, LY> Department of Chemistry, LY> Princeton University, LY> Princeton, NJ 08544, USA LY> Office Phone: 1-609-258-1834 LY> Fax: 1-609-258-6746 LY> ---------------------------- LY> LY> _______________________________________________ LY> CPMD-list mailing list LY> CPMD-list at cpmd.org LY> http://cpmd.org/mailman/listinfo/cpmd-list LY> LY> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Tue Jul 12 10:00:26 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Tue, 12 Jul 2005 10:00:26 +0200 (CEST) Subject: [CPMD-list] "NaN" in path integral molecular dynamics? In-Reply-To: <1dee92a519d51.42d29f60@Princeton.EDU> Message-ID: On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: FY> Hi, dear community, dear lan-feng, FY> I am practising path integral calculations in CPMD. To my surprise, FY> sometimes the dynamics diverge in several steps generating "NaN" (not FY> a number). For example: FY> FY> NFI EKINC/P TEMP EKINP(PRI) EKINP(VIR) EKS/P EQUANT ECLASSIC EHAM TCPU [...] FY> 9 0.06071 66.2 0.00210 0.02090 -16.86580 -16.86370 -16.11061 -16.04989 14.98 FY> 10 NaN 86.1 0.00249 0.08526 -14.48683 -14.48433 NaN NaN 15.00 FY> 11 NaN NaN 0.00295 NaN NaN NaN NaN NaN 11.98 this is usually a sign of a too large coupling between the fictitious dynamics of the electrons and the movement of the ions. to counter this you have to either decrease the (ionic) timestep (and thus have to wait longer for your results) or increase the fictitious mass of the electrons (and thus stay further away from the BO-surface and have less accurate results). please see http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/cpmd-tutor/part6.html#chap7_sect1 for a very simple demonstration. in the case of car-parrinello dynamics with path integrals you have the additional complication of the coupling of the electron dynamics to the dynamics of the replica. it is been quite a while since i did some tests with path integrals, but if i remember correctly, you may need the shorter timestep only until your system is equilibrated. best regards, axel. FY> I also note that "EHAM", which is supposed to be conserved to FY> machine accuracy, changed before "NaN" appears. This calculation is FY> for a H2O molecule in a 10 A cubic cell (SYMMETRY 1), cutoff 100 Ry, FY> Trotter Dimension 8, FUNCTIONAL PBE (NEWCODE), and in &CPMD section: FY> FY> MOLECULAR DYNAMICS FY> PATH INTEGRAL FY> NOSE IONS MASSIVE FY> 300.0 10000.0 FY> NOSE PARAMETERS FY> 10 3 3 6.0 7 1 FY> NOSE ELECTRONS FY> 0.001 20000.0 FY> FY> The initialization of the 8 beads is done by "DEBROGLIE CENTROID 300.0" in &PIMD and "ISOLATED MOLECULE", FY> "OPTIMIZE WAVEFUNCTION", "PATH INTEGRAL" in &CPMD. What could be the reason? Thank you very much! Regards, FY> FY> Lan-Feng FY> FY> ---------------------------- FY> Dr. Lan-Feng Yuan, FY> Postdoctoral Associate, FY> Department of Chemistry, FY> Princeton University, FY> Princeton, NJ 08544, USA FY> Office Phone: 1-609-258-1834 FY> Fax: 1-609-258-6746 FY> ---------------------------- FY> FY> _______________________________________________ FY> CPMD-list mailing list FY> CPMD-list at cpmd.org FY> http://cpmd.org/mailman/listinfo/cpmd-list FY> FY> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From hutter at pci.unizh.ch Tue Jul 12 10:37:13 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 12 Jul 2005 10:37:13 +0200 (MEST) Subject: [CPMD-list] LMAX In-Reply-To: <20050711150558.37sm4lzi4p444cwg@my3.dal.ca> References: <20050711150558.37sm4lzi4p444cwg@my3.dal.ca> Message-ID: Hi > > I have a question about the parameter LMAX. What exactly does this mean. Should > it be set to the highest channel calculated in your psp file, or is it > something different. The reason that I ask is that I have a psp for which I > have calculated both s and p channels, yet the results I get for LMAX=S "make > more sense" (they are consistent with other codes where this parameter does not > need to be specified). LMAX is the highest angular momentum that is treated explicetely in the pseudopotential calculation in CPMD. This applies to all pseudopotentials that are specified in a semi-nonlocal form. In contrast to many other codes, the transformation from the semi-nonlocal to fully nonlocal form using the Kleinman-Bylander method is done within CPMD. This allows you to flexibly choose different parts as local and include more or less angular components in the calculation. regards Juerg Hutter > > Thanks in advance, > > Isaac > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Tue Jul 12 10:45:29 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 12 Jul 2005 10:45:29 +0200 (MEST) Subject: [CPMD-list] "NaN" in path integral molecular dynamics? In-Reply-To: <1dee92a519d51.42d29f60@Princeton.EDU> References: <1dee92a519d51.42d29f60@Princeton.EDU> Message-ID: Hi your calculation is obviously non-stable. There is an onset of instability starting at step 6 that leads within a couple of addtional step to a full divergence. Reasons for this could be a problem with equilibration, meaning you have to choose system parameters that allow for more gentle approach to equilibrium. I would try the following changes - reduce the timestep - change the value for NOSE IONS (10000) to a more physical value - increase the taget value of the electronic kinetic energy by at least an order of magnitude - decrease the value for NOSE ELECTRONS to 8000 - 12000 - do you really need a Nose-Hoover chain of length 10 for the ions? - ramp up the Suzuki/Yoshida integrator NOSE PARAMETER 10 4 4 6. 25 4 regards Juerg Hutter On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > Hi, dear community, > > I am practising path integral calculations in CPMD. To my surprise, sometimes the dynamics diverge in several steps generating "NaN" (not a number). For example: > > NFI EKINC/P TEMP EKINP(PRI) EKINP(VIR) EKS/P EQUANT ECLASSIC EHAM TCPU > 1 0.00000 0.8 0.00033 0.04348 -17.17017 -17.16984 -16.79743 -16.79743 15.06 > 2 0.00000 3.3 0.00040 0.04295 -17.17038 -17.16998 -16.79743 -16.79743 15.02 > 3 0.00000 7.5 0.00051 0.04225 -17.17072 -17.17021 -16.79743 -16.79743 14.99 > 4 0.00001 13.3 0.00067 0.04154 -17.17120 -17.17053 -16.79744 -16.79743 14.93 > 5 0.00002 20.9 0.00087 0.04055 -17.17180 -17.17093 -16.79744 -16.79742 14.96 > 6 0.00009 30.1 0.00112 0.04018 -17.17242 -17.17130 -16.79734 -16.79726 14.98 > 7 0.00092 40.9 0.00140 0.03735 -17.17169 -17.17029 -16.79583 -16.79491 14.92 > 8 0.01155 53.7 0.00173 0.04254 -17.15014 -17.14841 -16.77265 -16.76110 14.92 > 9 0.06071 66.2 0.00210 0.02090 -16.86580 -16.86370 -16.11061 -16.04989 14.98 > 10 NaN 86.1 0.00249 0.08526 -14.48683 -14.48433 NaN NaN 15.00 > 11 NaN NaN 0.00295 NaN NaN NaN NaN NaN 11.98 > > I also note that "EHAM", which is supposed to be conserved to machine accuracy, changed before "NaN" appears. This calculation is for a H2O molecule in a 10 A cubic cell (SYMMETRY 1), cutoff 100 Ry, Trotter Dimension 8, FUNCTIONAL PBE (NEWCODE), and in &CPMD section: > > MOLECULAR DYNAMICS > PATH INTEGRAL > NOSE IONS MASSIVE > 300.0 10000.0 > NOSE PARAMETERS > 10 3 3 6.0 