From balducci at dschsun1.univ.trieste.it Thu May 2 20:24:53 2002 From: balducci at dschsun1.univ.trieste.it (gabriele balducci) Date: Thu, 02 May 2002 19:24:53 +0100 Subject: [CPMD-list] (no subject) Message-ID: <200205021824.g42IOrGl011699@dschsun1.univ.trieste.it> Dear all, I'm quite new to this list and to QM calculations in general, so apologies if my question has an obvious answer. I'd like to study the CeO_2 system and have generated ultrasoft pseudo potentials for cerium and oxygen using Vanderbilt's code. However, from the CPMD (v.3.5) manual, I seem to understand that the maximum admissible L value is 2 (LMAX=D, p.38 of the manual). On the other hand, cerium has partially occupied 4f orbitals: have I to conclude that CPMD cannot handle lanthanides? Also, what is the meaning of the sentence on p. 39 of the manual: "Note: Also for Vanderbilt and Goedecker pseudopotentials this line has to be in a valid format, but the actual values are not used". Does this mean that Vanderbilt's PP bypass the LMAX limitation? I thank you very much in advance for any hint Ciao =========================================================================== Gabriele Balducci tel I-040-6763957 Dipartimento di Scienze Chimiche fax I-040-6763903 Via L. Giorgieri 1 e-mail balducci at univ.trieste.it I-34127 TRIESTE =========================================================================== From hutter at pci.unizh.ch Mon May 6 09:21:58 2002 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Mon, 6 May 2002 09:21:58 +0200 (MEST) Subject: [CPMD-list] (no subject) In-Reply-To: <200205021824.g42IOrGl011699@dschsun1.univ.trieste.it> Message-ID: On Thu, 2 May 2002, gabriele balducci wrote: > Dear all, > > I'm quite new to this list and to QM calculations in general, so > apologies if my question has an obvious answer. > > I'd like to study the CeO_2 system and have generated ultrasoft pseudo > potentials for cerium and oxygen using Vanderbilt's code. However, > from the CPMD (v.3.5) manual, I seem to understand that the maximum > admissible L value is 2 (LMAX=D, p.38 of the manual). On the other > hand, cerium has partially occupied 4f orbitals: have I to conclude > that CPMD cannot handle lanthanides? There is a limit to s, p, and d states for Vanderbilt PP in CPMD. The reason is that the Clebsch-Gordon coefficients are only programmed up to d-d (see aainit.F). Another restriction is that all angular momenta channels have to have the same number of projectors. In addition there seem to be problems with the current version when used with VDB PP. Several people work on this problem and also on the interface to the newest version of Vanderbilts atomic code. > > Also, what is the meaning of the sentence on p. 39 of the manual: > "Note: Also for Vanderbilt and Goedecker pseudopotentials this line > has to be in a valid format, but the actual values are not used". Does > this mean that Vanderbilt's PP bypass the LMAX limitation? > For these PP the values used in the code are provided by the PP file not the input file. Still, the line in the input file has to have a valid format in order that the input parser does not fail. Juerg > > I thank you very much in advance for any hint > > Ciao > =========================================================================== > Gabriele Balducci tel I-040-6763957 > Dipartimento di Scienze Chimiche fax I-040-6763903 > Via L. Giorgieri 1 e-mail balducci at univ.trieste.it > I-34127 TRIESTE > =========================================================================== > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From dyyang at po.iams.sinica.edu.tw Tue May 7 03:05:19 2002 From: dyyang at po.iams.sinica.edu.tw (dyyang) Date: Tue, 07 May 2002 09:05:19 +0800 Subject: [CPMD-list] problem with test run Message-ID: <3CD7284F.62BA5A01@po.iams.sinica.edu.tw> Dear all: I will be appreciate any body can teach us the following question. A. We have tried to test H2O single molecule in out deck XP1000 machine. The input file of our pseudopot. is &ATOM atom=O core=1 valence=2 lda = BLYP gcx = BECKE gcc = LYP states 1 S 2. 2 S 2. 2 P 4. MARTINS WAVEFUNCTIONS channels 3 0 0.70 0.0 1 0.80 0.0 2 0.80 0.0 rccore 0.50 &END and our CPMD input is &CPMD OPTIMIZE GEOMETRY HESSIAN DISCO CONVERGENCE ORBITALS GEOMETRY 5.D-4 5.D-3 GDIIS 50 STRUCTURE BONDS ANGLES DIHEDRALS ISOLATED MOLECULE &END &SYSTEM ANGSTROM SYMMETRY 0 CELL 11.0 1.0 1.0 0.0 0.0 0.0 CUTOFF 50. &END &ATOMS *O_bhs LMAX=P 1 10.761353 10.000000 8.485000 1 *H_bhs LMAX=S 2 10.761353 10.000000 7.525000 1 9.856418 10.000000 8.805455 1 &END &DFT OLDCODE GRADIENT CORRECTION BECKE88 LYP LDA CORRELATION LYP GC-CUTOFF 1.0E-08 &END &PROP PROJECT WAVEFUNCTION &END What we obtained are 1. in XP1000 it takes over 16 hrs to run. I think it is not right. 2. By adjusting the cutoff rc in the pseudopot. input, we can improve the angle of HOH to close to 104 degree, say 106 degree. But no matter what we have tried, we always get the OH bond legth is longer than 1.12 A. Can anybody teach me how to adjust the bond length. When I check the paper by M. Sprik, J. Hutter and M. Parrinello, JCP105(1996)1142. Table I shows OH bond legth is ca. 0.97 A. B. We also test Si surface by putting si atoms on a (100) surface. After long run, the si atoms increase their distance and makes the lattice structure disordered a little bit. We don't think we did it in a right way. Can anybody suggest how to fix this problem. &CPMD MOLECULAR DYNAMICS RHOOUT MOVIE MAXSTEP 500 EMASS 200 TIMESTEP 3 TEMPERATURE 300 TEMPCONTROL IONS 300 1 NOSE IONS 300 2000 NOSE PARAMETERS 4 4 4 6 15 2 &END &SYSTEM SYMMETRY 6 ANGSTROM CELL 29.0376 1.0000 0.4987 0.0 0.0 0.0 CUTOFF 13. &END &ATOMS *Si_bhs LMAX=P 112 8.0000 3.0000 7.2429 1 8.0000 6.8396 7.2429 1 8.0000 4.9198 8.6004 1 ......................................................... 17.5990 23.3736 4.7500 1 17.5990 21.0223 4.7500 1 &END &DFT FUNCTIONAL LDA LDA CORRECTION LYP GRADIENT CORRECTION BECKE88 GC-CUTOFF 1.D-7 &END &PROP PROJECT WAVEFUNCTION &END C. I red one of paper by N.L. Doltsinis and D. Max, PRL,88(2002)166402. They said that non-adiabatic CPMD is implemented in CPMD3.4. Can anybody tell me which part of manual we can find it? D. Can people run CPMD with localized initial charge distribution. For example, if we obtain a localized electron distribution from gaussian98 and then use it as an input distribution to see how these charge flows out or transfered. How to do it? Best regards. -------------------------- Sincerely Yours Dr. Dah-Yen Yang Institute of Atomic and Molecular Science Academia Sinica, Taipei, Taiwan Tel:886-2-2366-8239 Fax:886-2-2362-0200 E-mail:dyyang at po.iams.sinica.edu.tw http://www.iams.sinica.edu.tw/lab/dyyang/indexa.html From hutter at pci.unizh.ch Tue May 7 12:08:37 2002 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 7 May 2002 12:08:37 +0200 (MEST) Subject: [CPMD-list] problem with test run In-Reply-To: <3CD7284F.62BA5A01@po.iams.sinica.edu.tw> Message-ID: On Tue, 7 May 2002, dyyang wrote: > Dear all: > I will be appreciate any body can teach us the following question. > > A. We have tried to test H2O single molecule in out deck XP1000 machine. > > The input file of our pseudopot. is > &ATOM > atom=O > core=1 > valence=2 > lda = BLYP > gcx = BECKE > gcc = LYP > states > 1 S 2. > 2 S 2. > 2 P 4. > MARTINS > WAVEFUNCTIONS > channels 3 > 0 0.70 0.0 > 1 0.80 0.0 > 2 0.80 0.0 > rccore 0.50 > &END The rc values (0.7 - 0.8) are very small for oxygen. This PP most likely will need a very high cutoff (beyond 100 Ry) to give reasonable results. An rc of 1.05 and lmax=p with a cutoff of 70 Ry worked for us in the past, but other groups have used rc of 1.40 and lmax=d with lower cutoffs as well. > > and our CPMD input is > > &CPMD > OPTIMIZE GEOMETRY > HESSIAN DISCO > CONVERGENCE ORBITALS GEOMETRY > 5.D-4 5.D-3 If you optimize the orbitals only to 5.D-4 your forces will not be very accurate and it may