From mkosmows at mailbox.syr.edu Sun Jun 1 09:05:33 2003 From: mkosmows at mailbox.syr.edu (mkosmows) Date: Sun, 1 Jun 2003 03:05:33 -0400 Subject: [CPMD-list] k points Message-ID: <3EDC5033@OrangeMail> Dear CPMD community: When no KPOINTS keyword is specified, what does CPMD do? Also, can anyone recommend any sources to learn more about k points? I have the Monkhorst & Pack paper from the early 1970's as well as the Chadi et al. paper referenced therein and I am looking for a better understanding of the significance of k points in a calculation. Thank you, Mark Kosmowski Chemistry Department Syracuse University mkosmows at syr.edu From aijun_du at hotmail.com Tue Jun 3 14:25:29 2003 From: aijun_du at hotmail.com (Du Aijun) Date: Tue, 03 Jun 2003 20:25:29 +0800 Subject: [CPMD-list] Problem with population analysis! Message-ID: Dear all, I met the problem on population analysis. LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985200 PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 2] LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985320 PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 1] *** PHFAC| THE NEW SIZE OF THE PROGRAM IS $ This problem appeared in the CPMD-list before, but it was not resolved at that time. Here I want to know whether there is a solution to it now! Thanks in advance! With regards, Aijun Du _________________________________________________________________ ???????? MSN Explorer: http://explorer.msn.com/lccn/ From Ari.P.Seitsonen at iki.fi Tue Jun 3 14:53:45 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Tue, 3 Jun 2003 14:53:45 +0200 Subject: [CPMD-list] k points In-Reply-To: <3EDC5033@OrangeMail> (message from mkosmows on Sun, 1 Jun 2003 03:05:33 -0400) References: <3EDC5033@OrangeMail> Message-ID: <200306031253.h53Crj110005@magadino.cscs.ch> Dear Mark, > When no KPOINTS keyword is specified, what does CPMD do? Only the Gamma point (= 0,0,0) is used. This allows simplifications in the numerics (the wave functions can be chosen to be real in the real space, saving half of the wave function coefficient in the reciprocal space since c(-G) = c(G)^*, and two FFT's on the orbitals can be performed simultaneously etc). > Also, can anyone recommend any sources to learn more about k points? I > have the Monkhorst & Pack paper from the early 1970's as well as the Chadi et > al. paper referenced therein and I am looking for a better understanding of > the > significance of k points in a calculation. Well, the very basics are given in the standard textbooks on solid state physics, e.g. Ashcroft & Mermin. Besides that, of course the original papers which you mentioned. Otherwise I recommend some testing and playing around with the meshes, it's the best to have some own experience from many kind of systems (semi-conducting, metallic, 3D, 2D, ...) on how to choose the appropriate mesh of k points. Greetings from Zurigo sereno, apsi PS If you are interested in the specific case of two dimensional k point sets in the hexagonal cell there has been some discussion in the literature, in my thesis I generalised the Cunningham sets to include many more sets, plus compared them to the Monkhorst-Pack meshes: http://edocs.tu-berlin.de/diss/2000/seitsonen_ari.pdf -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From Ari.P.Seitsonen at iki.fi Tue Jun 3 19:04:59 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Tue, 3 Jun 2003 19:04:59 +0200 Subject: [CPMD-list] Problem with population analysis! In-Reply-To: (aijun_du@hotmail.com) References: Message-ID: <200306031704.h53H4xI11232@magadino.cscs.ch> Dear Aijun, > I met the problem on population analysis. > > LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985200 > > PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 2] > LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985320 > > PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 1] > *** PHFAC| THE NEW SIZE OF THE PROGRAM IS $ > > This problem appeared in the CPMD-list before, but it was not resolved at > that time. Here I want to know whether there is a solution to it now! Do you possibly use Vanderbilt pseudos for some of the atoms? That might not be implemented. Could you please send your input file so that we can inspect if we see something suspicious. Greetings, apsi -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From aijun_du at hotmail.com Tue Jun 3 19:12:22 2003 From: aijun_du at hotmail.com (Du Aijun) Date: Wed, 04 Jun 2003 01:12:22 +0800 Subject: [CPMD-list] Problem with population analysis! Message-ID: Dear Dr. Ari.P.Seitsonen, Thanks for your reply! I used the Vanderbilt pseudos. Now I have changed the VDB to MT psedudos. The population analysis is Okay! With regards, Aijun Du >From: Ari.P.Seitsonen at iki.fi >Reply-To: cpmd-list at cpmd.org >To: aijun_du at hotmail.com >CC: cpmd-list at cpmd.org >Subject: Re: [CPMD-list] Problem with population analysis! >Date: Tue, 3 Jun 2003 19:04:59 +0200 > >Dear Aijun, > > > I met the problem on population analysis. > > > > LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985200 > > > > PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 2] > > LSCR= 100 MAX(LRNLSM1,LRNLSM2) = 985320 > > > > PROGRAM STOPS IN SUBROUTINE RNLSM| SCRATCH ARRAY TOO SMALL [PROC= 1] > > *** PHFAC| THE NEW SIZE OF THE PROGRAM IS $ > > > > This problem appeared in the CPMD-list before, but it was not resolved at > > that time. Here I want to know whether there is a solution to it now! > >Do you possibly use Vanderbilt pseudos for some of the atoms? That >might not be implemented. Could you please send your input file so >that we can inspect if we see something suspicious. > > Greetings, > > apsi > >-- >-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- > Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ > Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 > Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) > Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich > Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland _________________________________________________________________ ?????????????????????????????? MSN Hotmail?? http://www.hotmail.com From veronique.vanspeybroeck at ugent.be Fri Jun 6 10:03:32 2003 From: veronique.vanspeybroeck at ugent.be (Veronique Van Speybroeck) Date: Fri, 06 Jun 2003 10:03:32 +0200 Subject: [CPMD-list] Re: MEMORY ALLOCATION References: <3EDFCC60.7090403@ipicyt.edu.mx> <3EE03B6B.3010700@ugent.be> Message-ID: <3EE04AD4.9050903@ugent.be> > Dear Jose, > > I did not receive any answer to my previous message. It is a rather > strange problems, depending on the machine (all linux machines ) I get > the error message or not. In any case I can allocate more when using > the parallel version of the cpmd code. But still in some cases this > is not sufficient. I do not know how to resolve the problem. > In the mean time I have started the calculations with a smaller cutoff > (20 Ry and not 25 Ry , for VDB pseudopotentials). Later on I will > have to do a follow up calculation to get the cutoff right. I am > planning to try that on an SGI workstation and hope this will work > > best regards > > veronique > > > JOSE LUIS RODRIGUEZ-LOPEZ wrote: > >> Dear Veronique, >> What did you do for to fix the reported problem like this: >> >> >> ---------------------------------------------------------------- >> [PEAK NUMBER 57] PEAK MEMORY 10592097 = 84.7 MBytes >> ================================================================ >> >> >> PROGRAM STOPS IN SUBROUTINE MEMORY| ALLOCATION FAILED (GDE) [PROC= 0] >> >> **************************************************************** >> PROCESSOR 64 ALLOCATION OF 1670740 WORDS OF MEMORY FAILED >> **************************************************************** >> >> *** MEMORY| CURRENT HEAP LENGTH 10805 KWords *** >> >> ================================================================ >> BIG MEMORY ALLOCATIONS >> PME 6682960 C2 1336584 >> C0 1336584 SCR 1185409 >> ATWFR 301200 HGPOT 223494 >> WORK 100400 INYH 84730 >> RHOPS 70656 GK 70656 >> ---------------------------------------------------------------- >> [PEAK NUMBER 59] PEAK MEMORY 10645298 = 85.2 MBytes >> ================================================================ >> >> >> >> Do you have the answer received?? >> Have any idea for help? Thanks >> >> Greeting from Mexico, >> JoseLuis >> > -- ----------------------------------------------------------------------- Dr. ir. Van Speybroeck Veronique Laboratorium voor Theoretische Fysica Universiteit Gent Proeftuinstraat 86 9000 Gent Tel +32-9-264.65.58 GSM : +32/474/259767 Fax +32-9-264.65.60 email : veronique.vanspeybroeck at ugent.be http://inwfaxp2.ugent.be /~web/onderzoek/proj_3.html ----------------------------------------------------------------------- From adrainzhou at yahoo.com.cn Fri Jun 6 10:15:45 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Fri, 6 Jun 2003 16:15:45 +0800 (CST) Subject: [CPMD-list] Geometry optimization! Message-ID: <20030606081545.55715.qmail@web15304.mail.bjs.yahoo.com> --------------------------------- Do You Yahoo!? "?????