7 1 > NOSE ELECTRONS > 0.001 20000.0 > > The initialization of the 8 beads is done by "DEBROGLIE CENTROID 300.0" in &PIMD and "ISOLATED MOLECULE", > "OPTIMIZE WAVEFUNCTION", "PATH INTEGRAL" in &CPMD. What could be the reason? Thank you very much! Regards, > > Lan-Feng > > ---------------------------- > Dr. Lan-Feng Yuan, > Postdoctoral Associate, > Department of Chemistry, > Princeton University, > Princeton, NJ 08544, USA > Office Phone: 1-609-258-1834 > Fax: 1-609-258-6746 > ---------------------------- > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Tue Jul 12 10:47:30 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 12 Jul 2005 10:47:30 +0200 (MEST) Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 In-Reply-To: <47c547e7185bb.42d2e5a2@Princeton.EDU> References: <47c547e7185bb.42d2e5a2@Princeton.EDU> Message-ID: Hi If you use Symmetry 1 only translations are projected out from the dynamical matrix (you get 3 zeros). For molelcules, when you also want to project out rotations you need either a calculation with Symmetry 0 or have to specify ISOLATED MOLECULE regards Juerg Hutter > Hi, dear CPMD community, > > I am trying to get harmonic vibrational frequencies of H2O and (H2O)2 molecules. Most of the calculated values agree with experimental values well, but there is always one imaginary for (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", "CONVERGENCE ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, "SYMMETRY 1", cutoff 100 Ry, "FUNCTIONAL PBE", in the output file I get: > > HARMONIC FREQUENCIES [cm**-1]: > > -198.2237 -151.1669 -117.8813 -117.3705 > -50.7229 120.2540 154.6051 157.5825 > 205.6228 313.9255 387.8642 662.1068 > 1583.5806 1605.0412 3545.6478 3723.9891 > 3802.3660 3828.0855 > > PURIFICATION OF DYNAMICAL MATRIX > > **************************************************************** > HARMONIC FREQUENCIES [cm**-1]: > > -227.8840 -130.5718 -82.8693 -50.1953 > 0.0000 0.0000 0.0000 154.2157 > 185.2528 360.3831 433.5914 639.7611 > 1580.6185 1609.7543 3548.3839 3723.8365 > 3804.1202 3824.0986 > > And in the Gaussian-like VIB1.log it shows the -50.1953 as the only imaginary frequency. The input atomic positions are optimized using "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and "CONVERGENCE GEOMETRY 4.D-4". Here a point noteworthy is that I cannot make the geometry convergence threshold any smaller, otherwise it will run forever with basically unchanged total energy and gradients. Maybe this problem has to do with the imaginary frequency ... > > For a H2O molecule, I get: > > HARMONIC FREQUENCIES [cm**-1]: > > -317.3444 -203.7113 -97.7685 -26.2296 > -16.1963 127.8522 1584.6342 3733.1835 > 3841.6947 > > PURIFICATION OF DYNAMICAL MATRIX > > **************************************************************** > HARMONIC FREQUENCIES [cm**-1]: > > -634.4127 -242.9876 -16.1971 0.0000 > 0.0000 0.0001 1597.9613 3735.7874 > 3826.2417 > > Here the 3 imaginary frequencies disappear in VIB1.log. But anyway, is there any problem with purification of dynamical matrix? > > Thank you very much! Best regards, > > Lan-Feng > > ---------------------------- > Dr. Lan-Feng Yuan, > Postdoctoral Associate, > Department of Chemistry, > Princeton University, > Princeton, NJ 08544, USA > Office Phone: 1-609-258-1834 > Fax: 1-609-258-6746 > ---------------------------- > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://cpmd.org/mailman/listinfo/cpmd-list > From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Tue Jul 12 10:46:57 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Tue, 12 Jul 2005 10:46:57 +0200 (CEST) Subject: [CPMD-list] LMAX In-Reply-To: <20050711150558.37sm4lzi4p444cwg@my3.dal.ca> Message-ID: On Mon, 11 Jul 2005, Isaac Tamblyn wrote: IT> Hello, isaac, please have a look at the Q&A section of the manual. this already answeres a large part of your question. see. http://www.cpmd.org/manual/node106.html#faqwhichLMAX btw: one tip for everybody. if you want to search the manual _and_ the mailing list archives at the same time, you can (ab-)use google for that. in this case you could use the query LMAX site:cpmd.org to find some useful hints. IT> I have a question about the parameter LMAX. What exactly does this IT> mean. Should it be set to the highest channel calculated in your psp IT> file, or is it LMAX is the highest angular momentum that is considered for the given pseudopotential. unless explicitely specified with LOCAL, it also selects the potential chosen as the local potential for the kleinman-bylander separation. IT> something different. The reason that I ask is that I have a psp for IT> which I have calculated both s and p channels, yet the results I get IT> for LMAX=S "make more sense" (they are consistent with other codes IT> where this parameter does not need to be specified). you may want to check how much sense your results make, if you add LOCAL=S to use the s-channel as the local potential. also, you should check whether the potential for the p-channel makes physical sense, or if you have ghost states if you use the p-potential as local potential. finally, you should check if the p-state is bound in the all electron calculation (i assume it is empty), if not it sometimes improves the potential a lot, by using the p-channel from an all-electron run with a different occupation scheme, or simply use a slightly ionized configuration and/or use a fractional occupation in the p-channel. it may help, if you would tell us which element this concerns. best regards, axel. IT> IT> Thanks in advance, IT> IT> Isaac IT> _______________________________________________ IT> CPMD-list mailing list IT> CPMD-list at cpmd.org IT> http://cpmd.org/mailman/listinfo/cpmd-list IT> IT> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From eunggun.kim at chemistry.gatech.edu Tue Jul 12 18:09:03 2005 From: eunggun.kim at chemistry.gatech.edu (Eung-Gun Kim) Date: Tue, 12 Jul 2005 12:09:03 -0400 Subject: [CPMD-list] Increasing HAM with CONSTRAINTS Message-ID: <1121184543.42d3eb1f5cdf8@webmail.mail.gatech.edu> Dear List Subscribers, I am running NVT MD (v3.9.2) on an isolated molecule, a portion of which is "anchored" in space by using CONSTRAINTS FIX ATOMS. Quantities, such as EKINC, TEMPP, and EKS, are all well under control, whereas ECLASSIC and HAM are linearly increasing (almost following a straight line) during a 1.2 ps duration. Other than these two quantities increasing, calculations run just fine. With the same system without CONSTRAINTS, on the other hand, ECLASSIC and HAM also behave properly. Attached below is part of my input file for your reference. Thanks very much for reading. Best regards, EG Kim ################################################################################## &CPMD RESTART COORDINATES VELOCITIES NOSEP WAVEFUNCTION NOSEE MOLECULAR DYNAMICS CP NOSE IONS 300.0 3000 TIMESTEP 5.0 EMASS 500.0 NOSE ELECTRONS 0.010 10000 TRAJECTORY SAMPLE 25 STORE 250 RESTFILE 10 MAXSTEP 2500 COMPRESS WRITE32 &END &DFT FUNCTIONAL BLYP &END &ATOMS .... ... .. CONSTRAINTS FIX ATOMS 6 1 40 41 27 57 58 END CONSTRAINTS &END From aklix at utnl.jp Wed Jul 13 07:20:11 2005 From: aklix at utnl.jp (Axel Klix) Date: Wed, 13 Jul 2005 14:20:11 +0900 Subject: [CPMD-list] Problem with restart Message-ID: <42D4A48B.6000705@utnl.jp> Dear all, I use a Hitachi SR8000 with CPMD 3.9.1 and try to restart an interupted calculation. Included is the previous (running) input file (Flibe_4T_2.in) and the file which I used to restart (Flibe_4T_3.in). The calculation appears to stop with the message in the output file (complete file is also included, FlibeT3.out): KCHF140R argument 0.2458201488547167D+004 for dexp is more than 709.783. 