happen that the geometry optimization will not converge. The default of 1.D-5 is a very reasonable number and you should only use larger values if there a convergence problems and then only at the beginning of geometry optimizations. > GDIIS > 50 > STRUCTURE BONDS ANGLES DIHEDRALS > ISOLATED MOLECULE > &END > &SYSTEM > ANGSTROM > SYMMETRY > 0 > CELL > 11.0 1.0 1.0 0.0 0.0 0.0 > CUTOFF > 50. > &END You have a rather big box (good to get accurate results) and this will be expensive especially if you use the Hockney Poisson solver (the default, so that's what you are using). For this type of system and box sizes you would get the same results by using the Tuckerman Poisson solver at a much lower cost. As mentioned above, 50 Ry is probably a too low cutoff for the PP you have generated. > &ATOMS > *O_bhs > LMAX=P > 1 > 10.761353 10.000000 8.485000 1 > *H_bhs > LMAX=S > 2 > 10.761353 10.000000 7.525000 1 > 9.856418 10.000000 8.805455 1 > &END I assume you have named your Martins-Trouiller PP O_bhs (bhs stands for Bachelet-Hamann-Schluter another type of PP) You are using (the default) Gauss-Hermite integration scheme for the nonlocal part of the PP. This is again much more expensive then using the KLEINMAN-BYLANDER scheme that works very niceley for this type of systems. In fact the GH scheme is only used for complicated electronic structures mostly in transition metal ions. > &DFT > OLDCODE Why using the OLDCODE keyword here. The default (NEWCODE) is probably much faster > GRADIENT CORRECTION BECKE88 LYP > LDA CORRELATION LYP > GC-CUTOFF > 1.0E-08 > &END > &PROP > PROJECT WAVEFUNCTION > &END > > What we obtained are > 1. in XP1000 it takes over 16 hrs to run. I think it is not right. > Was this due to bad convergence of the geometry optimization or the time for a single SCF iteration? With the changes mentioned above it should be possible to reduce both times considerably. > 2. By adjusting the cutoff rc in the pseudopot. input, we can improve > the angle of HOH to close to 104 degree, say 106 degree. > But no matter what we have tried, we always get the OH bond legth is > longer than 1.12 A. Can anybody teach me how to adjust > the bond length. When I check the paper by M. Sprik, J. Hutter and M. > Parrinello, JCP105(1996)1142. Table I shows OH bond legth is ca. 0.97 A. > See above, we used rc=1.05 for O and rc=0.5 for H > B. > We also test Si surface by putting si atoms on a (100) surface. After > long run, the si atoms increase their distance and makes the lattice > structure disordered a little bit. We don't think we did it in a right > way. Can anybody suggest how to fix this problem. > > &CPMD > MOLECULAR DYNAMICS > RHOOUT > MOVIE > MAXSTEP > 500 > EMASS > 200 > TIMESTEP > 3 I would have used emass=400 and timestep=5 but your values should work fine. I'm missing a RESTART statement here. Before doing MD you have to have a optimized wavefunction. > TEMPERATURE > 300 > TEMPCONTROL IONS > 300 1 > NOSE IONS > 300 2000 > NOSE PARAMETERS > 4 4 4 6 15 2 You can not have TEMPCONTROL IONS and NOSE IONS at the same time. They are competing methods that try to achive the same goal. > &END > &SYSTEM > SYMMETRY > 6 > ANGSTROM > CELL > 29.0376 1.0000 0.4987 0.0 0.0 0.0 > CUTOFF > 13. > &END I assume that you have checked the box and the geometry. How much empty space is there between the slabs and how many atomic layers are in your slab. These are the two most important questions in surface calculations. In general, surfaces are non-trivial and need special care. > &ATOMS > *Si_bhs > LMAX=P > 112 > 8.0000 3.0000 7.2429 1 > 8.0000 6.8396 7.2429 1 > 8.0000 4.9198 8.6004 1 > ......................................................... > 17.5990 23.3736 4.7500 1 > 17.5990 21.0223 4.7500 1 > &END Same problem as with the water calcultion. use the KLEINMAN-BYLANDER scheme > &DFT > FUNCTIONAL LDA > LDA CORRECTION LYP > GRADIENT CORRECTION BECKE88 > GC-CUTOFF > 1.D-7 > &END Here you are mixing two ways of specifying the functional. The FUNCTIONAL keyword is a shortcut to specify most common functionals, and you are asking for LDA. Then you additionaly specify other parts, I don't know what the functional (combination of parts of functionals) this results in. Have a look in the ouput if this gives anything reasonable (most likely not). My advice is to use only the FUNCTIONAL keyword (e.g. FUNCTIONAL BLYP) > &PROP > PROJECT WAVEFUNCTION > &END > > > C. I red one of paper by N.L. Doltsinis and D. Max, PRL,88(2002)166402. > They said that non-adiabatic CPMD is implemented in CPMD3.4. Can anybody > tell me which part of manual we can find it? > The methods described in this paper, surface hopping with excited state surfaces calculated from ROKS where implemented based on version 3.4 of CPMD. The ROKS method is part of CPMD 3.5 (see manual LOW SPIN EXCITATIONS) and is still being further developped. The non-adiabatic dynamics is not in the available code, as this new development has not (yet?) been submitted to be included. > D. > Can people run CPMD with localized initial charge distribution. For > example, if we obtain a localized electron distribution from gaussian98 > and then use it as an input distribution to see how these charge flows > out or transfered. How to do it? > CPMD uses atomic densities as initial guess to wavefunction optimisations. If you want to use other densities you have to change the code (wouldn't be too hard to do in this case) If you want to use these densities to do CPMD, this will not work as you will be too far from the Born-Oppenheimer surface to do reasonable dynamics. In addition, keep in mind, the CP dynamics of the electrons is ficticious and has no direct physical meaning. regards Juerg Hutter > Best regards. > > -------------------------- > Sincerely Yours > > Dr. Dah-Yen Yang > > Institute of Atomic and Molecular Science > Academia Sinica, Taipei, Taiwan > > Tel:886-2-2366-8239 > Fax:886-2-2362-0200 > E-mail:dyyang at po.iams.sinica.edu.tw > http://www.iams.sinica.edu.tw/lab/dyyang/indexa.html > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From atte.sillanpaa at oulu.fi Wed May 8 10:07:01 2002 From: atte.sillanpaa at oulu.fi (=?ISO-8859-1?Q?Atte_Sillanp=E4=E4?=) Date: Wed, 8 May 2002 11:07:01 +0300 (EEST) Subject: [CPMD-list] Disagreement with the test run Message-ID: Hello all, I'm trying to confirm, that my executable CPMD 3.5 is correct. I've downloaded and run the 32 water molecule test cases from www.cpmd.org. I get slightly different results with the exaple output. I have performed three trial runs with differently compiled executables. Namely, one of version 3.4.3 runnning on T3E, one compiled with optimization of the code turned on of the 3.5 version on IBM-SP4, and finally one of the 3.5 version with the debugging options turned on with the IBM-SP4 platform. The results of my runs give practically identical results, but these differ from the example output quite a bit. Here are the total energies for comparison: T3E -549.3152869 IBM-SP4 (optimized) -549.3152208 IBM-SP4 (debug) -549.3152208 sample output (w32_wfSP2) -548.7794744 Is this as things should be, i.e. should I worry about the compilations? I guess the total energies should not differ this much with different platforms. Sincerely, Atte atte.sillanpaa at oulu.fi +358 (0)8 553 1681 (work), KE 368 Dept. of Physical Chemistry +358 (0)40 592 7369 (gsm) Po-BOX 3000 FIN-90014 University of Oulu FINLAND | Entropy requires no maintenance. | From iivanov at cmm.chem.upenn.edu Wed May 8 16:49:56 2002 From: iivanov at cmm.chem.upenn.edu (Ivaylo Ivanov) Date: 08 May 2002 10:49:56 -0400 Subject: [CPMD-list] path integrals Message-ID: <1020869396.25065.113.camel@quicksilver> Hello all, I would like to find out if anyone has had trouble running path integral calculations on the Compaq machines in Pittsburgh (either Lemieux or TCSINI). Normal CPMD jobs seem to have no problems.Path integral jobs die in groups.f