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030606/87c57d26/attachment.html From adrainzhou at yahoo.com.cn Fri Jun 6 10:27:53 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Fri, 6 Jun 2003 16:27:53 +0800 (CST) Subject: [CPMD-list] Geometry optimization! Message-ID: <20030606082753.3901.qmail@web15307.mail.bjs.yahoo.com> Dear all, I optimize a geometry structure with VDB first and it reach convergence. But when starting from this geometry, I optimize it with TM potential, even the wavefunction never get convergene! The following is the main part of input and output file: &CPMD OPTIMIZE GEOMETRY XYZ rESTART COORDINATES WAVEFUNCTION LATEST HESSIAN UNIT CONVERGENCE GEOMETRY 1.D-4 LBFGS CONVERGENCE ADAPT 0.02 CONVERGENCE ENERGY 0.05 CONVERGENCE RELAX 5 MAXSTEP 5000 &END &SYSTEM ANGSTROM ....... ....... CUTOFF 35.0 CHARGE -1 ODIIS| Insufficient progress; reset! 24 6.046E-04 4.552E-05 -293.831117 -2.213E-03 54.45 25 5.961E-04 4.807E-05 -293.833460 -2.343E-03 41.65 RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 32.0 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 28 5.769E-04 5.110E-05 -293.835152 8.914E-03 51.56 29 5.814E-04 4.859E-05 -293.830313 4.839E-03 53.30 30 5.882E-04 4.593E-05 -293.828870 1.443E-03 57.19 RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 64.0 I also try a high energy cutoff and PCG, but it is in vain. Can anyone tell me what's the wrong with it? Many thanks in advance! With regards, Zhou --------------------------------- Do You Yahoo!? "?????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030606/b1024fd0/attachment.html From adrainzhou at yahoo.com.cn Fri Jun 6 10:28:00 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Fri, 6 Jun 2003 16:28:00 +0800 (CST) Subject: [CPMD-list] Geometry optimization! Message-ID: <20030606082800.47675.qmail@web15301.mail.bjs.yahoo.com> Dear all, I optimize a geometry structure with VDB first and it reach convergence. But when starting from this geometry, I optimize it with TM potential, even the wavefunction never get convergene! The following is the main part of input and output file: &CPMD OPTIMIZE GEOMETRY XYZ rESTART COORDINATES WAVEFUNCTION LATEST HESSIAN UNIT CONVERGENCE GEOMETRY 1.D-4 LBFGS CONVERGENCE ADAPT 0.02 CONVERGENCE ENERGY 0.05 CONVERGENCE RELAX 5 MAXSTEP 5000 &END &SYSTEM ANGSTROM ....... ....... CUTOFF 35.0 CHARGE -1 ODIIS| Insufficient progress; reset! 24 6.046E-04 4.552E-05 -293.831117 -2.213E-03 54.45 25 5.961E-04 4.807E-05 -293.833460 -2.343E-03 41.65 RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 32.0 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 28 5.769E-04 5.110E-05 -293.835152 8.914E-03 51.56 29 5.814E-04 4.859E-05 -293.830313 4.839E-03 53.30 30 5.882E-04 4.593E-05 -293.828870 1.443E-03 57.19 RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 64.0 I also try a high energy cutoff and PCG, but it is in vain. Can anyone tell me what's the wrong with it? Many thanks in advance! With regards, Zhou --------------------------------- Do You Yahoo!? "?????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030606/379d9138/attachment.html From s.meloni at caspur.it Fri Jun 6 19:43:30 2003 From: s.meloni at caspur.it (Simone Meloni) Date: Fri, 6 Jun 2003 17:43:30 +0000 (UTC) Subject: [CPMD-list] intel mkl & cpmd Message-ID: Dear CPMDer, recently INTEL released the new version of their mathematic: MKL 6.0. In this version the simbol izamax_ has changed size and now in incompatible with the izamax_ in util.F. If you want to compile using the new library (I have not yet run performance test jobs ;-) needs to modify the conditional compilation about the function IZAMAX. It need to be changed line 727 of util.F from #if !(defined(__SGI)) && !(defined(__HP)) to #if !(defined(__SGI)) && !(defined(__HP)) && !(defined(__MKL60)) and define the varriable __MKL60 in the make file (CPPFLAGS += -D__MKL60) Bye Simone -- +------------------------------------------------------------------------+ PhD Simone Meloni | Achab: Lascio una scia Bianca e inquieta, CASPUR | acque pallide, facce piu' pallide Via Dei Tizii 6b | dovunque passo. Le onde invidiose 00185 Rome (Italy) | si gonfiano ai lati per sommergere Phone: +39-06-44486-711 | la mia traccia: facciano, ma prima Fax: +39-06-4957083 | io passo. From Ari.P.Seitsonen at iki.fi Tue Jun 10 13:48:46 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Tue, 10 Jun 2003 13:48:46 +0200 Subject: [CPMD-list] Re: MEMORY ALLOCATION In-Reply-To: <3EE04AD4.9050903@ugent.be> (message from Veronique Van Speybroeck on Fri, 06 Jun 2003 10:03:32 +0200) References: <3EDFCC60.7090403@ipicyt.edu.mx> <3EE03B6B.3010700@ugent.be> <3EE04AD4.9050903@ugent.be> Message-ID: <200306101148.h5ABmkR29981@magadino.cscs.ch> Dear Veronique, Could this be a problem of the limits (memory/stack/data) in the shell? Otherwise, could you please send us (= the list) your input file for CPMD? Greetings from sultry Zurich, apsi > > Dear Jose, > > > > I did not receive any answer to my previous message. It is a rather > > strange problems, depending on the machine (all linux machines ) I get > > the error message or not. In any case I can allocate more when using > > the parallel version of the cpmd code. But still in some cases this > > is not sufficient. I do not know how to resolve the problem. > > In the mean time I have started the calculations with a smaller cutoff > > (20 Ry and not 25 Ry , for VDB pseudopotentials). Later on I will > > have to do a follow up calculation to get the cutoff right. I am > > planning to try that on an SGI workstation and hope this will work > > > > best regards > > > > veronique > > > > > > JOSE LUIS RODRIGUEZ-LOPEZ wrote: > > > >> Dear Veronique, > >> What did you do for to fix the reported problem like this: > >> > >> > >> ---------------------------------------------------------------- > >> [PEAK NUMBER 57] PEAK MEMORY 10592097 = 84.7 MBytes > >> ================================================================ > >> > >> > >> PROGRAM STOPS IN SUBROUTINE MEMORY| ALLOCATION FAILED (GDE) [PROC= 0] > >> > >> **************************************************************** > >> PROCESSOR 64 ALLOCATION OF 1670740 WORDS OF MEMORY FAILED > >> **************************************************************** > >> > >> *** MEMORY| CURRENT HEAP LENGTH 10805 KWords *** > >> > >> ================================================================ > >> BIG MEMORY ALLOCATIONS > >> PME 6682960 C2 1336584 > >> C0 1336584 SCR 1185409 > >> ATWFR 301200 HGPOT 223494 > >> WORK 100400 INYH 84730 > >> RHOPS 70656 GK 70656 > >> ---------------------------------------------------------------- > >> [PEAK NUMBER 59] PEAK MEMORY 10645298 = 85.2 MBytes > >> ================================================================ > >> > >> > >> > >> Do you have the answer received?? > >> Have any idea for help? Thanks > >> > >> Greeting from Mexico, > >> JoseLuis > > > > -- > ----------------------------------------------------------------------- > Dr. ir. Van Speybroeck Veronique > Laboratorium voor Theoretische Fysica > Universiteit Gent > Proeftuinstraat 86 > 9000 Gent > Tel +32-9-264.65.58 GSM : +32/474/259767 > Fax +32-9-264.65.60 > email : veronique.vanspeybroeck at ugent.be > http://inwfaxp2.ugent.be /~web/onderzoek/proj_3.html > ----------------------------------------------------------------------- > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From Ari.P.Seitsonen at iki.fi Tue Jun 10 13:56:29 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Tue, 10 Jun 2003 13:56:29 +0200 Subject: [CPMD-list] Geometry optimization! In-Reply-To: <20030606082753.3901.qmail@web15307.mail.bjs.yahoo.com> (message from =?gb2312?q?Adrain=20Zhou?