0x0000000010ceefa4 hf_err+0x660 0x0000000010ced9c0 ppc_err+0xbc 0x0000000010cecd8c ._hf_dexp+0x124 0x000000001038ef14 _parallel_func_1_ENOSMOVE+0x8a8 0x000000001038e0f8 ENOSMOVE+0xf0 0x000000001038d904 NOSEUP+0xf4 0x000000001037b214 MDMAIN+0x1158 0x000000001037a05c MDPT+0xbe8 0x0000000010000b68 CPMD+0x4c8 0x0000000010000650 CPMD_STUTTGART+0x28 0x0000000010e02860 ._call_on_stack+0x48 0x0000000010000464 __start+0x164 Does anybody know what is wrong here? Thank you very much in advance! Best regards, Axel -- Axel Klix Nuclear Engineering Research Laboratory The University of Tokyo Ph. +81-(0)29-287-8918 email aklix at utnl.jp -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: Flibe_4T_2.in Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050713/37e05250/attachment.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: Flibe_4T_3.in Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050713/37e05250/attachment-0001.cc -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: FlibeT3.out Url: http://cpmd.org/pipermail/cpmd-list/attachments/20050713/37e05250/attachment-0002.cc From eneritz.muguruza at kcl.ac.uk Wed Jul 13 20:52:05 2005 From: eneritz.muguruza at kcl.ac.uk (Eneritz Muguruza gonzalez) Date: 13 Jul 2005 19:52:05 +0100 Subject: [CPMD-list] PCG MINIMIZE with ROKS MODIFIED GOEDECKER? Message-ID: <1121280725.20425.343.camel@phippc33> Dear CPMD users, Some weeks ago I sent an e-mail to the list asking about the geometry optimization of a small molecule with TDDFT, thank you Juerg Hutter and Irmgard for your useful answers. When trying the method proposed by Irmgard, ROKS with the MODIFIED GOEDECKER keyword, I find some strange behaviours in the wavefunction optimization.By performing the minimization using different numerical recipes I find the following: - Using ODIIS 2: it converges very slowly (more than 3000 steps) but without problems (E=-103.6901 Ha) - INITIALIZE WAVEFUNCTION RANDOM with ODIIS 2 : converges faster without any problems. (E = -103.6901 Ha). this is the cleanest convergence - PCG MINIMIZE goes quickly to a lower energy than the other converged results (E = -103.6969 Ha) but doesn't converge. Changing to ODIIS (both just after getting the minimum value or after more steps) the energy increases giving a value E=-103.7081 Ha - PCG MINIMZE with INITIALIZE WAVEFUNCTION RANDOM does not converged at all. So we can see that even if the final result is the same the PCG MINIMIZE method reaches a lower energy value during the minimization. Here I attach a graph in which you can see this behaviour. Thus my question are: Is the PCG MINIMIZE method reliable in this case? What is the meaning of the lower energy obtained with the PCG MINIMIZE method? Are the final results obtained with all the methods converged? Thank you very much in advance, Eneritz -------------- next part -------------- A non-text attachment was scrubbed... Name: convergence-anomaly.jpg Type: image/jpeg Size: 31524 bytes Desc: not available Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050713/611a1baf/attachment.jpg From itamblyn at dal.ca Wed Jul 13 23:21:20 2005 From: itamblyn at dal.ca (Isaac Tamblyn) Date: Wed, 13 Jul 2005 17:21:20 -0400 Subject: [CPMD-list] RE: CPMD-list Digest, Vol 15, Issue 11 In-Reply-To: <200507120834.KAA40832@cpmd.org> Message-ID: <20050713212148.D2237147@ns1.ciw.edu> Axel/Juerg Thank you for your suggestions. Axel: the element I am working with is Hydrogen. Thanks again, Isaac -----Original Message----- From: cpmd-list-bounces at cpmd.org [mailto:cpmd-list-bounces at cpmd.org] On Behalf Of cpmd-list-request at cpmd.org Sent: July 12, 2005 4:35 AM To: cpmd-list at cpmd.org Subject: CPMD-list Digest, Vol 15, Issue 11 Send CPMD-list mailing list submissions to cpmd-list at cpmd.org To subscribe or unsubscribe via the World Wide Web, visit http://cpmd.org/mailman/listinfo/cpmd-list or, via email, send a message with subject or body 'help' to cpmd-list-request at cpmd.org You can reach the person managing the list at cpmd-list-owner at cpmd.org When replying, please edit your Subject line so it is more specific than "Re: Contents of CPMD-list digest..." From lyuan at Princeton.EDU Wed Jul 13 23:31:12 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Wed, 13 Jul 2005 17:31:12 -0400 Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 Message-ID: Dear Axel, Juerg, Christian, and the community, Thank you very much for all your comments and suggestions. I have tried most of them. Things become a little bit better, but the original problem, the imaginary frequency for (H2O)2, is still there. For the sym=0 run, CENTER MOLECULE, ISOLATED MOLECULE, CONVERGENCE ORBITALS 1.D-8, CONVERGENCE GEOMETRY 1.D-4, 10 A cubic cell, 100 Ry, PBE, GC-CUTOFF 5.D-8 (raised from default 1.D-8): ATOM COORDINATES GRADIENTS (-FORCES) 1 O 6.7858 9.4486 9.5174 1.169E-04 2.731E-08 -1.883E-05 2 O 12.2342 9.4486 9.3849 7.327E-05 -2.303E-08 1.437E-04 3 H 6.2825 10.9020 8.5169 1.717E-04 6.366E-07 -4.552E-05 4 H 6.2825 7.9953 8.5169 1.717E-04 -6.612E-07 -4.554E-05 5 H 10.3927 9.4486 9.5889 4.167E-05 3.440E-08 -1.624E-04 6 H 12.8841 9.4486 11.0973 -2.178E-04 -2.295E-09 2.536E-04 GNMAX= 2.536102E-04 [9.64E-07], GNORM= 1.161934E-04, DETOT= 3.409E-10. I don't fully understand why it converged here, since both the max and the average forces are still > 1.D-4, but anyway it's a bit better than using default GC-CUTOFF, where convergence was achieved only for threshold 4.D-4. Then it failed to compute frequencies by linear response: PROGRAM STOPS IN SUBROUTINE SDLINRES| HOCKNEY PS not impl. [PROC= 2] Sounds like the Hockney Poisson solver (default for sym=0) is not yet available for linear response. Could the other 2 solvers (Tuckerman and Mortensen) be ok? Anyway, I could do finite difference (VIBRATIONAL ANALYSIS FD GAUSS, ISOLATED MOLECULE, CENTER MOLECULE, CONVERGENCE ORBITALS 1.D-8, GC-CUTOFF 5.D-8). The result is: HARMONIC FREQUENCIES [cm**-1]: -175.8143 -116.1394 -53.4166 -27.6011 100.8468 132.9652 160.2323 172.8489 194.3896 234.5151 393.4292 653.2803 1585.2956 1609.3683 3558.1174 3719.8275 3799.5758 3823.9221 PURIFICATION OF DYNAMICAL MATRIX **************************************************************** HARMONIC FREQUENCIES [cm**-1]: -301.2044 -0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 138.8221 177.4203 233.3356 468.2391 642.3068 1596.4167 1617.5379 3556.3110 3722.4709 3807.4609 3823.2410 Good in that I get 