apparently at the first call to MY_BCAST.I don't think that the problem is with the BCAST itself since it is a very simple wrapper for calling MPI_Bcast and it is used throughout the rest of the code.More likely it is related to the construction of the processor groups in pi_init.f: CALL MY_GROUP(NPROC,PGROUP(1),LABEL,ALLGRP) CALL MY_SYNC(SUPERGROUP)...etc The code is running just fine on the IBM SP3 and T3E architectures.I would greatly appreciate any input or suggestions that you can give me. Thanks very much in advance. Sincerely, Ivaylo Ivanov -- Ivaylo Ivanov Center for Molecular Modeling | University of Pennsylvania 231 S.34 Street, | Philadelphia,PA 19104-6323 tel: (215) 573 8697 | fax: (215) 573 6233 e-mail: iivanov at cmm.chem.upenn.edu From alex.merchant at rmit.edu.au Thu May 9 03:48:16 2002 From: alex.merchant at rmit.edu.au (Alex Merchant) Date: Thu, 09 May 2002 11:48:16 +1000 Subject: [CPMD-list] Lahey F90 Compiler Message-ID: Hi everyone, My questions are: * Has anybody sucessfully used Lahey-Fujitsu to compile CPMD ? If so what compiler options did you use ? * Can anyone suggest how I might go about passing the "-Y" variable into the cpp compiler using Lahey Fortran ? See below for details: Our department has recently migrated to PC-Linux boxes running [I think] RedHat (at least my env says HOSTTYPE=i586-linux, VENDOR=intel, OSTYPE=linux MACHTYPE=i586). Anyway, my experience is using COMPAQ SC80 and DEC-ALPHA ./Configure options not PC-IFC, PC-PGI, PC-ABSOFT! Regardless we are using Lahey-Fujitsu fortran compilers not ABSOFT or IFC. Using the compiler option "ABSOFT" comes close but we get the error: >> /usr/bin/f90 -c -B108 -B100 -YEXT_NAMES=LCS -O ./util.f -o ./util.o >> f95: warning: Invalid suboption specified for -Y. >> 2018-S: "./util.f", line 183: When IMPLICIT NONE is specified, I must be declared in a type declaration statement. >> Encountered 1 error, 0 warnings in file ./util.f. >> rm -f timetag.f >> /lib/cpp -P -C -D__NOINT8 -D__Linux -D__ABSOFT -DLAPACK -DFFT_DEFAULT ./timetag.F ./timetag.f >> /usr/bin/f90 -c -B108 -B100 -YEXT_NAMES=LCS -O ./timetag.f >> f95: warning: Invalid suboption specified for -Y. >> Encountered 0 errors, 0 warnings in file ./timetag.f. >> etc I understand that the "-Y" option is a variable passed to the "cpp" compiler but I cannot find what the equivalent Lahey F90 command to pass the variable to "cpp". Any help would be most appreciated. I've used the standard INCLUDES from the lahey-f90 before compiling. Alex From hutter at pci.unizh.ch Thu May 9 18:06:19 2002 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Thu, 9 May 2002 18:06:19 +0200 (MEST) Subject: [CPMD-list] Disagreement with the test run In-Reply-To: Message-ID: Hi The test runs are rather old and it might be that some default values have changed. But most likely the difference comes from different real space grids. Check the number of grid points used in the calculations. The program usually chooses the smallest possible grid compatible with the FFT library. The FFT libraries of different vendors and the default library in CPMD don't allow always for the same grids. About the defaults: check the number of spline points used to expand the pseudopotential and the XC functional, especially was there a tabulation of the functional. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Wed, 8 May 2002, Atte Sillanp?? wrote: > Hello all, > > I'm trying to confirm, that my executable CPMD 3.5 is correct. I've downloaded > and run the 32 water molecule test cases from www.cpmd.org. I get slightly > different results with the exaple output. I have performed three trial runs with > differently compiled executables. Namely, one of version 3.4.3 runnning on T3E, > one compiled with optimization of the code turned on of the 3.5 version on > IBM-SP4, and finally one of the 3.5 version with the debugging options turned on > with the IBM-SP4 platform. The results of my runs give practically identical > results, but these differ from the example output quite a bit. Here are the > total energies for comparison: > > T3E -549.3152869 > IBM-SP4 (optimized) -549.3152208 > IBM-SP4 (debug) -549.3152208 > sample output (w32_wfSP2) -548.7794744 > > Is this as things should be, i.e. should I worry about the compilations? I guess > the total energies should not differ this much with different platforms. > > Sincerely, > > Atte > > atte.sillanpaa at oulu.fi +358 (0)8 553 1681 (work), KE 368 > Dept. of Physical Chemistry +358 (0)40 592 7369 (gsm) > Po-BOX 3000 > FIN-90014 University of Oulu > FINLAND > | Entropy requires no maintenance. | > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Thu May 9 18:10:30 2002 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Thu, 9 May 2002 18:10:30 +0200 (MEST) Subject: [CPMD-list] Lahey F90 Compiler In-Reply-To: Message-ID: Hi can't help you directly with the Lahey compiler, but the -Y option of the Absoft compiler is not used by the cpp. It determines how the compiler names subroutines internally (lower case, upper case, appending underscore, etc). This is important to match names with external libraries. Most likely you should ignore this option for your compiler or replace it with the appropriate option of the Lahey compiler. This option has nothing to do with CPMD directly. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Thu, 9 May 2002, Alex Merchant wrote: > Hi everyone, > > My questions are: > > * Has anybody sucessfully used Lahey-Fujitsu to compile CPMD ? If so > what compiler options did you use ? > * Can anyone suggest how I might go about passing the "-Y" variable > into the cpp compiler using Lahey Fortran ? > > See below for details: > > Our department has recently migrated to PC-Linux boxes running [I > think] RedHat > (at least my env says HOSTTYPE=i586-linux, VENDOR=intel, OSTYPE=linux > MACHTYPE=i586). > > Anyway, my experience is using COMPAQ SC80 and DEC-ALPHA ./Configure > options not PC-IFC, PC-PGI, PC-ABSOFT! Regardless we are using > Lahey-Fujitsu > fortran compilers not ABSOFT or IFC. Using the compiler option "ABSOFT" > comes > close but we get the error: > > >> /usr/bin/f90 -c -B108 -B100 -YEXT_NAMES=LCS -O ./util.f -o > ./util.o > >> f95: warning: Invalid suboption specified for -Y. > >> 2018-S: "./util.f", line 183: When IMPLICIT NONE is specified, I > must be declared in a type declaration statement. > >> Encountered 1 error, 0 warnings in file ./util.f. > >> rm -f timetag.f > >> /lib/cpp -P -C -D__NOINT8 -D__Linux -D__ABSOFT -DLAPACK > -DFFT_DEFAULT ./timetag.F ./timetag.f > >> /usr/bin/f90 -c -B108 -B100 -YEXT_NAMES=LCS -O ./timetag.f > >> f95: warning: Invalid suboption specified for -Y. > >> Encountered 0 errors, 0 warnings in file ./timetag.f. > >> etc > > I understand that the "-Y" option is a variable passed to the "cpp" > compiler but I cannot find > what the equivalent Lahey F90 command to pass the variable to "cpp". > > Any help would be most appreciated. I've used the standard INCLUDES > from the lahey-f90 > before compiling. > > Alex > > > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From kulhanek at chemi.muni.cz Thu May 9 22:40:30 2002 From: kulhanek at chemi.muni.cz (=?iso-8859-2?Q?Petr_Kulh=E1nek?