= on Fri, 6 Jun 2003 16:27:53 +0800 (CST)) References: <20030606082753.3901.qmail@web15307.mail.bjs.yahoo.com> Message-ID: <200306101156.h5ABuTE30024@magadino.cscs.ch> Dear Zhou, So you start from scratch and don't try to do a restart from the Vanderbilt calculation, as I see from the input? That's the right strategy. Did you try 'PCG MINIMIZE' or just 'PCG'? Does your system contain some "difficult" materials, namely elements for which the Troullier-Martins pseudo potential would not work properly (so that the electronic gap is closed)? Did you check that the total energy makes sense (i.e. that it should be around -293.x Hartree), and did you check that your coordinates are in the right units (Bohr from the output in CPMD, you request having them in Angstroms)? With the charge -1, does the system have a closed shell structure? Just some thoughts, I don't have any other ideas for the time being... Greetings from sunny Zurich, apsi > I optimize a geometry structure with VDB first and it reach > convergence. But when starting from this geometry, I optimize it > with TM potential, even the wavefunction never get convergene! > > The following is the main part of input and output file: > &CPMD > OPTIMIZE GEOMETRY XYZ > rESTART COORDINATES WAVEFUNCTION LATEST > HESSIAN UNIT > CONVERGENCE GEOMETRY > 1.D-4 > LBFGS > CONVERGENCE ADAPT > 0.02 > CONVERGENCE ENERGY > 0.05 > CONVERGENCE RELAX > 5 > MAXSTEP > 5000 > &END > &SYSTEM > ANGSTROM > ....... > ....... > CUTOFF > 35.0 > CHARGE > -1 > > ODIIS| Insufficient progress; reset! > 24 6.046E-04 4.552E-05 -293.831117 -2.213E-03 54.45 > 25 5.961E-04 4.807E-05 -293.833460 -2.343E-03 41.65 > RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 32.0 > 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 > 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 > 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 > 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 > 28 5.769E-04 5.110E-05 -293.835152 8.914E-03 51.56 > 29 5.814E-04 4.859E-05 -293.830313 4.839E-03 53.30 > 30 5.882E-04 4.593E-05 -293.828870 1.443E-03 57.19 > RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 64.0 > > I also try a high energy cutoff and PCG, but it is in vain. Can > anyone tell me what's the wrong with it? > > Many thanks in advance! > > With regards, > Zhou > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From WANGYI at engr.sc.edu Wed Jun 11 19:18:22 2003 From: WANGYI at engr.sc.edu (WANG, YIXUAN ) Date: Wed, 11 Jun 2003 13:18:22 -0400 Subject: [CPMD-list] TM pseudopotential of Pt Message-ID: <99296F29E0A01A40A0503EB30507ECF91E9A2D@mail1.engr.sc.edu> Dear Colleagues, A lot of publications are using TM PP for Pt, most probably generated by fhi98PP code, however I failed to find a paper that deals with the parameters employed for PP generation and/or compares the results with different PPs. We tested two types of TM PPs for molecular adsorptions on Pt surface. One is from the input (iexc=8, attached below for discussion) in fhi98PP, and DFT is shifted from LDA to BLYP ( iexc=9) for the other one. They give clear different results. In principle I should believe the latter one. My question is that anybody has used different parameters than the original one, especially (rnlc, rct and et), and how the effect is. I appreciate your comments. Regards, Yixuan 78.00 12 2 8 0.00 : z nc nv iexc rnlc 1 0 2.00 : n l f 2 0 2.00 2 1 6.00 3 0 2.00 3 1 6.00 3 2 10.00 4 0 2.00 4 1 6.00 4 2 10.00 4 3 14.00 5 0 2.00 5 1 6.00 5 2 9.00 6 0 1.00 2 t : lmax s_pp_def 1 0.00 10.00 t : lt rct et s_pp_type From tdeutsch at cea.fr Thu Jun 12 10:13:13 2003 From: tdeutsch at cea.fr (Thierry Deutsch) Date: Thu, 12 Jun 2003 10:13:13 +0200 Subject: [CPMD-list] Difference between IBM-SP3-SMP and IBM-SP3 Message-ID: <200306121013.13629.tdeutsch@cea.fr> Hello, In the Configure file, there is two options for IBM-SP3: IBM-SP3-SMP which seems to use a SMP library for essl IBM-SP3 which uses a serial library for essl. I would like to have some recommendations from our lovely staff from IBM Zurich about using SMP or not in a IBM-SP3. Thanks in advance. 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 shaposh at isp.nsc.ru Fri Jun 13 08:22:48 2003 From: shaposh at isp.nsc.ru (shaposh) Date: Fri, 13 Jun 2003 13:22:48 +0700 Subject: [CPMD-list] Problem compiling CPMD - PC+IFC7+MKL6 Message-ID: <000201c33174$365f2bc0$5ab2e2c2@quantum> Dear All! I have a P4 PC with IFC 7.1 and Intel MKL on RedHat9. Trying to compile CPMD 3.7.1, i'v got an error. My configuration in Makefile is as following: #--------------- Default Configuration for PC-IFC --------------- SRC = . DEST = . BIN = . #QMMM_FLAGS = -D__QMECHCOUPL #QMMM_LIBS = -L. -lmm FFLAGS = -c -r8 -w90 -w95 -O2 -pc64 -axM -ip -tpp7 CPPFLAGS = -D__MKL60 -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC LFLAGS = -L$(MKLPATH) -lmkl_lapack -lmkl_p4 -lguide -lpthread -Vaxlib $(QMMM_LIBS) CFLAGS = -c -O2 -Wall CPP = /lib/cpp -P -C -traditional NOOPT_FLAG = CC = gcc FC = ifc LD = ifc AR = #----------------------------------------------------------------------- ----- The error is: opt/intel/compiler70/ia32/lib/libIEPCF90.a(f90fioerr.o)(.text+0x4d3): In function `f_f77ioerr': : undefined reference to `__ctype_b' I will very appreciate any help with this. By the way, it seems to me that the env. variable LD_LIBRARY_PATH is not working in my system as it has no effect in assigning any value to it (and exporting it).What can be the source of the problem? Thank you in advance. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030613/6359d0d8/attachment.html From adrainzhou at yahoo.com.cn Fri Jun 13 18:19:25 2003 From: adrainzhou at yahoo.com.cn (=?gb2312?q?Adrain=20Zhou?=) Date: Sat, 14 Jun 2003 00:19:25 +0800 (CST) Subject: [CPMD-list] Geometry optimization! In-Reply-To: <200306101156.h5ABuTE30024@magadino.cscs.ch> Message-ID: <20030613161925.80998.qmail@web15301.mail.bjs.yahoo.com> Dear Dr. Ari.P.Seitsonen, Yes, I just start the geometry from the Vanderbilit calculation. This system involves Carbon, Silicon and Hydrogen atomic species. With Charge -1, this system contains 153 states with 306 electrons. I also try PCG minimize, the output is as follows, NFI GEMAX CNORM ETOT DETOT TCPU 1 2.669E-02 1.693E-03 -290.095240 0.000E+00 38.97 LINE SEARCH : LAMBDA=.664 ENERGY = -293.42791 2 2.346E-02 5.299E-04 -293.028027 -2.933E+00 39.34 LINE SEARCH : LAMBDA=1.65 ENERGY = -293.61716 3 7.462E-03 4.567E-04 -293.428667 -4.006E-01 39.25 LINE SEARCH : LAMBDA=.664 ENERGY = -293.65894 4 7.290E-03 2.232E-04 -293.617055 -1.884E-01 38.53 LINE SEARCH : LAMBDA=.864 ENERGY = -293.68680 5 4.220E-03 1.464E-04 -293.658951 -4.190E-02 54.67 LINE SEARCH : LAMBDA=.871 ENERGY = -293.70538 6 3.500E-03 1.169E-04 -293.686883 -2.793E-02 68.70 LINE SEARCH : LAMBDA=1.40 ENERGY = -293.72920 7 3.382E-03 1.276E-04 -293.705251 -1.837E-02 60.05 LINE SEARCH : LAMBDA=1.41 ENERGY = -293.76193 8 3.662E-03 1.672E-04 -293.729179 -2.393E-02 57.30 LINE SEARCH : LAMBDA=1.15 ENERGY = -293.79379 9 5.450E-03 1.570E-04 -293.761992 -3.281E-02 63.50 LINE SEARCH : LAMBDA=.896 ENERGY = -293.82154 10 4.128E-03 1.463E-04 -293.794566 -3.257E-02 68.07 LINE SEARCH : LAMBDA=2.75 ENERGY = -293.87424 11 4.693E-03 1.491E-04 -293.822917 -2.835E-02 45.30 LINE SEARCH : LAMBDA=6.24 ENERGY = -294.33177 12 8.708E-03 2.020E-04 -293.873630 -5.071E-02 39.56 LINE SEARCH : LAMBDA=3.14 ENERGY = -214.09080 13 2.425E-02 9.565E-04 -294.397957 -5.243E-01 38.83 LINE SEARCH : LAMBDA=1.95 ENERGY = -106.59824 14 2.890E-01 5.159E-03 -216.248872 7.815E+01 39.24 LINE SEARCH : LAMBDA=4.29 ENERGY = -121.31615 15 4.229E-01 9.960E-03 -117.419573 9.883E+01 38.92 LINE SEARCH : LAMBDA=6.43 ENERGY = -130.29987 16 4.191E-01 9.784E-03 -120.890101 -3.471E+00 39.29 LINE SEARCH : LAMBDA=9.63 ENERGY = -138.51088 17 4.134E-01 9.500E-03 -129.268811 -8.379E+00 38.03 LINE SEARCH : LAMBDA=14.4 ENERGY = -146.51132 18 4.077E-01 9.208E-03 -137.493951 -8.225E+00 41.25 LINE SEARCH : LAMBDA=21.7 ENERGY = -154.30916 19 4.018E-01 8.911E-03 -145.516237 -8.022E+00 61.41 The energy becomes very higher! So I Stoped at here! With regards, Zhou Ari.P.Seitsonen at iki.fi wrote: Dear Zhou, So you start from scratch and don't try to do a restart from the Vanderbilt calculation, as I see from the input? That's the right strategy. Did you try 'PCG MINIMIZE' or just 'PCG'? Does your system contain some "difficult" materials, namely elements for which the Troullier-Martins pseudo potential would not work properly (so that the electronic gap is closed)? Did you check that the total energy makes sense (i.e. that it should be around -293.x Hartree), and did you check that your coordinates are in the right units (Bohr from the output in CPMD, you request having them in Angstroms)? With the charge -1, does the system have a closed shell structure? Just some thoughts, I don't have any other ideas for the time being... Greetings from sunny Zurich, apsi > I optimize a geometry structure with VDB first and it reach > convergence. But when starting from this geometry, I optimize