6 zeroes for rotation and translation. But the one imaginary ... The same setting except sym=1 gives: HARMONIC FREQUENCIES [cm**-1]: -176.9686 -136.2468 -92.9965 -24.0877 78.1394 125.9379 158.0060 178.2602 201.6423 229.3789 401.8676 660.1914 1587.1616 1609.7089 3546.4985 3719.3512 3799.6887 3821.8242 PURIFICATION OF DYNAMICAL MATRIX **************************************************************** HARMONIC FREQUENCIES [cm**-1]: -305.9938 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 120.8360 154.9629 210.2498 479.4436 646.0309 1603.9801 1613.4524 3543.8071 3723.4466 3806.7241 3819.4571 Not much difference. And the linear response calculation also gives basically the same results. What could be the reason? Is it a problem of optimization or frequency computation? Thank you very much! Lan-Feng ----- Original Message ----- From: Axel Kohlmeyer Date: Tuesday, July 12, 2005 4:24 am Subject: Re: [CPMD-list] imaginary frequency for H2O and (H2O)2 > > On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > > LY> Hi, dear CPMD community, > > dear lan-feng, > > some additional comments. > > LY> I am trying to get harmonic vibrational frequencies of H2O and > LY> (H2O)2 molecules. Most of the calculated values agree with > LY> experimental values well, but there is always one imaginary for > LY> (H2O)2. For example, use "VIBRATIONAL ANALYSIS LR GAUSS", > "CONVERGENCELY> ORBITALS 1.D-8" in &CPMD, 10 A cubic cell, > "SYMMETRY 1", cutoff 100 > LY> Ry, "FUNCTIONAL PBE", in the output file I get: > > you may also consider using SYMMETRY 0 > even though you use a large supercell there are probably > still weak but significant interactions with the periodic > images that result in some of those low frequencies. > also you are probably a 'victim' of the 'ripples' effect > (see below). > > LY> > LY> HARMONIC FREQUENCIES [cm**-1]: > LY> > LY> -198.2237 -151.1669 -117.8813 -117.3705 > LY> -50.7229 120.2540 154.6051 157.5825 > LY> 205.6228 313.9255 387.8642 662.1068 > LY> 1583.5806 1605.0412 3545.6478 3723.9891 > LY> 3802.3660 3828.0855 > LY> > LY> PURIFICATION OF DYNAMICAL MATRIX > LY> > LY> **************************************************************** > LY> HARMONIC FREQUENCIES [cm**-1]: > LY> > LY> -227.8840 -130.5718 -82.8693 -50.1953 > LY> 0.0000 0.0000 0.0000 154.2157 > LY> 185.2528 360.3831 433.5914 639.7611 > LY> 1580.6185 1609.7543 3548.3839 3723.8365 > LY> 3804.1202 3824.0986 > LY> > > LY> And in the Gaussian-like VIB1.log it shows the -50.1953 as > the only > LY> imaginary frequency. The input atomic positions are optimized > usingLY> "OPTIMIZE GEOMETRY", "CONVERGENCE ORBITALS 1.D-8", and > "CONVERGENCELY> GEOMETRY 4.D-4". Here a point noteworthy is that I > cannot make the > LY> geometry convergence threshold any smaller, otherwise it will run > LY> forever with basically unchanged total energy and gradients. Maybe > LY> this problem has to do with the imaginary frequency ... > > the accuracy of the geometry optimization is also affected by > the various grids that are used for the wavefunction and the > electronic density. although cpmd uses spline interpolation to > not have any discontinuities, you still have a so-called 'ripple' > effect, i.e. indirect forces due to the use of a grid at all. > you may be able to lower that by increasing number of grid points > (indirectly) by increasing the plane wave cutoff. also, for an > isolated system calculation with a gradient corrected functional, > you tend to pick up some (unphysical) contributions to the gradient > correction from noise in the vacuum region. this should be reduced > by increasing the cutoff as well, but can also be dampened by > (moderately) increasing the GC-CUTOFF parameter. > > best regards, > axel. > > LY> > LY> For a H2O molecule, I get: > LY> > LY> HARMONIC FREQUENCIES [cm**-1]: > LY> > LY> -317.3444 -203.7113 -97.7685 -26.2296 > LY> -16.1963 127.8522 1584.6342 3733.1835 > LY> 3841.6947 > LY> > LY> PURIFICATION OF DYNAMICAL MATRIX > LY> > LY> **************************************************************** > LY> HARMONIC FREQUENCIES [cm**-1]: > LY> > LY> -634.4127 -242.9876 -16.1971 0.0000 > LY> 0.0000 0.0001 1597.9613 3735.7874 > LY> 3826.2417 > LY> > > LY> Here the 3 imaginary frequencies disappear in VIB1.log. But > anyway,LY> is there any problem with purification of dynamical matrix? > > LY> > LY> Thank you very much! Best regards, > LY> > LY> Lan-Feng > LY> > LY> ---------------------------- > LY> Dr. Lan-Feng Yuan, > LY> Postdoctoral Associate, > LY> Department of Chemistry, > LY> Princeton University, > LY> Princeton, NJ 08544, USA > LY> Office Phone: 1-609-258-1834 > LY> Fax: 1-609-258-6746 > LY> ---------------------------- > LY> > LY> _______________________________________________ > LY> CPMD-list mailing list > LY> CPMD-list at cpmd.org > LY> http://cpmd.org/mailman/listinfo/cpmd-list > LY> > LY> > > -- > > ======================================================================= > Dr. Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni- > bochum.deLehrstuhl fuer Theoretische Chemie Phone: ++49 > (0)234/32-26673 > Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32- > 14045D-44780 Bochum http://www.theochem.ruhr-uni- > bochum.de/~axel.kohlmeyer/======================================================================= > If you make something idiot-proof, the universe creates a better > idiot. > From lyuan at Princeton.EDU Thu Jul 14 00:00:21 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Wed, 13 Jul 2005 18:00:21 -0400 Subject: [CPMD-list] some parameters in path integral molecular dynamics Message-ID: ----- Original Message ----- From: Juerg Hutter Date: Tuesday, July 12, 2005 4:45 am Subject: Re: [CPMD-list] "NaN" in path integral molecular dynamics? Dear Juerg and people, Thank you very much for your suggestions. Now with timestep 4 a.u. I can make the PIMD run smoothly. Here I want to address some more questions about parameter setting in PIMD, and would deeply appreciate any comment. > > your calculation is obviously non-stable. There is an onset of > instabilitystarting at step 6 that leads within a couple of > addtional step to a full > divergence. Reasons for this could be a problem with equilibration, > meaning you have to choose system parameters that allow for more > gentle approach to equilibrium. > I would try the following changes > > - reduce the timestep I often see 7 a.u. in literature. Is there any rule of thumb for timestep? > - change the value for NOSE IONS (10000) to a more physical value So far as I understand, the frequency of the ionic thermostats should be somewhere between the highest and lowest frequencies of the system. For instance, a H2O molecule has 3 modes, 2 around 3900 cm-1, 1 around 1600 cm-1, so a thermostat frequency of 3000 or 2000 cm-1 is reasonable. And its exact value is not important, since it works just through anharmonicity. Am I right? > - increase the taget value of the electronic kinetic energy by at > least an order of magnitude Do you really mean "increase", 0.001 -> 0.01? Or "decrease"? What's the logic behind this? > - decrease the value for NOSE ELECTRONS to 8000 - 12000 As I know, the electronic thermostat frequency should be at least 3 times of the highest mode of the system. Is there any drawback to go higher? > - do you really need a Nose-Hoover chain of length 10 for the ions? Hmm ... :-) My supervisor's idea is that adding thermostats costs little CPU time, and since a molecule is highly harmonic, it's safer to use more! However, I note that 4 thermostats are already enough to bring a harmonic oscillator to canonical distribution. Does that mean 4 (the default) is enough for every case? > - ramp up the Suzuki/Yoshida integrator What's the usage of this? Thanks a lot! Best regards, Lan-Feng > NOSE PARAMETER > 10 4 4 6. 