=) Date: Thu, 9 May 2002 22:40:30 +0200 Subject: [CPMD-list] Rotation in Isolated Molecule MD Message-ID: <003b01c1f799$c2489c40$663cfb93@botulin> Dear cpmd users, I have some problems with isolated molecule MD. When I start MD the molecule always gets rotational moment.I think that this is not right because rotational energy cannot be exchange with vibrational one (no colisions exist in isolated molecule aproach). Do you have any idea how this rotational moment can be removed or at least quantified ? Thanks in advance, Petr Kulhanek ****************************************************************** Petr Kulhanek e-mail: kulhanek at chemi.muni.cz fax: +420-5-41129506 National Centre for Biomolecular Research Faculty of Science Kotlarska 2 611 37 Brno Czech Republic ****************************************************************** From francesco.filippone at mlib.cnr.it Fri May 10 09:31:14 2002 From: francesco.filippone at mlib.cnr.it (Francesco Filippone) Date: Fri, 10 May 2002 09:31:14 +0200 (CEST) Subject: [CPMD-list] Rotation in Isolated Molecule MD In-Reply-To: <003b01c1f799$c2489c40$663cfb93@botulin> Message-ID: Dear Petr, in CPMD only cartesian coordinates are used, i.e. the initialization of the velocities in a MD run is not done with any internal coordinate at all. And using cartesian coordinates inevitably means adding some contribution from translations and rotations of the molecule. In secder.F and phonons_p.F there are two "twin" couples of routines, settrans and setrotat, and settras and setrot that build the "a priori" translations and rotations to be used in the vibrational analysis. They are basically the same. You can try to use them as starting point either to implement a new MD constraint either to project out these degrees of freedom from the TRAJECTORY in the post-processing. Ciao, F. On Thu, 9 May 2002, [iso-8859-2] Petr Kulh?nek wrote: > Dear cpmd users, > > I have some problems with isolated molecule MD. When I start MD the molecule > always gets rotational moment.I think that this is not right because > rotational energy cannot be exchange with vibrational one (no colisions > exist in isolated molecule aproach). Do you have any idea how this > rotational moment can be removed or at least quantified ? > Thanks in advance, > > Petr Kulhanek > -- ---------------------------------------------------------------------------- Francesco Filippone | Tanto gentile e CNR - ISTITUTO DI STRUTTURA DELLA MATERIA | tant'onesta pare v. Salaria Km 29,300 - C.P. 10 | la donna mia quand'ella I 00016 - Monterotondo Stazione (RM) | altrui saluta... Tel + 39 06 90672342 - Fax +39 06 90672316 | e-mail: | Dante ---------------------------------------------------------------------------- From zcfeng at dicp.ac.cn Wed May 15 06:14:10 2002 From: zcfeng at dicp.ac.cn (zhaochifeng) Date: Wed, 15 May 2002 12:14:10 +0800 Subject: [CPMD-list] (no subject) Message-ID: <200205150413.g4F4Doj12105@smtp.dicp.ac.cn> Dear Sir, I send my eamil to the mailing list Best regards! zcfeng at dicp.ac.cn website: www.canli.dicp.ac.cn State Key Laboratory of Catalysis Dalian Institute of Chemical Physics Chinese Academy of Sciences Dalian 116023, P.R.China Tel: (86)-411-4379302 (Lab) Fax: (86)-411-4694447 zhaochifeng zcfeng at dicp.ac.cn From weiz at mail.rochester.edu Wed May 15 19:13:11 2002 From: weiz at mail.rochester.edu (weizhuang) Date: Wed, 15 May 2002 13:13:11 -0400 Subject: [CPMD-list] help on parellel computing Message-ID: <200205151711.g4FHBTsT14989221@mail.rochester.edu> Hi, friends: I was running CPMD program on a linux cluster, I set the restart file to be saved every 5 steps. however, every time when the machine is going to save the restart file. the job is crashed. and following is the information. could anybody give me some clue about what is wrong here and any suggestion of how to solve it. thanks a lot. wei zhuang ----------------- MPI_Recv: process in local group is dead (rank 4, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 1, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 8, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 2, MPI_COMM_WORLD) Rank (1, MPI_COMM_WORLD): Call stack within LAM: Rank (1, MPI_COMM_WORLD): - MPI_Recv() Rank (1, MPI_COMM_WORLD): - MPI_Barrier() Rank (1, MPI_COMM_WORLD): - main() Rank (2, MPI_COMM_WORLD): Call stack within LAM: Rank (2, MPI_COMM_WORLD): - MPI_Recv() Rank (2, MPI_COMM_WORLD): - MPI_Barrier() Rank (2, MPI_COMM_WORLD): - main() MPI_Recv: process in local group is dead (rank 5, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 9, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 10, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 3, MPI_COMM_WORLD) Rank (4, MPI_COMM_WORLD): Call stack within LAM: Rank (4, MPI_COMM_WORLD): - MPI_Recv() Rank (4, MPI_COMM_WORLD): - MPI_Barrier() Rank (4, MPI_COMM_WORLD): - main() Rank (3, MPI_COMM_WORLD): Call stack within LAM: Rank (3, MPI_COMM_WORLD): - MPI_Recv() Rank (3, MPI_COMM_WORLD): - MPI_Barrier() Rank (3, MPI_COMM_WORLD): - main() Rank (5, MPI_COMM_WORLD): Call stack within LAM: Rank (5, MPI_COMM_WORLD): - MPI_Recv() Rank (5, MPI_COMM_WORLD): - MPI_Barrier() Rank (5, MPI_COMM_WORLD): - main() MPI_Recv: process in local group is dead (rank 6, MPI_COMM_WORLD) Rank (6, MPI_COMM_WORLD): Call stack within LAM: Rank (6, MPI_COMM_WORLD): - MPI_Recv() Rank (6, MPI_COMM_WORLD): - MPI_Barrier() Rank (6, MPI_COMM_WORLD): - main() MPI_Recv: process in local group is dead (rank 11, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 12, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 13, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 14, MPI_COMM_WORLD) MPI_Recv: process in local group is dead (rank 7, MPI_COMM_WORLD) Rank (9, MPI_COMM_WORLD): Call stack within LAM: Rank (9, MPI_COMM_WORLD): - MPI_Recv() Rank (9, MPI_COMM_WORLD): - MPI_Barrier() Rank (9, MPI_COMM_WORLD): - main() Rank (7, MPI_COMM_WORLD): Call stack within LAM: Rank (7, MPI_COMM_WORLD): - MPI_Recv() Rank (7, MPI_COMM_WORLD): - MPI_Barrier() Rank (7, MPI_COMM_WORLD): - main() Rank (8, MPI_COMM_WORLD): Call stack within LAM: Rank (8, MPI_COMM_WORLD): - MPI_Recv() Rank (8, MPI_COMM_WORLD): - MPI_Barrier() Rank (8, MPI_COMM_WORLD): - main() Rank (10, MPI_COMM_WORLD): Call stack within LAM: Rank (10, MPI_COMM_WORLD): - MPI_Recv() Rank (10, MPI_COMM_WORLD): - MPI_Barrier() Rank (10, MPI_COMM_WORLD): - main() Rank (12, MPI_COMM_WORLD): Call stack within LAM: Rank (12, MPI_COMM_WORLD): - MPI_Recv() Rank (12, MPI_COMM_WORLD): - MPI_Barrier() Rank (12, MPI_COMM_WORLD): - main() Rank (11, MPI_COMM_WORLD): Call stack within LAM: Rank (11, MPI_COMM_WORLD): - MPI_Recv() Rank (11, MPI_COMM_WORLD): - MPI_Barrier() Rank (11, MPI_COMM_WORLD): - main() Rank (14, MPI_COMM_WORLD): Call stack within LAM: Rank (14, MPI_COMM_WORLD): - MPI_Recv() Rank (14, MPI_COMM_WORLD): - MPI_Barrier() Rank (14, MPI_COMM_WORLD): - main() Rank (13, MPI_COMM_WORLD): Call stack within LAM: Rank (13, MPI_COMM_WORLD): - MPI_Recv() Rank (13, MPI_COMM_WORLD): - MPI_Barrier() Rank (13, MPI_COMM_WORLD): - main() MPI_Recv: process in local group is dead (rank 15, MPI_COMM_WORLD) Rank (15, MPI_COMM_WORLD): Call stack within LAM: Rank (15, MPI_COMM_WORLD): - MPI_Recv() Rank (15, MPI_COMM_WORLD): - MPI_Barrier() Rank (15, MPI_COMM_WORLD): - main() From fparnold at mercury.chem.northwestern.edu Wed May 15 21:45:21 2002 From: fparnold at mercury.chem.northwestern.edu (Fred P. Arnold) Date: Wed, 15 May 2002 14:45:21 -0500 (CDT) Subject: [CPMD-list] CPMD on Itanium? Message-ID: Hello, Pardon the simple question, but has anyone modified the CPMD 3.5 package to run on Intel IA-64 under Linux? Thank you for your time. -Fred Arnold Frederick P. Arnold, Jr. NUIT, Northwestern U. f-arnold at northwestern.edu From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Thu May 16 12:28:32 2002 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (axel.kohlmeyer at theochem.ruhr-uni-bochum.de) Date: Thu, 16 May 2002 12:28:32 +0200 Subject: [CPMD-list] help on parellel computing In-Reply-To: Your message of "Wed, 15 May 2002 13:13:11 EDT." <200205151711.g4FHBTsT14989221@mail.rochester.edu> Message-ID: <200205161028.g4GASWV02182@yello.theochem.ruhr-uni-bochum.de> >>> "WZ" == weiz writes: WZ> Hi, friends: WZ> I was running CPMD program on a linux cluster, I set the restart file to be WZ> saved every 5 steps. however, every time when the machine is going to save WZ> the restart file. the job is crashed. and following is the information. could WZ> anybody give me some clue about what is wrong here and any suggestion of how WZ> to solve it. thanks a lot. WZ> wei zhuang