it > with TM potential, even the wavefunction never get convergene! > > The following is the main part of input and output file: > &CPMD > OPTIMIZE GEOMETRY XYZ > rESTART COORDINATES WAVEFUNCTION LATEST > HESSIAN UNIT > CONVERGENCE GEOMETRY > 1.D-4 > LBFGS > CONVERGENCE ADAPT > 0.02 > CONVERGENCE ENERGY > 0.05 > CONVERGENCE RELAX > 5 > MAXSTEP > 5000 > &END > &SYSTEM > ANGSTROM > ....... > ....... > CUTOFF > 35.0 > CHARGE > -1 > > ODIIS| Insufficient progress; reset! > 24 6.046E-04 4.552E-05 -293.831117 -2.213E-03 54.45 > 25 5.961E-04 4.807E-05 -293.833460 -2.343E-03 41.65 > RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 32.0 > 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 > 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 > 26 5.865E-04 4.957E-05 -293.840304 -6.843E-03 48.00 > 27 6.009E-04 5.068E-05 -293.844066 -3.763E-03 59.75 > 28 5.769E-04 5.110E-05 -293.835152 8.914E-03 51.56 > 29 5.814E-04 4.859E-05 -293.830313 4.839E-03 53.30 > 30 5.882E-04 4.593E-05 -293.828870 1.443E-03 57.19 > RELAXING WAVEFUNCTION TOLERANCE, NEW FACTOR: 64.0 > > I also try a high energy cutoff and PCG, but it is in vain. Can > anyone tell me what's the wrong with it? > > Many thanks in advance! > > With regards, > Zhou > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland --------------------------------- Do You Yahoo!? "?????????????????????????? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030614/7b9c6814/attachment.html From audrius.alkauskas at unibas.ch Mon Jun 16 15:30:54 2003 From: audrius.alkauskas at unibas.ch (Audrius Alkauskas) Date: Mon, 16 Jun 2003 15:30:54 +0200 Subject: [CPMD-list] Generating Gaussian ECPs from PPs Message-ID: <3EEDC68E.2000203@unibas.ch> Hello I have generated files for TM BLYP pseudopotentials with fhi98PP. I use TM PPs in CPMD, but I would also like to use them with localized basis. Would it be anyhow justified to fit the pseudopotential by least squares fit to a sum of Gaussians (Stuttgart-Dresden type EPCs) in order I could use these pseudopotentials in Gaussian98? Also, can basis set be generated from pseudoelectron orbitals also fitting the output to several contractions of Gaussians? Regards Audrius Alkauskas From proffess at mail.ru Mon Jun 16 16:14:03 2003 From: proffess at mail.ru (Sergey) Date: Mon, 16 Jun 2003 18:14:03 +0400 Subject: [CPMD-list] Error on HP mpi Message-ID: Dear CPMD users, I tried to run CPMD on HP-PA RISC SUperdome using HP MPI. I got: mpirun -np 2 cpmd.x test.in mpid: MPI Application exited before MPI_Finalize() with status 5 cpmd.x: Rank 0: MPI_Comm_rank: Invalid communicator: Null communicator What doest it mean? BEst wishes, Sergey From resumes at workwondersstaffing.net Tue Jun 17 23:06:47 2003 From: resumes at workwondersstaffing.net (Work Wonders Staffing) Date: Tue, 17 Jun 2003 14:06:47 -0700 Subject: [CPMD-list] Senior Drug Discovery Applications Scientist Message-ID: Senior Drug Discovery Applications Scientist I am assisting to recruit a Senior Drug Discovery Applications Scientist for a client with offices in Oregon, California and New York. This client is recognized as a high-growth leader in the development and delivery computational methods for drug discovery and development. This company offers excellent compensation and generous benefits. Here is more information: They are seeking a scientist for their Drug Discovery Applications Group. The successful candidate will become part of a multidisciplinary drug discovery team working on both in-house efforts to improve and apply drug discovery methodologies and on collaborative projects with pharmaceutical/biotechnology partners. Job Description: Actively participate in Drug Discovery Applications Group projects Apply drug discovery software and perform analysis of the results Effectively communicate results to other members of the Group Work closely with the Development Team to test and improve Drug Discovery technologies Essential Qualifications and Experience: Ph.D. in computational chemistry, organic chemistry, or biochemistry Computational experience in at least 1 of the following areas: molecular modeling, structure-based drug design, virtual ligand screening, computationally-driven lead optimization, ligand-based design, or homology modeling Excellent verbal and written communication skills Ability to work independently Desirable Skills: Programming skills: Perl script writing, in particular. Experience with chemical databases Benefits include: Medical, Dental, 401K, Flexible Spending Account, 3+ weeks vacation and tuition reimbursement. What do you think? If you are interested in pursuing this opportunity, please forward to resumes at workwondersstaffing.net: -your resume in word or text format -your salary requirement It is important to use the position title as the subject of your email to enable your resume to reach me as quickly as possible. If you aren't interested or feel you are not qualified, do you know anyone who would be interested? We do offer generous referral fees. Thanks for your time and have a great day! Khristine Anderson Senior Recruiter Work Wonders Staffing www.workwondersstaffing.net resumes at workwondersstaffing.net --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.490 / Virus Database: 289 - Release Date: 6/16/03 From uludogan at ising.physics.metu.edu.tr Wed Jun 18 02:26:20 2003 From: uludogan at ising.physics.metu.edu.tr (uludogan) Date: Tue, 17 Jun 2003 20:26:20 -0400 Subject: [CPMD-list] Calculations on Gd Message-ID: <3EEFB1AC.CB245C9A@ising.physics.metu.edu.tr> Hello, I am using CPMD version 3.7 for a couple of weeks and I am trying to understand the code.My concern is to do some calculations on Gd which has hexagonal structure and the basis contains two atoms located one at origin the other at 1/3(a+b)+1/2 c( a,b,c are vectors).. I used Goedeckers PP (only this pseudopotential is available for Gd.). Before going into the calculations deeply , I observed the efficient CUTOFF for this system. I selected the final CUTOFF=60 au . After 60 au, at 70 au cutoff, optimization process faile for wavefunction optimization. And also during the optimization process I could not include the Monchorst Pack k-meshes into the system(another failure- all of these are discussed in this mail list before but i would like to ask if there is a new solution to do the optimization with the other k-points included to Gamma point) Finally I wanted to obtain the optimim c/a ratio for hexagonal structure. The easiest way is to do it by hand. So, I kept the volume constant and changed the c/a ratio. The experimental values are used as an input i.e. a=6.8766 au and c/a is 1.59 au and obtained c/a (au) Total E(au) 1.59 -201.48303666 1.58 -201.48378383 1.57 -201.48443310 1.56 -201.48517790 1.55 -201.48608282 1.54 -201.48682439 1.50 -201.48937426 1.40 -201.49614118 1.30 -201.50164759 I know that there should be an optimum c/a ratio where the energy is minimum but here the dependence of the total energy to c/a ratio seemed to me linear.Here is my typical input file &CPMD OPTIMIZE WAVEFUNCTION RESTART WAVEFUNCTION ODIIS 20.0 MAXSTEP 100000 TIMESTEP 20.0 &END &SYSTEM POINT GROUP AUTO SYMMETRY 4 CELL 7.35616249436 1.000 1.300 0.00 0.00 -0.50 CUTOFF 60 SCALE CARTESIAN TESR 7 &END &ATOMS *Gd-q18 LMAX=D 2 0.000000 0.000000 0.000000 1.0/3.0 1.0/3.0 0.500000 &END &DFT FUNCTIONAL LDA &END and one of the output is PROGRAM CPMD STARTED AT: Wed Jun 18 02:43:22 2003 ****** ****** **** **** ****** ******* ******* ********** ******* *** ** *** ** **** ** ** *** ** ** *** ** ** ** ** ** ** ******* ** ** ** ** *** ****** ** ** ** *** ******* ** ** ** ******* ****** ** ** ** ****** VERSION 3.7.1 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 *** Jun 8 2003 -- 04:52:22 *** THE INPUT FILE IS: inp2atoms3 THIS JOB RUNS ON: ising.physics.metu.edu.tr THE CURRENT DIRECTORY IS: /data2/uludogan/cpmd_trial/newrun3/scaling/a_c1.30 THE TEMPORARY DIRECTORY IS: /data2/uludogan/cpmd_trial/newrun3/scaling/a_c1.30 THE PROCESS ID IS: 21098 THE JOB WAS SUBMITTED BY: uludogan SINGLE POINT DENSITY OPTIMIZATION PATH TO THE RESTART FILES: ./ RESTART WITH OLD ORBITALS GRAM-SCHMIDT ORTHOGONALIZATION MAXIMUM NUMBER OF STEPS: 100000 STEPS PRINT INTERMEDIATE RESULTS EVERY 100001 STEPS STORE INTERMEDIATE RESULTS EVERY 100001 STEPS NUMBER OF DISTINCT RESTART FILES: 1 TEMPERATURE IS CALCULATED ASSUMING EXTENDED BULK BEHAVIOR FICTITIOUS ELECTRON MASS: 400.0000 TIME STEP FOR ELECTRONS: 20.0000 TIME STEP FOR IONS: 20.0000 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: 20 STEPS UNTIL DIIS RESET ON POOR PROGRESS: 20 FULL ELECTRONIC GRADIENT IS