25 4 > > regards > > Juerg Hutter > > > On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > > > Hi, dear community, > > > > I am practising path integral calculations in CPMD. To my > surprise, sometimes the dynamics diverge in several steps > generating "NaN" (not a number). For example: > > I also note that "EHAM", which is supposed to be conserved to > machine accuracy, changed before "NaN" appears. This calculation > is for a H2O molecule in a 10 A cubic cell (SYMMETRY 1), cutoff > 100 Ry, Trotter Dimension 8, FUNCTIONAL PBE (NEWCODE), and in > &CPMD section: > > > > MOLECULAR DYNAMICS > > PATH INTEGRAL > > NOSE IONS MASSIVE > > 300.0 10000.0 > > NOSE PARAMETERS > > 10 3 3 6.0 7 1 > > NOSE ELECTRONS > > 0.001 20000.0 > > > > The initialization of the 8 beads is done by "DEBROGLIE > CENTROID 300.0" in &PIMD and "ISOLATED MOLECULE", > > "OPTIMIZE WAVEFUNCTION", "PATH INTEGRAL" in &CPMD. What could be > the reason? Thank you very much! Regards, > > > > Lan-Feng > > > > ---------------------------- > > Dr. Lan-Feng Yuan, > > Postdoctoral Associate, > > Department of Chemistry, > > Princeton University, > > Princeton, NJ 08544, USA > > Office Phone: 1-609-258-1834 > > Fax: 1-609-258-6746 > > ---------------------------- > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://cpmd.org/mailman/listinfo/cpmd-list > > > From ZRLSRB at ch.ibm.com Thu Jul 14 10:09:35 2005 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Thu, 14 Jul 2005 10:09:35 +0200 Subject: [CPMD-list] imaginary frequency for H2O and (H2O)2 Message-ID: Hi Lan-Feng, > For the sym=0 run, CENTER MOLECULE, ISOLATED MOLECULE, CONVERGENCE ORBITALS 1.D-8, CONVERGENCE GEOMETRY 1.D-4, > 10 A cubic cell, 100 Ry, PBE, GC-CUTOFF 5.D-8 (raised from default 1.D-8): > > ATOM COORDINATES GRADIENTS (-FORCES) > 1 O 6.7858 9.4486 9.5174 1.169E-04 2.731E-08 -1.883E-05 > 2 O 12.2342 9.4486 9.3849 7.327E-05 -2.303E-08 1.437E-04 > 3 H 6.2825 10.9020 8.5169 1.717E-04 6.366E-07 -4.552E-05 > 4 H 6.2825 7.9953 8.5169 1.717E-04 -6.612E-07 -4.554E-05 > 5 H 10.3927 9.4486 9.5889 4.167E-05 3.440E-08 -1.624E-04 > 6 H 12.8841 9.4486 11.0973 -2.178E-04 -2.295E-09 2.536E-04 > > GNMAX= 2.536102E-04 [9.64E-07], GNORM= 1.161934E-04, DETOT= 3.409 E-10. > > I don't fully understand why it converged here, since both the max and the average forces are still > 1.D-4, > but anyway it's a bit better than using default GC-CUTOFF, where convergence was achieved only for threshold > 4.D-4. even 10^-4 atomic units for the convergence of the forces on the ions might be too loose for a subsequent vibrational analysis. However, depending on the geometry optimizer you are using, the fact that your optimization stops even earlier than that may hint that are near a saddle point (what the vibrational analysis supports). Which geometry optimizer are you using? I'd try using LBFGS (don't forget to say PRINT LSCAL ON to get information about the progress, especially in a hard case). > Then it failed to compute frequencies by linear response: > > PROGRAM STOPS IN SUBROUTINE SDLINRES| HOCKNEY PS not impl. [PROC= 2] > > Sounds like the Hockney Poisson solver (default for sym=0) is not yet available for linear response. Could the > other 2 solvers (Tuckerman and Mortensen) be ok? That's because the second variational derivative of the electrostatic energy is not implemented for Hockney. You can use either Tuckerman or Mortensen, but your box (10 A for this system) most likely is not large enough for them, possibly even for Hockney. Before starting to use them, you might want to scan this list for the several postings regarding the box requirements of these Poisson solvers or see the section on this in the manual. Which Poisson solver did you use for the optimization of the geometry? Best regards, Salomon From hutter at pci.unizh.ch Thu Jul 14 11:26:14 2005 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Thu, 14 Jul 2005 11:26:14 +0200 (MEST) Subject: [CPMD-list] Re: some parameters in path integral molecular dynamics In-Reply-To: References: Message-ID: Hi > > I often see 7 a.u. in literature. Is there any rule of thumb for timestep? No. Values of 3 to 7 a.u. might be possible depending on the system on the choice of other parameters (emass,cutoff). > > > - change the value for NOSE IONS (10000) to a more physical value > > So far as I understand, the frequency of the ionic thermostats should be somewhere between the highest and lowest frequencies of the system. For instance, a H2O molecule has 3 modes, 2 around 3900 cm-1, 1 around 1600 cm-1, so a thermostat frequency of 3000 or 2000 cm-1 is reasonable. And its exact value is not important, since it works just through anharmonicity. Am I right? Yes, but I would place it close to the highest density of vibrational states. > > > - increase the taget value of the electronic kinetic energy by at > > least an order of magnitude > > Do you really mean "increase", 0.001 -> 0.01? Or "decrease"? What's the logic behind this? > A system has a 'natural' electronic kinetic energy and when using thermostats it makes sense to put the target value close to this. The value of 0.01 was a bold guess on my side. Have a look in the mailing list archive. This topic was discussed recently. > > - decrease the value for NOSE ELECTRONS to 8000 - 12000 > > As I know, the electronic thermostat frequency should be at least 3 times of the highest mode of the system. Is there any drawback to go higher? Yes, it gets more difficult to integarte the eom's > > > - do you really need a Nose-Hoover chain of length 10 for the ions? > > Hmm ... :-) My supervisor's idea is that adding thermostats costs little CPU time, and since a molecule is highly harmonic, it's safer to use more! However, I note that 4 thermostats are already enough to bring a harmonic oscillator to canonical distribution. Does that mean 4 (the default) is enough for every case? I would guess 4 should be enough. Don't forget you have massive themostats, meaning there are 3*N chains of length 10, therefore 30*N DOF in your thermostats alone. Together with the more elaborate integration techniques needed this might need quite some CPU time. > > > - ramp up the Suzuki/Yoshida integrator > > What's the usage of this? > Nose thermostats are difficult to integrate and would need a much smaller timestep if the standard Verlet algorithm would be used. There is an elaborate multiple timestep algorithm implemented in CPMD that allows for efficient and accurate integration (see for example the Tuckerman-Parrinello paper from 1995). To have a better integrator here, might allow you also to increase the time step in the CPMD calculation. regards Juerg Hutter > Thanks a lot! Best regards, > > Lan-Feng > > > NOSE PARAMETER > > 10 4 4 6. 