WZ> ----------------- WZ> MPI_Recv: process in local group is dead (rank 4, MPI_COMM_WORLD) WZ> MPI_Recv: process in local group is dead (rank 1, MPI_COMM_WORLD) WZ> MPI_Recv: process in local group is dead (rank 8, MPI_COMM_WORLD) WZ> MPI_Recv: process in local group is dead (rank 2, MPI_COMM_WORLD) hiho! it looks as if you have overloaded your network and the ethernet driver skipped a few packages (or the pci bus lost some interrupts). some combinations of pc hardware get severly overloaded if you run large parallel cpmd jobs, especially with TCP/IP networking. can you describe the kind of network (network cards, switch) and driver software (linux kernel, network driver) you are using. also it would be helpful to know, whether the same job runs single cpu and how 'large' the job, i.e. how big the restart file is. cheers, axel. -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From tdeutsch at cea.fr Thu May 16 13:29:37 2002 From: tdeutsch at cea.fr (Thierry Deutsch) Date: Thu, 16 May 2002 13:29:37 +0200 Subject: [CPMD-list] CPMD on Itanium? In-Reply-To: References: Message-ID: <15587.38945.395379.492014@absinthe.ceng.cea.fr> Hello, In the CVS repository, I will add in the Configure file the correct options to compile that are: "ITANIUM-EFC") echo 'We use the Intel Library Kernel.' >&2 if [ "${MKL_LIB}" = "" ]; then echo "The variable MKL_LIB gives the location of this library." >&2 echo "The variable MKL_LIB does not exist or is empty." >&2 exit 2 fi IRAT=2 CFLAGS='' CPP='/lib/cpp -P -C -traditional' CPPFLAGS='-D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC -DPOINTER8' FFLAGS=' -pc64 -w90 -w95' LFLAGS='-L${MKL_LIB} -lmkl_lapack -lmkl_itp -lpthread -Vaxlib' FFLAGS_GROMOS='-Dgood_luck' if [ $debug ]; then FC='efc -c -g' CC='gcc -g -Wall' LD='efc -g' else FC='efc -c -O3' CC='gcc -O2 -Wall' LD='efc -O3' fi ;; In my case, I will use the MKL library from Intel in order to have Lapack. The only difference from PC-IFC is to use -DPOINTER8. Good luck, Thierry Deutsch -- ----------------------------------------------------------------------- Thierry Deutsch Laboratoire de simulation atomistique (L_Sim) DRFMC/SP2M T?l:(33) 04 38 78 34 06 C.E.A.Grenoble Fax:(33) 04 38 78 51 97 17, Avenue des Martyrs mailto:TDeutsch at cea.fr 38054 GRENOBLE CEDEX 9 FRANCE http://www-drfmc.cea.fr/SP2M ----------------------------------------------------------------------- From ccdiazt at kaliman.iimatercu.unam.mx Wed May 22 00:42:49 2002 From: ccdiazt at kaliman.iimatercu.unam.mx (=?ISO-8859-1?Q?C=E9sar?= Carlos =?ISO-8859-1?Q?D=EDaz_Torrej=F3n?=) Date: 21 May 2002 17:42:49 -0500 Subject: [CPMD-list] Error on PC-Linux Clusters Compiler Message-ID: <1022020969.13252.8.camel@kaliman.iimatercu.unam.mx> Dear all, I am trying to compiler CPMD v. 3.5 on an cluster of 10 of 1GHz Intel PC machines running Redhat 7.2. I'm using next Makefile: #---------------------------------------------------------------------------- # Makefile for cpmd.x (plane wave electronic calculation) # Configuration: PC-PGI-MPI # Creation of Makefile: May 21 2002 # on Linux cluster.iimatercu.unam.mx 2.4.7-10 #1 Thu Sep 6 17:27:27 EDT 2001 i686 unknown # Author: cluster.iimatercu.unam.mx!ccdiazt #---------------------------------------------------------------------------- # SHELL = /bin/sh # #--------------- Default Configuration for PC-PGI-MPI --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = -Mr8 -pc64 -byteswapio -Msignextend -Msecond_underscore LFLAGS = -L/usr/pgi/linux86/lib -llapack \ -L/usr/local/mpich-1.2.4/lib -lmpich-pgf77 \ -L/usr/local/lib/ATLAS -lcblas -lf77blas -latlas -byteswapio CFLAGS = CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT \ -DPARALLEL -DMP_LIBRARY=__MPI CC = /usr/local/mpich-1.2.4/bin/mpicc -O2 -Wall FC = /usr/local/mpich-1.2.4/bin/mpif77-pgf77 -c -fast LD = /usr/local/mpich-1.2.4/bin/mpif77-pgf77 -fast AR = #---------------------------------------------------------------------------- and the end of compilation I have the next error: my_para.o: In function `my_start__': my_para.o(.text+0x1c): undefined reference to `mpi_init__' my_para.o: In function `my_end__': my_para.o(.text+0xcd7): undefined reference to `mpi_finalize__' my_para.o: In function `my_group__': my_para.o(.text+0xd1e): undefined reference to `mpi_comm_group__' my_para.o(.text+0xd4d): undefined reference to `mpi_group_incl__' my_para.o(.text+0xd74): undefined reference to `mpi_comm_create__' my_para.o: In function `my_stopall__': my_para.o(.text+0xddd): undefined reference to `mpi_abort__' my_para.o: In function `my_sync__': my_para.o(.text+0xe0f): undefined reference to `mpi_barrier__' my_para.o: In function `my_environ__': my_para.o(.text+0xe8d): undefined reference to `mpi_comm_rank__' my_para.o(.text+0xeab): undefined reference to `mpi_comm_size__' my_para.o: In function `my_bcast__': my_para.o(.text+0xf74): undefined reference to `mpi_bcast__' my_para.o: In function `my_bsend__': my_para.o(.text+0x133d): undefined reference to `mpi_send__' my_para.o: In function `my_brecv__': my_para.o(.text+0x1416): undefined reference to `mpi_recv__' my_para.o: In function `my_concat__': my_para.o(.text+0x14b1): undefined reference to `mpi_allgather__' my_para.o: In function `my_trans__': my_para.o(.text+0x15b0): undefined reference to `mpi_alltoall__' my_para.o: In function `my_combine__': my_para.o(.text+0x16fc): undefined reference to `mpi_allreduce__' my_para.o(.text+0x1753): undefined reference to `mpi_allreduce__' my_para.o(.text+0x17bc): undefined reference to `mpi_allreduce__' my_para.o(.text+0x1825): undefined reference to `mpi_allreduce__' my_para.o(.text+0x187a): undefined reference to `mpi_allreduce__' my_para.o: In function `my_shift__': my_para.o(.text+0x1951): undefined reference to `mpi_comm_rank__' my_para.o(.text+0x199e): undefined reference to `mpi_comm_size__' my_para.o(.text+0x1a64): undefined reference to `mpi_isend__' my_para.o(.text+0x1adc): undefined reference to `mpi_recv__' make: *** [cpmd.x] Error 1 I use a MPI for communications with nodes. Many thanks, Best regards, C.C.D?az -- _________________________________________________________ F?s. C?sar Carlos D?az Torrej?n Administrador de Sistemas Unix Instituto de Investigaciones en Materiales, UNAM Oficina: A-007 e-mail: ccdiazt at kaliman.iimatercu.unam.mx tel: (5)6224611 (oficina) ________________________________________________________ From t.laino at sns.it Wed May 22 14:57:27 2002 From: t.laino at sns.it (Teodoro Laino) Date: Wed, 22 May 2002 14:57:27 +0200 (CEST) Subject: [CPMD-list] Error on PC-Linux Clusters Compiler In-Reply-To: <1022020969.13252.8.camel@kaliman.iimatercu.unam.mx> Message-ID: Dear Cesar, The first thing to do is to check whether all the paths in the Makefile point to the correct place in your system (just check the mpi library paths!). If everythings looks nice then I think your problem involves how MPI was built on your cluster... that is what kind of compiler was used to build mpich-1.2.4 . The problem is that different fortran compilers, notably g77 (pgf77 too I checked it on my system just few minutes ago and its behaviour is similar to g77) and pgf90, have different ways of encoding the subroutine name "mpi_init" (for example...); pgf90 produces "mpi_init_", while g77 and pgf77 produce "mpi_init__". (Note the different amount of appended underscores.) You can try : "nm /mpich/installation/path/on/your/system/libmpich.a | grep mpi_init" and see how many underscores your library assumes. If you see just one underscore i.e. "mpi_init_" , than you can try to compile all your source code with pgf90 compiler (and see what happens...) instead of mpif77-pgf77 script, in essentially two ways : 1) modifying this script (mpif77-pgf77) specifing pgf90 as new compiler 2) writing explicitly all the mpi FFLAGS with pgf90 in command line (The 2 ways are totally equivalent..) I think that is possible with PGI compilers to switch their underscore appending style according one or the other fashion.. Have a look for example in PGI Workstation User's Guide. Hope this helps Regards Teo , , %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% /( )` %% \ \___ / | %% /- _ `-/ ' %% (/\/ \ \ /\ Teodoro Laino %% / / | ` \ %% O O ) / | PhD. in Molecular Biophysics %% `-^--'`< ' NEST - Physics Labs %% (_.) _ ) / Scuola Normale Superiore di Pisa %% `.