USED SPLINE INTERPOLATION IN G-SPACE FOR PSEUDOPOTENTIAL FUNCTIONS NUMBER OF SPLINE POINTS: 5000 EXCHANGE CORRELATION FUNCTIONALS LDA EXCHANGE: NONE LDA XC THROUGH PADE APPROXIMATION S.GOEDECKER, J.HUTTER, M.TETER PRB 54 1703 (1996) *** DETSP| THE NEW SIZE OF THE PROGRAM IS 1460 kBYTES *** AUTOMATIC DETERMINATION OF THE POINT GROUP: THE CRYSTAL SYSTEM IS HEXAGONAL WITH 4 OPERATIONS: 1 2[ 0 0 1] -1 -2[ 0 0 1] THE SPACE GROUP IS NON-SYMMORPHIC, (SUM OF TRANSLATION VECTORS= 2.718234) POINT GROUP: GROUP ORDER= 4 SUBGROUP OF 6/mmm(d6h) [INDEX=24] NUMBER OF PRIMITIVE CELL: 1 THE POINT GROUP OF THE CRYSTAL CONTAINS THE INVERSION SYMMETRY UNIQUE (INEQUIVALENT) ATOM: 1 REQUIRED PRECISION FOR SYMMETRY: 1.00E-06 NUMBER OF IRREDUCIBLE REPRESENTATIONS: 4 DIMENSION OF IR: 1 1 1 1 ***************************** ATOMS **************************** NR TYPE X(bohr) Y(bohr) Z(bohr) MBL 1 Gd 0.000000 0.000000 0.000000 3 2 Gd 2.452054 2.452054 4.781506 3 **************************************************************** NUMBER OF STATES: 18 NUMBER OF ELECTRONS: 36.00000 CHARGE: 0.00000 ELECTRON TEMPERATURE(KELVIN): 0.00000 OCCUPATION 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 Goedecker/Hartwigsen s ? PP^M **************************************************************** * ATOM MASS RAGGIO NLCC PSEUDOPOTENTIAL * * Gd 157.2500 1.2000 NO GOEDECKER S NONLOCAL * * P NONLOCAL * * D NONLOCAL * * F NONLOCAL * * G LOCAL * **************************************************************** *** RGGEN| THE NEW SIZE OF THE PROGRAM IS 3180 kBYTES *** ************************** SUPERCELL *************************** SYMMETRY: HEXAGONAL POINT GROUP : SUBGROUP OF 6/mmm(d6h) GROUP ORDER= 4 LATTICE CONSTANT(a.u.): 7.35616 CELL DIMENSION: 7.3562 1.0000 1.3000 0.0000 0.0000 -0.5000 VOLUME(OMEGA IN BOHR^3): 448.15467 LATTICE VECTOR A1(BOHR): 7.3562 0.0000 0.0000 LATTICE VECTOR A2(BOHR): -3.6781 6.3706 0.0000 LATTICE VECTOR A3(BOHR): 0.0000 0.0000 9.5630 RECIP. LAT. VEC. B1(2Pi/BOHR): 0.1359 0.0785 0.0000 RECIP. LAT. VEC. B2(2Pi/BOHR): 0.0000 0.1570 0.0000 RECIP. LAT. VEC. B3(2Pi/BOHR): 0.0000 0.0000 0.1046 REAL SPACE MESH: 40 40 48 WAVEFUNCTION CUTOFF(RYDBERG): 60.00000 DENSITY CUTOFF(RYDBERG): (DUAL= 4.00) 240.00000 NUMBER OF PLANE WAVES FOR WAVEFUNCTION CUTOFF: 1761 NUMBER OF PLANE WAVES FOR DENSITY CUTOFF: 14058 NUMBER OF PLANE WAVES FOR BROYDEN MIXING: 14058 **************************************************************** *** RINFORCE| THE NEW SIZE OF THE PROGRAM IS 7336 kBYTES *** *** FFTPRP| THE NEW SIZE OF THE PROGRAM IS 9340 kBYTES *** GENERATE ATOMIC BASIS SET Gd SLATER ORBITALS 4F ALPHA= 8.2162 OCCUPATION= 7.00 5S ALPHA= 3.0750 OCCUPATION= 2.00 5P ALPHA= 3.0750 OCCUPATION= 6.00 5D ALPHA= 2.7500 OCCUPATION= 1.00 6S ALPHA= 0.7143 OCCUPATION= 2.00 INITIALIZATION TIME: 0.40 SECONDS *** WFOPTS| THE NEW SIZE OF THE PROGRAM IS 10128 kBYTES *** RV30! SUPERCELL DIMENSIONS HAVE CHANGED: NUMBER= 1 RV30! SUPERCELL DIMENSIONS HAVE CHANGED: NUMBER= 3 RV30! X REAL SPACE MESH HAS CHANGED 36 40 RV30! Y REAL SPACE MESH HAS CHANGED 36 40 RV30! Z REAL SPACE MESH HAS CHANGED 54 48 RESTART INFORMATION READ ON FILE ./RESTART *** PHFAC| THE NEW SIZE OF THE PROGRAM IS 10916 kBYTES *** **************************************************************** * ATOMIC COORDINATES * **************************************************************** 1 Gd 0.000000 0.000000 0.000000 2 Gd 2.452054 2.452054 4.781506 **************************************************************** CPU TIME FOR WAVEFUNCTION INITIALIZATION: 0.01 SECONDS *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 10936 kBYTES *** EWALD SUM IN REAL SPACE OVER 8* 8* 8 CELLS TOTAL INTEGRATED ELECTRONIC DENSITY IN G-SPACE = 35.999992 IN R-SPACE = 35.999992 (K+E1+L+N+X) TOTAL ENERGY = -162.84890499 A.U. (K) KINETIC ENERGY = 53.21308237 A.U. (E1=A-S+R) ELECTROSTATIC ENERGY = -185.89705748 A.U. (S) ESELF = 215.42882286 A.U. (R) ESR = 0.00010972 A.U. (L) LOCAL PSEUDOPOTENTIAL ENERGY = -9.34700606 A.U. (N) N-L PSEUDOPOTENTIAL ENERGY = -5.30479396 A.U. (X) EXCHANGE-CORRELATION ENERGY = -15.51312988 A.U. NFI GEMAX CNORM ETOT DETOT TCPU 1 1.005E+00 9.520E-02 -162.848905 0.000E+00 2.43 2 5.180E-01 4.154E-02 -192.036500 -2.919E+01 2.42 . . . 63 1.076E-05 2.278E-06 -201.501648 -4.424E-08 2.87 64 8.561E-06 2.057E-06 -201.501648 -3.323E-08 2.87 RESTART INFORMATION WRITTEN ON FILE ./RESTART.1 *** RWFOPT| THE NEW SIZE OF THE PROGRAM IS 21180 kBYTES *** **************************************************************** * * * FINAL RESULTS * * * **************************************************************** ATOM COORDINATES GRADIENTS (-FORCES) 1 Gd 0.0000 0.0000 0.0000 -0.123E-01 0.742E-02 0.000E+00 2 Gd 2.4521 2.4521 4.7815 0.123E-01 -0.742E-02 0.000E+00 **************************************************************** ELECTRONIC GRADIENT: MAX. COMPONENT = 8.68416E-06 NORM = 1.91454E-06 NUCLEAR GRADIENT: MAX. COMPONENT = 1.23294E-02 NORM = 8.30682E-03 TOTAL INTEGRATED ELECTRONIC DENSITY IN G-SPACE = 36.000000 IN R-SPACE = 36.000000 (K+E1+L+N+X) TOTAL ENERGY = -201.50164759 A.U. (K) KINETIC ENERGY = 53.22860012 A.U. (E1=A-S+R) ELECTROSTATIC ENERGY = -207.43204195 A.U. (S) ESELF = 215.42882286 A.U. (R) ESR = 0.00010972 A.U. (L) LOCAL PSEUDOPOTENTIAL ENERGY = -17.12110044 A.U. (N) N-L PSEUDOPOTENTIAL ENERGY = -10.50958120 A.U. (X) EXCHANGE-CORRELATION ENERGY = -19.66752412 A.U. **************************************************************** ================================================================ BIG MEMORY ALLOCATIONS PME 634120 SCR 224928 TWNS 200000 YF 164738 XF 164738 PSI 164738 GDE 158530 SCR 110685 ATWFR 100400 WORK 100400 ---------------------------------------------------------------- [PEAK NUMBER 83] PEAK MEMORY 2473324 = 19.8 MBytes ================================================================ Calculating the stress tensor for electronic calculations is also important to get the minimum energy structure (--> hydrostatic pressure case).Is it possible too get the stress tensor without doing MD? because when I tried to do MD, the program failed to continue and gave an error message of "PROGRAM STOPS IN SUBROUTINE DYLMR | L NOT PROGRAMMED". which is an error message obtained in the calculations of f-electrons ( from cpmd mail-archive.) This problem is mentioned to be solved easily but is there anyone who solved it for f-electrons case? Therefore I would like to learn what is wrong with my calculations and your suggestions. I will appreciate any help. Thanks. Mustafa Uludogan School of Materials Science And Engineering Georgia Institute of Technology 771 Ferst Dr. SW Atlanta, GA 30332-0245 From Ari.P.Seitsonen at iki.fi Wed Jun 18 11:52:15 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Wed, 18 Jun 2003 11:52:15 +0200 Subject: [CPMD-list] Calculations on Gd In-Reply-To: <3EEFB1AC.CB245C9A@ising.physics.metu.edu.tr> (message from uludogan on Tue, 17 Jun 2003 20:26:20 -0400) References: <3EEFB1AC.CB245C9A@ising.physics.metu.edu.tr> Message-ID: <200306180952.h5I9qF724498@magadino.cscs.ch> Dear Mustafa, There are several issues here: - The pseudo potential must be extremely hard, I imagine, with a semi-core and f electrons. Thus I expect a very high cut-off energy necessary for convergence, 150-200 Ry maybe - With Gamma point only in such a crystal one cannot assume a reasonable value from the lattice constants or any of the properties - Are you sure about the lattice structure? In the reference which I saw the ground state was said to be hcp, but that is not what you have (I think then the coordinates should read as 0,0,0 & 1/3,2/3,1/2 - Your system is probably metallic, right? Thus what you need are fractional occupations (FREE ENERGY FUNCTIONAL does not currently work with k points though, neither can stress be calculated with them) Maybe trying these out will help your results. One solution to the high cutoff might be the Vanderbilt pseudos (however then fractional occupations or k points are not available). Greetings from the warm Zurich, apsi > I am using CPMD version 3.7 for a couple of weeks and I am trying to > understand the code.My concern is to do some calculations on Gd which > has hexagonal structure and the basis contains two atoms located one at > origin the other at > 1/3(a+b)+1/2 c( a,b,c are vectors).. I used Goedeckers PP (only this > pseudopotential is available for Gd.). > Before going into the calculations deeply , I observed the efficient > CUTOFF for this system. > I selected the final CUTOFF=60 au . > After 60 au, at 70 au cutoff, optimization process faile for > wavefunction optimization. > And also during the optimization process