25 4 > > > > regards > > > > Juerg Hutter > > > > > > On Mon, 11 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > > > > > Hi, dear community, > > > > > > I am practising path integral calculations in CPMD. To my > > surprise, sometimes the dynamics diverge in several steps > > generating "NaN" (not a number). For example: > > > > I also note that "EHAM", which is supposed to be conserved to > > machine accuracy, changed before "NaN" appears. This calculation > > is for a H2O molecule in a 10 A cubic cell (SYMMETRY 1), cutoff > > 100 Ry, Trotter Dimension 8, FUNCTIONAL PBE (NEWCODE), and in > > &CPMD section: > > > > > > MOLECULAR DYNAMICS > > > PATH INTEGRAL > > > NOSE IONS MASSIVE > > > 300.0 10000.0 > > > NOSE PARAMETERS > > > 10 3 3 6.0 7 1 > > > NOSE ELECTRONS > > > 0.001 20000.0 > > > > > > The initialization of the 8 beads is done by "DEBROGLIE > > CENTROID 300.0" in &PIMD and "ISOLATED MOLECULE", > > > "OPTIMIZE WAVEFUNCTION", "PATH INTEGRAL" in &CPMD. What could be > > the reason? Thank you very much! Regards, > > > > > > Lan-Feng > > > > > > ---------------------------- > > > Dr. Lan-Feng Yuan, > > > Postdoctoral Associate, > > > Department of Chemistry, > > > Princeton University, > > > Princeton, NJ 08544, USA > > > Office Phone: 1-609-258-1834 > > > Fax: 1-609-258-6746 > > > ---------------------------- > > > > > > _______________________________________________ > > > CPMD-list mailing list > > > CPMD-list at cpmd.org > > > http://cpmd.org/mailman/listinfo/cpmd-list > > > > > > > From eneritz.muguruza at kcl.ac.uk Thu Jul 14 11:26:47 2005 From: eneritz.muguruza at kcl.ac.uk (Eneritz Muguruza gonzalez) Date: 14 Jul 2005 10:26:47 +0100 Subject: [CPMD-list] PCG MINIMIZE with ROKS (MODIFIED GOEDECKER) Message-ID: <1121333207.22299.4.camel@phippc33> Hello, This is me again, I just notice that I made a mistake in the mail I sent. The final energy obtained combining the PCG MINIMIZE and ODIIS methods is the same obtained with the other methods(ODIIS): E=-103.6901 Ha. Sorry about the confusion, Eneritz From lyuan at Princeton.EDU Thu Jul 14 21:33:05 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Thu, 14 Jul 2005 15:33:05 -0400 Subject: [CPMD-list] SUBTRACT COMVEL ROLVEL and Nose? Message-ID: <476f1ff927137.42d685b1@Princeton.EDU> Hi, dear CPMD users, I want to make sure that the energies from CP and path integral MD only include vibrational contributions, not translational and rotational. It seems that the keyword SUBTRACT COMVEL ROLVEL would subtract global momentuam and angular momentum every given steps. The manual says that the use of these keywords is strongly recommended for long runs (e.g. >10 ps) and/or low density systems (e.g. isolated molecules, gas phase). Since I want to model water monomer and dimer, its use is "stronly recommended". However, the manual also says that since the subtracted kinetic energy is put back into the system by simple rescaling of the ionic velocities, these options is not fully compatible with NOSE thermostats ... But I surely need Nose chains in PIMD. So what to do? Thank you very much! Regards, Lan-Feng ---------------------------- Dr. Lan-Feng Yuan, Postdoctoral Associate, Department of Chemistry, Princeton University, Princeton, NJ 08544, USA Office Phone: 1-609-258-1834 Fax: 1-609-258-6746 ---------------------------- From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Thu Jul 14 22:37:49 2005 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Thu, 14 Jul 2005 22:37:49 +0200 (CEST) Subject: [CPMD-list] SUBTRACT COMVEL ROLVEL and Nose? In-Reply-To: <476f1ff927137.42d685b1@Princeton.EDU> Message-ID: On Thu, 14 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: LY> Hi, dear CPMD users, LY> dear lan-feng, LY> I want to make sure that the energies from CP and path integral MD LY> only include vibrational contributions, not translational and LY> rotational. It seems that the keyword SUBTRACT COMVEL ROLVEL would LY> subtract global momentuam and angular momentum every given steps. The LY> manual says that the use of these keywords is strongly recommended for LY> long runs (e.g. >10 ps) and/or low density systems (e.g. isolated LY> molecules, gas phase). Since I want to model water monomer and dimer, LY> its use is "stronly recommended". However, the manual also says that LY> since the subtracted kinetic energy is put back into the system by LY> simple rescaling of the ionic velocities, these options is not fully LY> compatible with NOSE thermostats ... But I surely need Nose chains in LY> PIMD. So what to do? actually, SUBTRACT COMVEL/ROTVEL is not supported by the PI-MD drivers at all. so there is not much of a choice. in general, you should be aware of the fact, that the path-integral code has not been significantly updated for years and is somewhat 'lagging behind' in terms of features compared to the other MD drivers. as for using SUBTRACT COMVEL/ROTVEL in combination with nose-hoover chains. you _can_ use it, but due to the rescaling, the ensemble is not exactly preserved anymore. there may be a smarter way of doing it, but so far nobody has come up with it. on the other hand, if you use a fequency between 20 and 100 the rescaling should be rather small. in the end it, is a trade-off between keeping the correct ensemble and losing kinetic energy into rotation and translation. best regards, axel. LY> LY> Thank you very much! Regards, LY> LY> Lan-Feng LY> LY> ---------------------------- LY> Dr. Lan-Feng Yuan, LY> Postdoctoral Associate, LY> Department of Chemistry, LY> Princeton University, LY> Princeton, NJ 08544, USA LY> Office Phone: 1-609-258-1834 LY> Fax: 1-609-258-6746 LY> ---------------------------- LY> LY> LY> _______________________________________________ LY> CPMD-list mailing list LY> CPMD-list at cpmd.org LY> http://cpmd.org/mailman/listinfo/cpmd-list LY> LY> -- ======================================================================= 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/~axel.kohlmeyer/ ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From lyuan at Princeton.EDU Thu Jul 14 23:26:41 2005 From: lyuan at Princeton.EDU (Lan-Feng Yuan (lyuan@Princeton.EDU)) Date: Thu, 14 Jul 2005 17:26:41 -0400 Subject: [CPMD-list] origin and error bar of PIMD energies Message-ID: <5ff09f1527f11.42d6a051@Princeton.EDU> Dear Axel and CPMD users, Thank you very much. So there is no choice at all ... Anyway, if I do simulations of 1-2 ps for isolated molecules, how much would these translational and rotational degrees of freedom do? Will each of them obtain 1/2kT and therefore 3kT altogether? Or we can just ignore them? In a recent test of mine, the nuclear masses of H2O are changed to 30, 30, 1, and the frequencies become 885, 1063, 3533 cm-1, whose ZPEs add up to 0.340 eV. With 8 beads, 300 K, 4 a.u. timestep, after 10000 steps of PIMD, the average quantum energy is higher than the ground state energy by 0.351 or 0.358 eV (virial or primitive estimator). 