___/` / Piazza dei Cavalieri 7 %% `-----' / 56100 PISA (Italy) %% <----. __ / __ \ %% <----|====O)))==) \) /==== %% <----' `--' `.__,' \ Tel. +39-3299624004 (Mobile) %% | | Tel. +39-050-509412 (Office) %% \ / %% ______( (_ / \______ %% ,' ,-----' | \ %% `--{__________) \/ %% %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% From ccdiazt at kaliman.iimatercu.unam.mx Wed May 22 19:47:04 2002 From: ccdiazt at kaliman.iimatercu.unam.mx (=?ISO-8859-1?Q?C=E9sar?= Carlos =?ISO-8859-1?Q?D=EDaz_Torrej=F3n?=) Date: 22 May 2002 12:47:04 -0500 Subject: [CPMD-list] Error on PC-Linux Clusters Compiler In-Reply-To: <200205220943.10873.jwh@science.uva.nl> References: <1022020969.13252.8.camel@kaliman.iimatercu.unam.mx> <200205220943.10873.jwh@science.uva.nl> Message-ID: <1022089624.20131.101.camel@kaliman.iimatercu.unam.mx> Hello All: Thanks for your answer: I use a mpich-1.2.4 and pgf77, and I try with the pgf90 and the error is the same. The problem are after compilation in the moment of the linked for generated the cpmd.x binary. I see the problem of the underscore in my library and this have "mpi_init_" but when compiling my_para.F with both pgf77 and pgf90 this have "mpi_init__". I try compiling only my_para.F with pgf77 and pgf90 (using mpif77-pgf77 and mpif90) the compilation is good but the end the same error is presente. Finally I recompiling the ATLAS libraries using pgf90 and the problem persist. Thanks again Saludos C.C.D?az On Wed, 2002-05-22 at 02:43, Jan-Willem Handgraaf wrote: > On Wednesday 22 May 2002 00:42, C?sar Carlos D?az Torrej?n wrote: > > I use a MPI for communications with nodes. > > The problem occurs when the mpi libraries are linked, after the compilation of > all the source? Which compiler did you use for the compilation of > mpich-1.2.4, pgf90 or pgf77? Perhaps there is an incompatibility? Have you > try to compile the CPMD source with the mpif90 script? > > Regards, > > Jan-Willem Handgraaf > > -- > Jan-Willem Handgraaf > Department of Chemical Engineering > University of Amsterdam > Nieuwe Achtergracht 166 > 1018 WV Amsterdam > The Netherlands > > Room: C6.18 > Tel: +31-20-5256492 > Fax: +31-20-5255604 > Email: jwh at science.uva.nl -- _________________________________________________________ F?s. C?sar Carlos D?az Torrej?n Administrador de Sistemas Unix Instituto de Investigaciones en Materiales, UNAM Oficina: A-007 e-mail: ccdiazt at kaliman.iimatercu.unam.mx tel: (5)6224611 (oficina) ________________________________________________________ From ccdiazt at kaliman.iimatercu.unam.mx Thu May 23 01:29:03 2002 From: ccdiazt at kaliman.iimatercu.unam.mx (=?ISO-8859-1?Q?C=E9sar?= Carlos =?ISO-8859-1?Q?D=EDaz_Torrej=F3n?=) Date: 22 May 2002 18:29:03 -0500 Subject: [CPMD-list] I can compiling in a cluster Message-ID: <1022110143.20131.108.camel@kaliman.iimatercu.unam.mx> Hello All: Finally I can compiling de cpmdv3.5 using MPI libraries, the change was using lam-6.5.6 with pgf90 and the next Makefile: # Author: cluster.iimatercu.unam.mx!ccdiazt #---------------------------------------------------------------------------- # SHELL = /bin/sh # #--------------- Default Configuration for PC-PGI-MPI --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = -Mr8 -pc64 -byteswapio -Msignextend -Msecond_underscore LFLAGS = -L/usr/pgi/linux86/lib -llapack \ -L/usr/local/lib/ATLAS -lf77blas -latlas -byteswapio CFLAGS = CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT \ -DPARALLEL -DMP_LIBRARY=__MPI CC = mpicc -O2 -Wall FC = mpif77 -c -fast LD = mpif77 -fast AR = #---------------------------------------------------------------------------- and the result is very good. The compilation was very clean. The libraries liblamf77mpi.a of lam have the mpi_init__ that is a old problem. Best Regards Saludos C.C.D?az -- _________________________________________________________ F?s. C?sar Carlos D?az Torrej?n Administrador de Sistemas Unix Instituto de Investigaciones en Materiales, UNAM Oficina: A-007 e-mail: ccdiazt at kaliman.iimatercu.unam.mx tel: (5)6224611 (oficina) ________________________________________________________ From markwick at NMR.EMBL-Heidelberg.DE Thu May 23 10:54:43 2002 From: markwick at NMR.EMBL-Heidelberg.DE (Phineus Markwick) Date: Thu, 23 May 2002 10:54:43 +0200 Subject: [CPMD-list] CPMD SGI-ORIGIN PROBLEM Message-ID: <3CECAE53.6040507@NMR.EMBL-Heidelberg.DE> Hallo CPMD mail list! We are having problems with CPMD on our SGI Origin 2000 (28 Proccessors, 6GB real memory). The program does compile, but CPMD aborts with a Bus error part-way through. In a simple test on H2O doing a wavefunction optimisation we get the output file below. We have tried compiling with no optimisation (-O) and with optimisation (-O3) but with the same results. Can anyone help please. Best regards Phineus and Chris. PROGRAM CPMD STARTED AT: Thu May 23 10:39:21 2002 ****** ****** **** **** ****** ******* ******* ********** ******* *** ** *** ** **** ** ** *** ** ** *** ** ** ** ** ** ** ******* ** ** ** ** *** ****** ** ** ** *** ******* ** ** ** ******* ****** ** ** ** ****** VERSION 3.5.1 COPYRIGHT IBM RESEARCH DIVISION MPI FESTKOERPERFORSCHUNG STUTTGART The CPMD consortium WWW: http://www.cpmd.org Mailinglist: cpmd-list at cpmd.org E-mail: cpmd at cpmd.org *** May 22 2002 -- 15:34:13 *** THE INPUT FILE IS: h2otest.inp THIS JOB RUNS ON: birdie.EMBL-Heidelberg.DE THE CURRENT DIRECTORY IS: /nmr/sattler5/data/markwick/h2otest THE TEMPORARY DIRECTORY IS: /nmr/sattler5/data/markwick/h2otest THE PROCESS ID IS: 47710 THE JOB WAS SUBMITTED BY: markwick SINGLE POINT DENSITY OPTIMIZATION PATH TO THE RESTART FILES: ./ GRAM-SCHMIDT ORTHOGONALIZATION MAXIMUM NUMBER OF STEPS: 1000 STEPS PRINT INTERMEDIATE RESULTS EVERY 10001 STEPS STORE INTERMEDIATE RESULTS EVERY 10001 STEPS NUMBER OF DISTINCT RESTART FILES: 1 TEMPERATURE IS CALCULATED ASSUMING EXTENDED BULK BEHAVIOR FICTITIOUS ELECTRON MASS: 400.0000 TIME STEP: 4.0000 CONVERGENCE CRITERIA FOR WAVEFUNCTION OPTIMIZATION: 1.0000E-05 WAVEFUNCTION OPTIMIZATION BY PRECONDITIONED DIIS THRESHOLD FOR THE WF-HESSIAN IS 0.5000 MAXIMUM NUMBER OF VECTORS RETAINED FOR DIIS: 10 FULL ELECTRONIC GRADIENT IS USED SPLINE INTERPOLATION IN G-SPACE FOR PSEUDOPOTENTIAL FUNCTIONS NUMBER OF SPLINE POINTS: 501 ================================================================ = UNKNOWN KEYWORDS = = mOLECULAR dYNAMICS = = rESTART lATEST wAVEFUNCTION cOORDINATE = = S vELOCITIES rESTART nOSEP = = = = rESCALE oLD vELOCITIES = = iSOLATED mOLECULE = = pRINT oN wANNIER = = = = wANNIER wFNOUT aLL = ================================================================ EXCHANGE CORRELATION FUNCTIONALS LDA EXCHANGE: SLATER (ALPHA = 0.66667) LDA CORRELATION: LEE, YANG & PARR [C.L. LEE, W. YANG, AND R.G. PARR, PRB 37 785 (1988)] GRADIENT CORRECTED FUNCTIONAL DENSITY THRESHOLD: 1.00000E-07 EXCHANGE ENERGY [A.D. BECKE, PHYS. REV. A 38, 3098 (1988)] PARAMETER BETA: 0.004200 CORRELATION ENERGY [LYP: C.L. LEE ET AL. PHYS. REV. B 37, 785 (1988)] *** DETSP| THE NEW SIZE OF THE PROGRAM IS 24674 kBYTES *** ***************************** ATOMS **************************** NR TYPE X(bohr) Y(bohr) Z(bohr) MBL 1 H 4.070145 4.511445 6.611041 3 2 H 4.070145 4.511445 3.388959 3 3 O 5.000000 5.183334 5.000000 3 **************************************************************** NUMBER OF STATES: 4 NUMBER OF ELECTRONS: 8.00000 CHARGE: 0.00000 ELECTRON TEMPERATURE(KELVIN): 0.00000 OCCUPATION 2.0 2.0 2.0 2.0 ============================================================ | hydrogen pseudopotential with | | Exchange-Correlation Functional : | | Slater exchange : .6667 | | LDA correlation : Lee-Yang-Parr | | Exchange GC : Becke (1988) | | Correlation GC : Lee-Yang-Parr | | Car (Giannozzi) normconserving PP | | obtained from M. Odelius (<-Michiel Sprik) October 1995 | ============================================================ ============================================================ | oxygen pseudopotential with | | Exchange-Correlation Functional : | | Slater exchange : .6667 | | LDA correlation : Lee-Yang-Parr | | Exchange GC : Becke (1988) | | Correlation GC : Lee-Yang-Parr | | Trouiller-Martins normconserving PP | | obtained from M. Odelius (<-Michiel Sprik) October 1995 | ============================================================ **************************************************************** * ATOM MASS RAGGIO NLCC PSEUDOPOTENTIAL * * H 1.0080 1.2000 NO S LOCAL * * O 15.9994 1.2000 NO KLEINMAN S NONLOCAL * * P LOCAL * **************************************************************** *** RGGEN| THE NEW SIZE OF THE PROGRAM IS 26935 kBYTES *** ************************** SUPERCELL *************************** SYMMETRY: SIMPLE CUBIC LATTICE CONSTANT(a.u.): 10.00000 CELL DIMENSION: 10.0000 1.0000 1.0000 0.0000 0.0000 0.0000 VOLUME(OMEGA IN BOHR3): 1000.00000 LATTICE VECTOR A1(BOHR): 10.0000 0.0000 0.0000 LATTICE VECTOR A2(BOHR): 0.0000 10.0000 0.0000 LATTICE VECTOR A3(BOHR): 0.0000 0.0000 10.0000 RECIP. LAT. VEC. B1(2Pi/BOHR): 0.1000 0.0000 0.0000 RECIP. LAT. VEC. B2(2Pi/BOHR): 0.0000 0.1000 0.0000 RECIP. LAT. VEC. B3(2Pi/BOHR): 0.0000 0.0000 0.1000 REAL SPACE MESH: 54 54 54 WAVEFUNCTION CUTOFF(RYDBERG): 70.00000 DENSITY CUTOFF(RYDBERG): (DUAL= 4.00) 280.00000 NUMBER OF PLANE WAVES FOR WAVEFUNCTION CUTOFF: 4922 NUMBER OF PLANE WAVES FOR DENSITY CUTOFF: 39559 **************************************************************** *** RINFORCE| THE NEW SIZE OF THE PROGRAM IS 28868 kBYTES *** *** FFTPRP| THE NEW SIZE OF THE PROGRAM IS 34177 kBYTES *** *** NOSALLOC| THE NEW SIZE OF THE PROGRAM IS 34177 47710 Bus error From M.A.Zwijnenburg at tnw.tudelft.nl Mon May 27 14:05:04 2002 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Mon, 27 May 2002 14:05:04 +0200 Subject: [CPMD-list] solid state optimization Message-ID: <3CF23CF8.28258.66F4ABEC@localhost> Hi, If I use CPMD to optimize the geometry of a solid does it then only optimize the positions of the atoms in the cell or also the cell volume? If yes, should I then input the cell parameter in the &system part. Regards, Martijn Zwijnenburg ------------------------------------------------------------------------- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From hutter at pci.unizh.ch Mon May 27 15:13:56 2002 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Mon, 27 May 2002 15:13:56 +0200 (MEST) Subject: [CPMD-list] CPMD SGI-ORIGIN PROBLEM In-Reply-To: <3CECAE53.6040507@NMR.EMBL-Heidelberg.DE> Message-ID: Hi can you send us some information on the compiler and linker options you used? Maybe also the version of the operating system and compiler. I have a SGI Origin 200 and no problems running CPMD. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- From szefczyk at pkmk486.ch.pwr.wroc.pl Tue May 28 10:43:57 2002 From: szefczyk at pkmk486.ch.pwr.wroc.pl (Borys Szefczyk) Date: Tue, 28 May 2002 10:43:57 +0200 Subject: [CPMD-list] Simulations with cobalt Message-ID: <20020528084357.GA11228@pkmk486.ch.pwr.wroc.pl> Dear CPMD users, Currently, I am doing simulations for a system containing the cobalt surface. I try to optimize my structure but I have many problems with the wavefunction convergence. After excluding other possible reasons, I came to the idea, that the cobalt is responsible for my troubles. I have tried to optimize small hexagonal cell containing only two cobalt atoms and even this didn't work. I am using BP functional and Goedecker-type pseudopotential. Does anybody have any experience with cobalt? Is this element especially difficult? I am beginer in CPMD, therefore I don't exclude other technical errors in my jobs. Thanks for any advice, Borys Szefczyk -- Molecular Modeling Laboratory, Institute of Physical & Theoretical Chemistry Wroclaw University of Technology szefczyk at mml.ch.pwr.wroc.pl http://www.mml.ch.pwr.wroc.pl From M.A.Zwijnenburg at tnw.tudelft.nl Tue May 28 15:06:12 2002 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Tue, 28 May 2002 15:06:12 +0200 Subject: [CPMD-list] no RESTART file Message-ID: <3CF39CCB.15310.6C52FBE5@localhost> Hi, Yesterday I started an optimization job, which went fine, but created no RESTART nor LATEST file. This to my great suprise since all previous calculations always created RESTART, LATEST and HESSIAN files. The difference with previous calculations are that I now included STEEPEST DESCENT ELECTRONS IONS CELL and changed the symmetry to 14 rather then 1. Does above settings make a RESTART file impossible or do i have to include a specific keyword to generate a RESTART file. cheers, Martijn ------------------------------------------------------------------------- Martijn Zwijnenburg Lab. of Applied Organic Chemistry and Catalysis Delft University of Technology Julianalaan 136 2628 BL Delft The Netherlands Tel: 0031-(0)152782691 Fax: 0031-(0)152784700 e-mail: M.A.Zwijnenburg at tnw.tudelft.nl web page: http://come.to/tock From srb at zurich.ibm.com Tue May 28 16:11:41 2002 From: srb at zurich.ibm.com (srb at zurich.ibm.com) Date: Tue, 28 May 2002 16:11:41 +0200 Subject: [CPMD-list] Re: [CPMD-list] Message-ID: <200205281411.g4SEBfj02694@verscio.zurich.ibm.com> Hi, with your settings, CPMD should write a RESTART file when convergence or the maximum time/number of steps are reached, whichever comes first. By using steepest descent for ions and electrons at the same time, you are selecting another optimization scheme (the ionic positions are updated every wavefunction optimization cycle rather than only after convergence of the wavefunction). Did your job finish normally? You have three possibilities: 1. Use the very efficient DIIS scheme for electron optimization instead of steepest descent. This will most likely save you many cycles but can fail in very rare cases. 2. Keep combined steepest descent scheme and specify STORE 20 (or any other frequency) to write the RESTART file on the fly (in case your job does not finish normally) 3. Use steepest descent for all degrees of freedom but return to the conventional scheme which writes the RESTART file each time the wavefunction is converged. In rgmopt.F, you have to change the line IF(TSDP .AND. TSDE) THEN into IF(.FALSE.) THEN and recompile the code. I'd try option 1 first. Best wishes, Salomon From hwsheng at jhu.edu Tue May 28 21:37:33 2002 From: hwsheng at jhu.edu (HW Sheng) Date: Tue, 28 May 2002 15:37:33 -0400 Subject: [CPMD-list] pseudopotential generation problem References: <200205281411.g4SEBfj02694@verscio.zurich.ibm.com> Message-ID: <003401c2067f$1d00cc40$a567dc80@mrs.jhu.edu> Dear all, I was trying to generate a Hartwigsen-Goedeker-Hutter type of pseudopotential for lanthanum using the code included in /Godecker/atom and /Godecker/pseudo. The all-electron calculations have been successfully made with provided electron configurations. However, I encounter the following problem when involving /Goedecker/pseudo/pseudo: "zion" from atom.ae and psp.par are not identical. I checked the source code of atom.f. I found in line 265, atom.f, the variable zion (valence electrons) has not been predefined, which invariably gives naught in psp.par as output. Does that mean I have to modify "zion" in psp.par manually before proceeding to pseudo? Can anyone tell me how to get around this problem? Thanks in advance! Howard Sheng Dept. Mater. Sci. Eng. Johns Hopkins Univ. Baltimore, MD 21218 From mqtan at zjuem.zju.edu.cn Mon May 13 06:20:11 2002 From: mqtan at zjuem.zju.edu.cn (Prof. Dr. Ming-Qiu Tan) Date: Mon, 13 May 2002 12:20:11 +0800 Subject: [CPMD-list] Crash under IRIX 6.5 References: <200205101001.MAA14344@cpmd.org> Message-ID: <001b01c1fa35$79670590$c77a0e0a@condmatter> Dear all: I have found the crash reason for the CPMD under SGI IRIX 6.5. Please devide the BYLM_p.F into several small subroutines. It seems that this subroutine is too large to accomodate some compilies, e.g., Intel IFC and MipsPro. Anyone who interests on such a modified BYLM_p.F could contact me. Ming-Qiu Tan ------------------------------ Department of Physics Zhejiang University Hangzhou 310027 P. R. China Phone: +86-571-87951594 Fax: +86-571-87951328 ------------------------------- From markwick at NMR.EMBL-Heidelberg.DE Thu May 23 10:20:42 2002 From: markwick at NMR.EMBL-Heidelberg.DE (Phineus Markwick) Date: Thu, 23 May 2002 10:20:42 +0200 Subject: [CPMD-list] CPMD SGI-ORIGIN PROBLEM Message-ID: <3CECA65A.7010004@NMR.EMBL-Heidelberg.DE> Hallo CPMD mail list! We are having problems with CPMD on our SGI Origin 2000 (28 Proccessors, 6GB real memory). The program does compile, but CPMD aborts with a Bus error part-way through. In a simple test on H2O doing a wavefunction optimisation we get the output file below. We have tried compiling with no optimisation (-O) and with optimisation (-O3) but with the same results. Can anyone help please. Best regards Phineus and Chris. PROGRAM CPMD STARTED AT: Thu May 23 10:39:21 2002 ****** ****** **** **** ****** ******* ******* ********** ******* *** ** *** ** **** ** ** *** ** ** *** ** ** ** ** ** ** ******* ** ** ** ** *** ****** ** ** ** *** ******* ** ** ** ******* ****** ** ** ** ****** VERSION 3.5.1 COPYRIGHT IBM RESEARCH DIVISION MPI FESTKOERPERFORSCHUNG STUTTGART The CPMD consortium WWW: http://www.cpmd.org Mailinglist: cpmd-list at cpmd.org E-mail: cpmd at cpmd.org *** May 22 2002 -- 15:34:13 *** THE INPUT FILE IS: h2otest.inp THIS JOB RUNS ON: birdie.EMBL-Heidelberg.DE THE CURRENT DIRECTORY IS: /nmr/sattler5/data/markwick/h2otest THE TEMPORARY DIRECTORY IS: /nmr/sattler5/data/markwick/h2otest THE PROCESS ID IS: 47710 THE JOB WAS SUBMITTED BY: markwick SINGLE POINT DENSITY OPTIMIZATION PATH TO THE RESTART FILES: ./ GRAM-SCHMIDT ORTHOGONALIZATION MAXIMUM NUMBER OF STEPS: 1000 STEPS PRINT INTERMEDIATE RESULTS EVERY 10001 STEPS STORE INTERMEDIATE RESULTS EVERY 10001 STEPS NUMBER OF DISTINCT RESTART FILES: 1 TEMPERATURE IS CALCULATED ASSUMING EXTENDED BULK BEHAVIOR FICTITIOUS ELECTRON MASS: 400.0000 TIME STEP: 4.0000 CONVERGENCE CRITERIA FOR WAVEFUNCTION OPTIMIZATION: 1.0000E-05 WAVEFUNCTION OPTIMIZATION BY PRECONDITIONED DIIS THRESHOLD FOR THE WF-HESSIAN IS 0.5000 MAXIMUM NUMBER OF VECTORS RETAINED FOR DIIS: 10 FULL ELECTRONIC GRADIENT IS USED SPLINE INTERPOLATION IN G-SPACE FOR PSEUDOPOTENTIAL FUNCTIONS NUMBER OF SPLINE POINTS: 501 ================================================================ = UNKNOWN KEYWORDS = = mOLECULAR dYNAMICS = = rESTART lATEST wAVEFUNCTION cOORDINATE = = S vELOCITIES rESTART nOSEP = = = = rESCALE oLD vELOCITIES = = iSOLATED mOLECULE = = pRINT oN wANNIER = = = = wANNIER wFNOUT aLL = ================================================================ EXCHANGE CORRELATION FUNCTIONALS LDA EXCHANGE: SLATER (ALPHA = 0.66667) LDA CORRELATION: LEE, YANG & PARR [C.L. LEE, W. YANG, AND R.G. PARR, PRB 37 785 (1988)] GRADIENT CORRECTED FUNCTIONAL DENSITY THRESHOLD: 1.00000E-07 EXCHANGE ENERGY [A.D. BECKE, PHYS. REV. A 38, 3098 (1988)] PARAMETER BETA: 0.004200 CORRELATION ENERGY [LYP: C.L. LEE ET AL. PHYS. REV. B 37, 785 (1988)] *** DETSP| THE NEW SIZE OF THE PROGRAM IS 24674 kBYTES *** ***************************** ATOMS **************************** NR TYPE X(bohr) Y(bohr) Z(bohr) MBL 1 H 4.070145 4.511445 6.611041 3 2 H 4.070145 4.511445 3.388959 3 3 O 5.000000 5.183334 5.000000 3 **************************************************************** NUMBER OF STATES: 4 NUMBER OF ELECTRONS: 8.00000 CHARGE: 0.00000 ELECTRON TEMPERATURE(KELVIN): 0.00000 OCCUPATION 2.0 2.0 2.0 2.0 ============================================================ | hydrogen pseudopotential with | | Exchange-Correlation Functional : | | Slater exchange : .6667 | | LDA correlation : Lee-Yang-Parr | | Exchange GC : Becke (1988) | | Correlation GC : Lee-Yang-Parr | | Car (Giannozzi) normconserving PP | | obtained from M. Odelius (<-Michiel Sprik) October 1995 | ============================================================ ============================================================ | oxygen pseudopotential with | | Exchange-Correlation Functional : | | Slater exchange : .6667 | | LDA correlation : Lee-Yang-Parr | | Exchange GC : Becke (1988) | | Correlation GC : Lee-Yang-Parr | | Trouiller-Martins normconserving PP | | obtained from M. Odelius (<-Michiel Sprik) October 1995 | ============================================================ **************************************************************** * ATOM MASS RAGGIO NLCC PSEUDOPOTENTIAL * * H 1.0080 1.2000 NO S LOCAL * * O 15.9994 1.2000 NO KLEINMAN S NONLOCAL * * P LOCAL * **************************************************************** *** RGGEN| THE NEW SIZE OF THE PROGRAM IS 26935 kBYTES *** ************************** SUPERCELL *************************** SYMMETRY: SIMPLE CUBIC LATTICE CONSTANT(a.u.): 10.00000 CELL DIMENSION: 10.0000 1.0000 1.0000 0.0000 0.0000 0.0000 VOLUME(OMEGA IN BOHR^3): 1000.00000 LATTICE VECTOR A1(BOHR): 10.0000 0.0000 0.0000 LATTICE VECTOR A2(BOHR): 0.0000 10.0000 0.0000 LATTICE VECTOR A3(BOHR): 0.0000 0.0000 10.0000 RECIP. LAT. VEC. B1(2Pi/BOHR): 0.1000 0.0000 0.0000 RECIP. LAT. VEC. B2(2Pi/BOHR): 0.0000 0.1000 0.0000 RECIP. LAT. VEC. B3(2Pi/BOHR): 0.0000 0.0000 0.1000 REAL SPACE MESH: 54 54 54 WAVEFUNCTION CUTOFF(RYDBERG): 70.00000 DENSITY CUTOFF(RYDBERG): (DUAL= 4.00) 280.00000 NUMBER OF PLANE WAVES FOR WAVEFUNCTION CUTOFF: 4922 NUMBER OF PLANE WAVES FOR DENSITY CUTOFF: 39559 **************************************************************** *** RINFORCE| THE NEW SIZE OF THE PROGRAM IS 28868 kBYTES *** *** FFTPRP| THE NEW SIZE OF THE PROGRAM IS 34177 kBYTES *** *** NOSALLOC| THE NEW SIZE OF THE PROGRAM IS 34177 47710 Bus error From weiz at mail.rochester.edu Fri May 24 22:58:01 2002 From: weiz at mail.rochester.edu (wei zhuang) Date: Fri, 24 May 2002 15:58:1 -0500 Subject: [CPMD-list] help Message-ID: <200205241957.g4OJvMEX18467896@mail.rochester.edu> Hi, all: I am running MD for a 16 glycerol system. but after several steps, it always said that op ----------------------- RORTHO:DIFGAM= nan ITER = 31 > 30 MAXIMUM NUMBER OF ITERATIONS EXCEEDED PROGRAM STOPS IN SUBROUTINE RORTOG| [PROC= 0] ----------------------- and then stoped. could anybody give me some hint about what happen? wei zhuang From ZRLSRB at ch.ibm.com Tue May 28 16:06:51 2002 From: ZRLSRB at ch.ibm.com (Salomon Billeter) Date: Tue, 28 May 2002 10:06:51 -0400 Subject: [CPMD-list] no RESTART file Message-ID: Hi, with your settings, CPMD should write a RESTART file when convergence or the maximum time/number of steps are reached, whichever comes first. By using steepest descent for ions and electrons at the same time, you are selecting another optimization scheme (the ionic positions are updated every wavefunction optimization cycle rather than only after convergence of the wavefunction). Did your job finish normally? You have three possibilities: 1. Use the very efficient DIIS scheme for electron optimization instead of steepest descent. This will most likely save you many cycles but can fail in very rare cases. 2. Keep combined steepest descent scheme and specify STORE 20 (or any other frequency) to write the RESTART file on the fly (in case your job does not finish normally) 3. Use steepest descent for all degrees of freedom but return to the conventional scheme which writes the RESTART file each time the wavefunction is converged. In rgmopt.F, you have to change the line IF(TSDP .AND. TSDE) THEN into IF(.FALSE.) THEN and recompile the code. I'd try option 1 first. Best wishes, Salomon