I could not include the > Monchorst Pack k-meshes into the system(another failure- all of these > are discussed in this mail list before but i would like to ask if there > is a new solution to do the optimization with the other k-points > included to Gamma point) > Finally I wanted to obtain the optimim c/a ratio for hexagonal > structure. The easiest way is to do it by hand. So, I kept the volume > constant and changed the c/a ratio. The experimental values are used as > an input i.e. a=6.8766 au and c/a is 1.59 au and obtained > c/a (au) Total E(au) > 1.59 -201.48303666 > 1.58 -201.48378383 > 1.57 -201.48443310 > 1.56 -201.48517790 > 1.55 -201.48608282 > 1.54 -201.48682439 > 1.50 -201.48937426 > 1.40 -201.49614118 > 1.30 -201.50164759 > > I know that there should be an optimum c/a ratio where the energy is > minimum but here the dependence of the total energy to c/a ratio seemed > to me linear.Here is my typical input file > > &CPMD > OPTIMIZE WAVEFUNCTION > RESTART WAVEFUNCTION > ODIIS > 20.0 > MAXSTEP > 100000 > TIMESTEP > 20.0 > &END > > > &SYSTEM > POINT GROUP > AUTO > SYMMETRY > 4 > CELL > 7.35616249436 1.000 1.300 0.00 0.00 -0.50 > CUTOFF > 60 > SCALE CARTESIAN > TESR > 7 > &END > > &ATOMS > *Gd-q18 > LMAX=D > 2 > 0.000000 0.000000 0.000000 > 1.0/3.0 1.0/3.0 0.500000 > &END > > &DFT > FUNCTIONAL LDA > &END > > > Calculating the stress tensor for electronic calculations is also > important to get the minimum energy structure (--> hydrostatic pressure > case).Is it possible too get the stress tensor without doing MD? because > when I tried to do MD, the program failed to continue and gave an error > message of > "PROGRAM STOPS IN SUBROUTINE DYLMR | L NOT PROGRAMMED". > which is an error message obtained in the calculations of f-electrons ( > from cpmd mail-archive.) This problem is mentioned to be solved easily > but is there anyone who solved it for f-electrons case? > > Therefore I would like to learn what is wrong with my calculations and > your suggestions. I will appreciate any help. > Thanks. > Mustafa Uludogan > > School of Materials Science And Engineering > Georgia Institute of Technology > 771 Ferst Dr. SW > Atlanta, GA 30332-0245 > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From dellsite at mpip-mainz.mpg.de Wed Jun 18 12:52:47 2003 From: dellsite at mpip-mainz.mpg.de (Luigi DelleSite) Date: Wed, 18 Jun 2003 12:52:47 +0200 Subject: [CPMD-list] Calculations on Gd References: <3EEFB1AC.CB245C9A@ising.physics.metu.edu.tr> <200306180952.h5I9qF724498@magadino.cscs.ch> Message-ID: <3EF0447F.CE29015B@mpip-mainz.mpg.de> Ari.P.Seitsonen at iki.fi wrote: > Dear Mustafa, > > There are several issues here: > > - The pseudo potential must be extremely hard, I imagine, with a > semi-core and f electrons. Thus I expect a very high cut-off energy > necessary for convergence, 150-200 Ry maybe > - With Gamma point only in such a crystal one cannot assume a > reasonable value from the lattice constants or any of the > properties > - Are you sure about the lattice structure? In the reference which I > saw the ground state was said to be hcp, but that is not what you > have (I think then the coordinates should read as 0,0,0 & > 1/3,2/3,1/2 > - Your system is probably metallic, right? Thus what you need are > fractional occupations (FREE ENERGY FUNCTIONAL does not currently > work with k points though, neither can stress be calculated with > them) > > Maybe trying these out will help your results. One solution to the > high cutoff might be the Vanderbilt pseudos (however then fractional > occupations or k points are not available). > > Greetings from the warm Zurich, > > apsi > > > I am using CPMD version 3.7 for a couple of weeks and I am trying to > > understand the code.My concern is to do some calculations on Gd which > > has hexagonal structure and the basis contains two atoms located one at > > origin the other at > > 1/3(a+b)+1/2 c( a,b,c are vectors).. I used Goedeckers PP (only this > > pseudopotential is available for Gd.). > > Before going into the calculations deeply , I observed the efficient > > CUTOFF for this system. > > I selected the final CUTOFF=60 au . > > After 60 au, at 70 au cutoff, optimization process faile for > > wavefunction optimization. > > And also during the optimization process I could not include the > > Monchorst Pack k-meshes into the system(another failure- all of these > > are discussed in this mail list before but i would like to ask if there > > is a new solution to do the optimization with the other k-points > > included to Gamma point) > > Finally I wanted to obtain the optimim c/a ratio for hexagonal > > structure. The easiest way is to do it by hand. So, I kept the volume > > constant and changed the c/a ratio. The experimental values are used as > > an input i.e. a=6.8766 au and c/a is 1.59 au and obtained > > c/a (au) Total E(au) > > 1.59 -201.48303666 > > 1.58 -201.48378383 > > 1.57 -201.48443310 > > 1.56 -201.48517790 > > 1.55 -201.48608282 > > 1.54 -201.48682439 > > 1.50 -201.48937426 > > 1.40 -201.49614118 > > 1.30 -201.50164759 > > > > I know that there should be an optimum c/a ratio where the energy is > > minimum but here the dependence of the total energy to c/a ratio seemed > > to me linear.Here is my typical input file > > > > &CPMD > > OPTIMIZE WAVEFUNCTION > > RESTART WAVEFUNCTION > > ODIIS > > 20.0 > > MAXSTEP > > 100000 > > TIMESTEP > > 20.0 > > &END > > > > > > &SYSTEM > > POINT GROUP > > AUTO > > SYMMETRY > > 4 > > CELL > > 7.35616249436 1.000 1.300 0.00 0.00 -0.50 > > CUTOFF > > 60 > > SCALE CARTESIAN > > TESR > > 7 > > &END > > > > &ATOMS > > *Gd-q18 > > LMAX=D > > 2 > > 0.000000 0.000000 0.000000 > > 1.0/3.0 1.0/3.0 0.500000 > > &END > > > > &DFT > > FUNCTIONAL LDA > > &END > > > > > > Calculating the stress tensor for electronic calculations is also > > important to get the minimum energy structure (--> hydrostatic pressure > > case).Is it possible too get the stress tensor without doing MD? because > > when I tried to do MD, the program failed to continue and gave an error > > message of > > "PROGRAM STOPS IN SUBROUTINE DYLMR | L NOT PROGRAMMED". > > which is an error message obtained in the calculations of f-electrons ( > > from cpmd mail-archive.) This problem is mentioned to be solved easily > > but is there anyone who solved it for f-electrons case? > > > > Therefore I would like to learn what is wrong with my calculations and > > your suggestions. I will appreciate any help. > > Thanks. > > Mustafa Uludogan > > > > School of Materials Science And Engineering > > Georgia Institute of Technology > > 771 Ferst Dr. SW > > Atlanta, GA 30332-0245 > > > > _______________________________________________ > > CPMD-list mailing list > > CPMD-list at cpmd.org > > http://www.cpmd.org/mailman/listinfo/cpmd-list > > > > -- > -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- > Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ > Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 > Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) > Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich > Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list Dear Ari and Mustafa Is it not possible to use CPMD in the FEMD (Free Energy Molecular Dynamics) version for your problem (just add the Key word FREE ENERGY FUNCTIONAL in the section &CPMD of the input file)? K-points can be used as in fact I do with metallic systems and you have fractional occupations. However I do not have an f-electron case but a d-electron case (Nickel) , this, I think should be not that different, and I am not currently using version 3.7 but an earlier one, so why k-points with the FREE ENERGY FUNCTIONAL should not work with this later version? In the worst case Mustafa can use an earlier version. Regards Luigi -- *********************************************************************** Luigi Delle Site dellsite at mpip-mainz.mpg.de Max-Planck-Institute for Polymer Research Phone:+49 (0)6131 379 328 Theory Group,PO Box 3148 FAX:+49 (0)6131 379 340 D-55021 Mainz,Germany *********************************************************************** From Ari.P.Seitsonen at iki.fi Wed Jun 18 13:53:15 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Wed, 18 Jun 2003 13:53:15 +0200 Subject: [CPMD-list] Calculations on Gd In-Reply-To: <3EF0447F.CE29015B@mpip-mainz.mpg.de> (message from Luigi DelleSite on Wed, 18 Jun 2003 12:52:47 +0200) References: <3EEFB1AC.CB245C9A@ising.physics.metu.edu.tr> <200306180952.h5I9qF724498@magadino.cscs.ch> <3EF0447F.CE29015B@mpip-mainz.mpg.de> Message-ID: <200306181153.h5IBrFT25146@magadino.cscs.ch> Dear