3kT here is 0.078 eV. So it seems that PIMD agrees with the harmonic vibrational energy to 0.01 or 0.02 eV, which looks good enough. From this I guess that that translation and rotation are absent in energies. But if so, it leads to another question: what is the internal error bar of PIMD? All the energy terms are vibrating depening on timestep, so there is an intrinsic spread for them. For example, in that test the quantum kinetic energy (virial) is 0.007838 +- 0.00165284 Ha, and I don't think more steps of simulation can reduce this fluctuation by much. Do we just accept it as part of error bar, or try to subtract its effect in error analysis? Thanks a lot! Regards, Lan-Feng > On Thu, 14 Jul 2005, Lan-Feng Yuan (lyuan at Princeton.EDU) wrote: > > LY> Hi, dear CPMD users, > LY> > > dear lan-feng, > > LY> I want to make sure that the energies from CP and path > integral MD > LY> only include vibrational contributions, not translational and > LY> rotational. It seems that the keyword SUBTRACT COMVEL ROLVEL would > LY> subtract global momentuam and angular momentum every given > steps. The > LY> manual says that the use of these keywords is strongly > recommended for > LY> long runs (e.g. >10 ps) and/or low density systems (e.g. isolated > LY> molecules, gas phase). Since I want to model water monomer and > dimer,LY> its use is "stronly recommended". However, the manual > also says that > LY> since the subtracted kinetic energy is put back into the > system by > LY> simple rescaling of the ionic velocities, these options is not > fullyLY> compatible with NOSE thermostats ... But I surely need > Nose chains in > LY> PIMD. So what to do? > > actually, SUBTRACT COMVEL/ROTVEL is not supported > by the PI-MD drivers at all. so there is not much > of a choice. in general, you should be aware of the > fact, that the path-integral code has not been > significantly updated for years and is somewhat > 'lagging behind' in terms of features compared to > the other MD drivers. > > as for using SUBTRACT COMVEL/ROTVEL in combination > with nose-hoover chains. you _can_ use it, but > due to the rescaling, the ensemble is not exactly > preserved anymore. there may be a smarter way of > doing it, but so far nobody has come up with it. > on the other hand, if you use a fequency between > 20 and 100 the rescaling should be rather small. > in the end it, is a trade-off between keeping the > correct ensemble and losing kinetic energy into > rotation and translation. > > best regards, > axel. > > LY> > LY> Thank you very much! Regards, > LY> > LY> Lan-Feng > LY> > LY> ---------------------------- > LY> Dr. Lan-Feng Yuan, > LY> Postdoctoral Associate, > LY> Department of Chemistry, > LY> Princeton University, > LY> Princeton, NJ 08544, USA > LY> Office Phone: 1-609-258-1834 > LY> Fax: 1-609-258-6746 > LY> ---------------------------- > LY> > LY> > LY> _______________________________________________ > LY> CPMD-list mailing list > LY> CPMD-list at cpmd.org > LY> http://cpmd.org/mailman/listinfo/cpmd-list > LY> > LY> > > -- > > ======================================================================= > Dr. Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni- > bochum.deLehrstuhl fuer Theoretische Chemie Phone: ++49 > (0)234/32-26673 > Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32- > 14045D-44780 Bochum http://www.theochem.ruhr-uni- > bochum.de/~axel.kohlmeyer/======================================================================= > If you make something idiot-proof, the universe creates a better > idiot. > From eneritz.muguruza at kcl.ac.uk Fri Jul 15 11:29:18 2005 From: eneritz.muguruza at kcl.ac.uk (Eneritz Muguruza gonzalez) Date: 15 Jul 2005 10:29:18 +0100 Subject: [CPMD-list] PCG MINIMIZE with ROKS (MODIFIED GOEDECKER) In-Reply-To: References: <1121333207.22299.4.camel@phippc33> Message-ID: <1121419758.24577.10.camel@phippc33> Dear Irmgard, > ROKS with MODIFIED GOEDECKER may show upwards convergence. That is > perfectly fine. The modification of the Goedecker algorithm was > necessary for excited states that have the same symmetry as the ground > state (e.g. pi-pi*). Thank you very much for your answer. In my PCG MINIMIZE calculations I observe the upward convergence, initially it might look like there is a lower energy minimum, but then the energy goes up no matter which recipe I use and converges to the same value obtained with ODIIS. But I don't understand this upward behaviour. Does this mean that the algorithm is not variational at each step? that it might give unphysical energies at intermediate steps? Is it possible that the modified Goedecker algorithm coupled to the line search of PCG MINIMIZE does not fully prevent the orbital rotation when is far from convergence, thus giving unphysical intermediate solutions (in fact lower in energy than the correct one)? The converged ROKS energy without the modified Goedecker is in fact lower (E=-103.7189 Ha). Many thanks in advance, Eneritz > Hello, > > > > > > This is me again, I just notice that I made a mistake in the mail I > > sent. The final energy obtained combining the PCG MINIMIZE and ODIIS > > methods is the same obtained with the other methods(ODIIS): E=-103.6901 > > Ha. > > > > Sorry about the confusion, > > > > > > Eneritz > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://cpmd.org/mailman/listinfo/cpmd-list > > From martin.konopka at stuba.sk Fri Jul 15 15:27:49 2005 From: martin.konopka at stuba.sk (Martin Konopka) Date: Fri, 15 Jul 2005 15:27:49 +0200 (CEST) Subject: [CPMD-list] transition state of a molecule Message-ID: Dear CPMD community, I am trying to locate a transition state of the CH3-N-N-CH3 molecule using the P-RFO optimizer. However, I can not see any progress towards the expected TS. The molecule essentially remains in its starting structure. In more details, the molecule has two isomers, the cis one: the trans one: C / N ------ N / / \ N ------ N / \ / C C / C and a potential energy barrier between them. I am mostly starting from the slightly randomized cis structure and trying to locate the top of the barrier. I tried several choices for reaction core and mode to be followed. Regardless of setting the mode, the result remains the same and unsatisfactory. In some cases the atoms in the reaction core do not move at all (for example in a special case when I started from a hand-modified configuration presumably closer to the TS.) I would greatly appreciate some hint to this problem. Below is my input file and modes from vibrational analysis. I tried the 4th, 7th, 10th mode and also the 1st one, which is the default. I also have a question: If I use, for example PRFO MODE 10 does it really mean that