Luigi, > Is it not possible to use CPMD in the FEMD (Free Energy Molecular > Dynamics) version for your problem (just add the Key word FREE > ENERGY FUNCTIONAL in the section &CPMD of the input file)? K-points > can be used as in fact I do with metallic systems and you have > fractional occupations. However I do not have an f-electron case but > a d-electron case (Nickel) , this, I think should be not that > different, and I am not currently using version 3.7 but an earlier > one, so why k-points with the FREE ENERGY FUNCTIONAL should not work > with this later version? Sorry, sorry, sorry!!! The FEMD _can_ be used with k points!!! Too little sleep last night... :( Greetings, apsi -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From f_xb at hotmail.com Thu Jun 19 08:14:12 2003 From: f_xb at hotmail.com (Xiao-Bing Feng) Date: Thu, 19 Jun 2003 06:14:12 +0000 Subject: [CPMD-list] How to compile CPMD with g77? Message-ID: Dear everyone, I'm a newcomer to the CPMD. Now I'm trying to installing CPMD 3.7 on my PC (Red Hat Linux 8.0, intel Pentium III) with only g77 available for compiling. I can't find suitalbe entry in the list generated by running Configure. Could anybody told me how to do next? And is there any library routines are needed for CPMD? I tried to compile the CPMD on another platform (Linux-PGI), but I can't run cpmd.x on my PC, it asked for some library file. By the way, I'll appreciate it very much If you can sent me a input file for a simple system, so that I can test it on my PC. Many thanks in advance. Sincerely yours, Xiaobing Feng >From: Ari.P.Seitsonen at iki.fi >Reply-To: CPMD-list at cpmd.org >To: dellsite at mpip-mainz.mpg.de >CC: CPMD-list at cpmd.org, uludogan at ising.physics.metu.edu.tr >Subject: Re: [CPMD-list] Calculations on Gd >Date: Wed, 18 Jun 2003 13:53:15 +0200 >Received: from mc7-f20.law1.hotmail.com ([65.54.253.27]) by >mc7-s10.law1.hotmail.com with Microsoft SMTPSVC(5.0.2195.5600); Wed, 18 Jun >2003 04:55:45 -0700 >Received: from cpmd.org ([195.176.20.84]) by mc7-f20.law1.hotmail.com with >Microsoft SMTPSVC(5.0.2195.5600); Wed, 18 Jun 2003 04:55:12 -0700 >Received: from cpmd.org (localhost [127.0.0.1])by cpmd.org >(AIX4.3/8.9.3p2/8.9.3) with ESMTP id NAA29734;Wed, 18 Jun 2003 13:54:10 >+0200 >Received: from internet-fence.zurich.ihost.com >(internet-fence.zurich.ihost.com [195.176.20.141])by cpmd.org >(AIX4.3/8.9.3p2/8.9.3) with ESMTP id NAA29728for ; Wed, >18 Jun 2003 13:53:28 +0200 >Received: from magadino.cscs.ch (magadino.cscs.ch [148.187.130.155]) by >internet-fence.zurich.ihost.com (AIX4.3/8.9.3p2/8.8.8) with ESMTP id >NAA07278 for ; Wed, 18 Jun 2003 13:53:29 +0200 >Received: (from aps at localhost)by magadino.cscs.ch (8.11.6/8.11.6) id >h5IBrFT25146;Wed, 18 Jun 2003 13:53:15 +0200 >X-Message-Info: KtxBqYfPyq2q+b5GbwSpcDYCuxNqEkmv >Message-Id: <200306181153.h5IBrFT25146 at magadino.cscs.ch> >In-reply-to: <3EF0447F.CE29015B at mpip-mainz.mpg.de> (message from >LuigiDelleSite on Wed, 18 Jun 2003 12:52:47 +0200) >References: <3EEFB1AC.CB245C9A at ising.physics.metu.edu.tr> ><200306180952.h5I9qF724498 at magadino.cscs.ch> ><3EF0447F.CE29015B at mpip-mainz.mpg.de> >Sender: cpmd-list-admin at cpmd.org >Errors-To: cpmd-list-admin at cpmd.org >X-BeenThere: cpmd-list at cpmd.org >X-Mailman-Version: 2.0.7 >Precedence: bulk >List-Help: >List-Post: >List-Subscribe: >, >List-Id: the mailing list of the CPMD program >List-Unsubscribe: >, >List-Archive: >Return-Path: cpmd-list-admin at cpmd.org >X-OriginalArrivalTime: 18 Jun 2003 11:55:13.0691 (UTC) >FILETIME=[7A1ED6B0:01C33590] > >Dear Luigi, > > > Is it not possible to use CPMD in the FEMD (Free Energy Molecular > > Dynamics) version for your problem (just add the Key word FREE > > ENERGY FUNCTIONAL in the section &CPMD of the input file)? K-points > > can be used as in fact I do with metallic systems and you have > > fractional occupations. However I do not have an f-electron case but > > a d-electron case (Nickel) , this, I think should be not that > > different, and I am not currently using version 3.7 but an earlier > > one, so why k-points with the FREE ENERGY FUNCTIONAL should not work > > with this later version? > >Sorry, sorry, sorry!!! The FEMD _can_ be used with k points!!! Too >little sleep last night... :( > > Greetings, > > apsi > >-- >-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- > Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ > Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 > Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich >(UniZh) > Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich > Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland >_______________________________________________ >CPMD-list mailing list >CPMD-list at cpmd.org >http://www.cpmd.org/mailman/listinfo/cpmd-list _________________________________________________________________ The new MSN 8: advanced junk mail protection and 2 months FREE* http://join.msn.com/?page=features/junkmail From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Thu Jun 19 11:23:21 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Thu, 19 Jun 2003 11:23:21 +0200 Subject: [CPMD-list] How to compile CPMD with g77? In-Reply-To: Message-ID: <200306190923.h5J9NL626635@yello.theochem.ruhr-uni-bochum.de> On Thu, 19 Jun 2003 06:14:12 +0000 "Xiao-Bing Feng" wrote: > Dear everyone, > I'm a newcomer to the CPMD. Now I'm trying to installing CPMD 3.7 > on my PC (Red Hat Linux 8.0, intel Pentium III) with only g77 available > for compiling. I can't find suitalbe entry in the list generated by running > Configure. hi! there is no entry because g77 is not sufficient to compile cpmd. > Could anybody told me how to do next? And is there any library routines are > needed you need a fortran 77 compiler which also has the cray-pointer extension implemented. this seems to be true for the portland group fortran, intel fortran, absoft and lahey, but not for g77. for optimal speed you also need a highly optimized BLAS/LAPACK library. search the mailing list archives for ATLAS and/or intel MKL to get some pointers (and recommendations). you can get an unsupported evaluation version of the intel linux compilers free of charge at > for CPMD? I tried to compile the CPMD on another platform (Linux-PGI), but I > can't run > cpmd.x on my PC, it asked for some library file. try adding '-Bstatic' to LFLAGS, delete cpmd.x and run make again. this should give you a static executable. > By the way, I'll appreciate it very much If you can sent me a input > file for a simple > system, so that I can test it on my PC. there are some example inputs in the cpmd-tests.tar.Z archive try the Si64 example for starters, should take 5 to 10 minutes on a current single cpu pc if you add SPLINE POINTS 1000 to the &CPMD section. cheers, axel kohlmeyer. > Many thanks in advance. > Sincerely yours, > Xiaobing Feng > > > > > >From: Ari.P.Seitsonen at iki.fi > >Reply-To: CPMD-list at cpmd.org > >To: dellsite at mpip-mainz.mpg.de > >CC: CPMD-list at cpmd.org, uludogan at ising.physics.metu.edu.tr > >Subject: Re: [CPMD-list] Calculations on Gd > >Date: Wed, 18 Jun 2003 13:53:15 +0200 > >Received: from mc7-f20.law1.hotmail.com ([65.54.253.27]) by > >mc7-s10.law1.hotmail.com with Microsoft SMTPSVC(5.0.2195.5600); Wed, 18 Jun > >2003 04:55:45 -0700 > >Received: from cpmd.org ([195.176.20.84]) by mc7-f20.law1.hotmail.com with > >Microsoft SMTPSVC(5.0.2195.5600); Wed, 18 Jun 2003 04:55:12 -0700 > >Received: from cpmd.org (localhost [127.0.0.1])by cpmd.org > >(AIX4.3/8.9.3p2/8.9.3) with ESMTP id NAA29734;Wed, 18 Jun 2003 13:54:10 > >+0200 > >Received: from internet-fence.zurich.ihost.com > >(internet-fence.zurich.ihost.com [195.176.20.141])by cpmd.org > >(AIX4.3/8.9.3p2/8.9.3) with ESMTP id NAA29728for ; Wed, > >18 Jun 2003 13:53:28 +0200 > >Received: from magadino.cscs.ch (magadino.cscs.ch [148.187.130.155]) by > >internet-fence.zurich.ihost.com (AIX4.3/8.9.3p2/8.8.8) with ESMTP id > >NAA07278 for ; Wed, 18 Jun 2003 13:53:29 +0200 > >Received: (from aps at localhost)by magadino.cscs.ch (8.11.6/8.11.6) id > >h5IBrFT25146;Wed, 18 Jun 2003 13:53:15 +0200 > >X-Message-Info: KtxBqYfPyq2q+b5GbwSpcDYCuxNqEkmv > >Message-Id: <200306181153.h5IBrFT25146 at magadino.cscs.ch> > >In-reply-to: <3EF0447F.CE29015B at mpip-mainz.mpg.de> (message from > >LuigiDelleSite on Wed, 18 Jun 2003 12:52:47 +0200) > >References: <3EEFB1AC.CB245C9A at ising.physics.metu.edu.tr> > ><200306180952.h5I9qF724498 at magadino.cscs.ch> > ><3EF0447F.CE29015B at mpip-mainz.mpg.de> > >Sender: cpmd-list-admin at cpmd.org > >Errors-To: cpmd-list-admin at