the mode corresponding to the 10th frequency of vibrational analysis is followed? If yes, then using PRFO MODE less than 7 (or 6 in special cases), which is the default, would have no sense since it corresponds to overall translations or rotations. Or do I misunderstand something? &CPMD OPTIMIZE GEOMETRY XYZ INITIALIZE WAVEFUNCTIONS ATOMS RANDOMIZE COORDINATES 0.1D0 RESTFILE 4 STORE 100 LBFGS PRFO MODE 10 PRFO CORE=4 1 2 9 10 CONVERGENCE 5.0D-7 5.0D-5 CONVERGENCE RELAX 15 SPLINE POINTS QFUNCTION 2001 ISOLATED MOLECULE CENTER MOLECULE OFF MEMORY BIG LSD &END &DFT FUNCTIONAL PBE GC-CUTOFF 5.0D-5 &END &SYSTEM SYMMETRY 0 POISSON SOLVER TUCKERMAN ANGSTROM CELL 10.0D0 1.0D0 1.0D0 0.0D0 0.0D0 0.0D0 CUTOFF 25.0D0 MULTIPLICITY 1 &END &ATOMS *N.uspp_ascii FORMATTED LMAX=P 2 4.525338176055 4.920747785642 4.949904563339 5.765416638683 5.048851794396 5.079925564279 *H.uspp_ascii FORMATTED LMAX=S 6 4.265466440716 2.788011128505 4.858413782037 2.878604693260 3.791408955331 4.306112314885 4.333398001463 3.598265838244 3.267273093363 6.467823612522 3.016976904998 5.093103687689 6.554809448253 3.823511424486 3.501228774397 7.703325625270 4.291321957329 4.803526469019 *C.uspp_ascii FORMATTED LMAX=P 2 3.972567677301 3.700817361171 4.307476081948 6.672717274443 3.979768413937 4.589493362517 &END vib. ana. for the optimized cis structure, after purification: HARMONIC FREQUENCIES [cm**-1]: 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 89.1253 160.4933 342.9720 461.3605 603.3003 854.8066 894.5454 918.9702 1033.3395 1041.1772 1122.1006 1310.9720 1322.4554 1397.1613 1406.4886 1414.6120 1437.0546 1535.1880 2896.9570 2900.3452 2961.5598 2972.3951 3036.4163 3036.7126 (The 10th mode, 461.3605 cm^-1 is particularly interesting since it changes the dihedral angle of CNNC.) I also attach my output file for convenience. Thanks. Martin Konopka. -------------- next part -------------- A non-text attachment was scrubbed... Name: H3CNNCH3.out.bz2 Type: application/octet-stream Size: 24154 bytes Desc: Url : http://cpmd.org/pipermail/cpmd-list/attachments/20050715/315a4b24/attachment.obj From eesantis at unity.ncsu.edu Fri Jul 15 16:03:25 2005 From: eesantis at unity.ncsu.edu (Erik Emilio Santiso) Date: Fri, 15 Jul 2005 10:03:25 -0400 (EDT) Subject: [CPMD-list] transition state of a molecule In-Reply-To: References: Message-ID: Dear Martin, What I have done in similar cases (searching for TS in a rotational isomerization) is to change the torsion angle by hand to a value that would be close to the maximum (in this case probably 90 degrees), and run a geometry optimization with the torsion angle constrained to this value. Then run a vibrational analysis, check that there is an imaginary frequency (or at least that the softest mode corresponds to the torsion) and then run the PRFO re-using the Hessian from the vibrational analysis. This usually works well. The only thing that worries me about your molecule is that it looks very similar to 2-butene (i.e., a double bond in the middle). When the molecule rotates you will break the double bond and end up with a diradical, which is tricky to model properly. I remember testing this with 2-butene a while ago, and what you get (even with LSD) is a much lower barrier than the real one. If anybody in the list has a suggestion on how to handle this particular problem, I'd like to hear about it too. Thanks! Erik Santiso. On Fri, 15 Jul 2005, Martin Konopka wrote: > Dear CPMD community, > > I am trying to locate a transition state of the CH3-N-N-CH3 molecule > using the P-RFO optimizer. However, I can not see any progress towards > the expected TS. The molecule essentially remains in its starting > structure. In more details, the molecule has two isomers, > > the cis one: the trans one: > > C > / > N ------ N / > / \ N ------ N > / \ / > C C / > C > > and a potential energy barrier between them. I am mostly starting from > the slightly randomized cis structure and trying to locate the top > of the barrier. I tried several choices for reaction core and mode to be > followed. Regardless of setting the mode, the result remains the same and > unsatisfactory. In some cases the atoms in the reaction core do not move > at all (for example in a special case when I started from a > hand-modified configuration presumably closer to the TS.) > > I would greatly appreciate some hint to this problem. > Below is my input file and modes from vibrational analysis. > > I tried the 4th, 7th, 10th mode and also the 1st one, which is the > default. I also have a question: If I use, for example > PRFO MODE > 10 > does it really mean that the mode corresponding to the 10th frequency of > vibrational analysis is followed? If yes, then using PRFO MODE less than > 7 (or 6 in special cases), which is the default, would have no sense since > it corresponds to overall translations or rotations. Or do I > misunderstand something? > > > &CPMD > OPTIMIZE GEOMETRY XYZ > INITIALIZE WAVEFUNCTIONS ATOMS > RANDOMIZE COORDINATES > 0.1D0 > RESTFILE > 4 > STORE > 100 > LBFGS > PRFO MODE > 10 > PRFO CORE=4 > 1 2 9 10 > CONVERGENCE > 5.0D-7 5.0D-5 > CONVERGENCE RELAX > 15 > SPLINE POINTS QFUNCTION > 2001 > ISOLATED MOLECULE > CENTER MOLECULE OFF > MEMORY BIG > LSD > &END > > &DFT > FUNCTIONAL PBE > GC-CUTOFF > 5.0D-5 > &END > > &SYSTEM > SYMMETRY > 0 > POISSON SOLVER TUCKERMAN > ANGSTROM > CELL > 10.0D0 1.0D0 1.0D0 0.0D0 0.0D0 0.0D0 > CUTOFF > 25.0D0 > MULTIPLICITY > 1 > &END > > &ATOMS > *N.uspp_ascii FORMATTED > LMAX=P > 2 > 4.525338176055 4.920747785642 4.949904563339 > 5.765416638683 5.048851794396 5.079925564279 > *H.uspp_ascii FORMATTED > LMAX=S > 6 > 4.265466440716 2.788011128505 4.858413782037 > 2.878604693260 3.791408955331 4.306112314885 > 4.333398001463 3.598265838244 3.267273093363 > 6.467823612522 3.016976904998 5.093103687689 > 6.554809448253 3.823511424486 3.501228774397 > 7.703325625270 4.291321957329 4.803526469019 > *C.uspp_ascii FORMATTED > LMAX=P > 2 > 3.972567677301 3.700817361171 4.307476081948 > 6.672717274443 3.979768413937 4.589493362517 > &END > > > vib. ana. for the optimized cis structure, after purification: > > HARMONIC FREQUENCIES [cm**-1]: > > 0.0000 0.0000 0.0000 0.0000 > 0.0000 0.0000 89.1253 160.4933 > 342.9720 461.3605 603.3003 854.8066 > 894.5454 918.9702 1033.3395 1041.1772 > 1122.1006 1310.9720 1322.4554 1397.1613 > 1406.4886 1414.6120 1437.0546 1535.1880 > 2896.9570 2900.3452 2961.5598 2972.3951 > 3036.4163 3036.7126 > > (The 10th mode, 461.3605 cm^-1 is particularly interesting since it > changes the dihedral angle of CNNC.) > I also attach my output file for convenience. > > Thanks. > Martin Konopka. > From ZRLSRB at ch.ibm.com Fri Jul 15 16:13:54 2005 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Fri, 15 Jul 2005 16:13:54 +0200 Subject: [CPMD-list] transition state of a molecule Message-ID: Hi Martin, > and a potential energy barrier between them. I am mostly starting from > the slightly randomized cis structure and trying to locate the top > of the barrier. I tried several choices for reaction core and mode