cpmd.org > >X-BeenThere: cpmd-list at cpmd.org > >X-Mailman-Version: 2.0.7 > >Precedence: bulk > >List-Help: > >List-Post: > >List-Subscribe: > >, > >List-Id: the mailing list of the CPMD program > >List-Unsubscribe: > >, > >List-Archive: > >Return-Path: cpmd-list-admin at cpmd.org > >X-OriginalArrivalTime: 18 Jun 2003 11:55:13.0691 (UTC) > >FILETIME=[7A1ED6B0:01C33590] > > > >Dear Luigi, > > > > > Is it not possible to use CPMD in the FEMD (Free Energy Molecular > > > Dynamics) version for your problem (just add the Key word FREE > > > ENERGY FUNCTIONAL in the section &CPMD of the input file)? K-points > > > can be used as in fact I do with metallic systems and you have > > > fractional occupations. However I do not have an f-electron case but > > > a d-electron case (Nickel) , this, I think should be not that > > > different, and I am not currently using version 3.7 but an earlier > > > one, so why k-points with the FREE ENERGY FUNCTIONAL should not work > > > with this later version? > > > >Sorry, sorry, sorry!!! The FEMD _can_ be used with k points!!! Too > >little sleep last night... :( > > > > Greetings, > > > > apsi > > > >-- > >-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- > > Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ > > Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 > > Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich > >(UniZh) > > Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich > > Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland > >_______________________________________________ > >CPMD-list mailing list > >CPMD-list at cpmd.org > >http://www.cpmd.org/mailman/listinfo/cpmd-list > > _________________________________________________________________ > The new MSN 8: advanced junk mail protection and 2 months FREE* > http://join.msn.com/?page=features/junkmail > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From mbdsslj2 at man.ac.uk Fri Jun 6 17:17:07 2003 From: mbdsslj2 at man.ac.uk (Luminita Jitariu) Date: Fri, 06 Jun 2003 15:17:07 +0000 Subject: [CPMD-list] NPT simulation Message-ID: <3EE0B072.32033.28D28511@localhost> An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030606/16c48e3b/attachment.html From barnes at wuchem.wustl.edu Thu Jun 12 20:01:22 2003 From: barnes at wuchem.wustl.edu (B Barnes) Date: Thu, 12 Jun 2003 13:01:22 -0500 (CDT) Subject: [CPMD-list] CPMD compile option -r8 Message-ID: Hello everyone. I am a new user of CPMD, and have been in e-mail contact with Axel Kohlmeyer, who has been of great assistance in compiling the program (I am using his libatlas.a for cpmd's BLAS/LAPACK on our Redhat 7.3 AMD 2200+ MP systems). He recommended that I forward this information to the mailing list. I have been testing various compiles of CPMD over the past few days. One particular ifc flag, -r8, makes a large difference in runtime and a slight difference in accuracy. All my tests were with the cpmd-test archive of input/output files which is available for download. using cpmd 3.7.1 and the following options in my makefile (a trivial modification of the default PC-IFC make): FFLAGS = -c -w90 -w95 -O3 -pc64 -axM -ip -tpp6 LFLAGS = -L. -latlas -Vaxlib CFLAGS = -c -O2 -Wall CPP = /lib/cpp -P -C -traditional CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC NOOPT_FLAG = CC = gcc FC = ifc LD = ifc gcc is 2.96 (redhat 7.3) and ifc is 7.1 I can perform the cpmd-test Si64/inp-1 calculation in approx. 10 minutes on a single AMD Athlon. This is an 'acceptable' amount of time -- the same benchmark completes in ~4 minutes using cpmd 3.5.4 and its default PC-IFC (with trivial modifications for libraries etc). However, if I use the -r8 flag with 3.7.1, which is present by default in PC-IFC makefile, the calculation ends after 798 iterations (terminated by the 1 hour runtime limit)! using cpmd 3.5.4, with the -r8 flag, the calculation time increases from 4 to 7 minutes (53 to 90 iterations needed for energy convergence). for reference, from the ifc .pdf manual: -r{8|16} defaults to OFF, and defines the KIND for real variables to be 8, or 16 bytes. essentially, this is forcing double precision for all real variables. double precision should be used in any quantum code anyway, right, so why is this messing things up so badly? At any rate, since -r8 is used by default in the 3.7.1 makefile, but seems to cause serious problems, I thought the list should know. -Brian From barnes at wuchem.wustl.edu Thu Jun 12 20:18:04 2003 From: barnes at wuchem.wustl.edu (B Barnes) Date: Thu, 12 Jun 2003 13:18:04 -0500 (CDT) Subject: [CPMD-list] cpmd-test calc questions Message-ID: Hello all, As mentioned in my immediately previous mail, I have been testing a new install of CPMD 3.7.1 recently, using the cpmd-test archive of input/output files available for download. It was compiled as a serial version using ifc 7.1 under Redhat 7.3, on an AMD 2200+ MP. There are a few peculiarities about the calculations in cpmd-test which I was hoping someone could answer. 1. Around half of the tests require a pseudopotential which is not distributed with 3.7.1, or in the additional pp archive on the download site (pp.tar.Z). Is there a place to download these? 2. the test job "std/c120/inp-2" crashes with a memory allocation error after the "SUPERCELL" information and "GENERATE ATOMIC BASIS SET" information is printed: INITIALIZATION TIME: 2.06 SECONDS **************************************************************** PROCESSOR 0 ALLOCATION OF 19332488 WORDS OF MEMORY FAILED **************************************************************** *** MEMORY| THE NEW SIZE OF THE PROGRAM IS 81040 kBYTES *** ================================================================ BIG MEMORY ALLOCATIONS PME 77329952 C2 38664968 C0 38664968 VR 3029698 YF 3029698 XF 3029698 GK 966768 SCG 644512 INYH 483385 SCR 322256 ---------------------------------------------------------------- [PEAK NUMBER 55] PEAK MEMORY 164865947 = 1318.9 MBytes ================================================================ PROGRAM STOPS IN SUBROUTINE MEMORY| ALLOCATION FAILED (GDE) ...this was being run on a machine with 1gig of memory - if not enough RAM was available, why can CPMD not switch to scratch disk? Is there a way to fix this behavior (looking at keywords regarding memory in the CPMD manual, it was not immediately obvious to me how to fix this problem)? 3. The test job "std/Si512/inp-1" fails with a virtually identical error, just slightly less descriptive: MEMORY| MEMORY REQUIRED: 801915002 WORDS PROGRAM STOPS IN SUBROUTINE MEMORY| TOO BIG VALUE ...I assume both problems #2 and #3 are related to not having enough physical RAM, but I do not know how to avoid this problem by having CPMD use scratch disk instead. Soon I will be attempting to compile a parallel version of the code, to see if I can increase the number of CPU's and distributed memory available, to sidestep this problem. -Brian From henning at sentef6.fi.uib.no Tue Jun 17 18:16:13 2003 From: henning at sentef6.fi.uib.no (Henning Heiberg-Andersen) Date: Tue, 17 Jun 2003 18:16:13 +0200 (CEST) Subject: [CPMD-list] Has any of you encountered this bug? Message-ID: Hi, The attached files shows an example of a CPMD bug which appears for large simulation cells. I have reproduced it with various input at different architectures, so I hope this information will be notified. The bug is also present in the 3.7.0 version. Best regards, Henning Heiberg-Andersen -------------- next part -------------- &CPMD OPTIMIZE GEOMETRY MAXITER 1 &END &DFT GC-CUTOFF 5.D-5 &END &SYSTEM ANGSTROM SYMMETRY 0 CELL 100.0 1.0 1.0 0.0 0.0 0.0 CUTOFF 25. &END &ATOMS *O_SG_LDA 0 0 0 1 0.0 0.0 0.0 *H_SG_LDA 0 0 0 2 1.188 0. 1.447 1.188 0. -1.447 &END -------------- next part -------------- PROGRAM CPMD STARTED AT: Wed May 28 15:29:39 2003 ****** ****** **** **** ****** ******* ******* ********** ******* *** ** *** ** **** ** ** *** ** ** *** ** ** ** ** ** ** ******* ** ** ** ** *** ****** ** ** ** *** ******* ** ** ** ******* ****** ** ** ** ****** VERSION 3.7.1 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 21 2003 -- 10:44:44 *** THE INPUT FILE IS: optimization.in THIS JOB RUNS ON: embla THE CURRENT DIRECTORY IS: /mnt/mldwork1/work1/henning/cpmdtest THE TEMPORARY DIRECTORY IS: /mnt/mldwork1/work1/henning/cpmdtest THE PROCESS ID IS: 254405 OPTIMIZATION OF IONIC POSITIONS PATH TO THE RESTART FILES: ./ GRAM-SCHMIDT ORTHOGONALIZATION MAXIMUM NUMBER OF STEPS: 10000 STEPS PRINT INTERMEDIATE RESULTS EVERY 10001 STEPS STORE INTERMEDIATE RESULTS EVERY 10001 STEPS STORE INTERMEDIATE RESULTS EVERY 10001 SELF-CONSISTENT STEPS NUMBER OF DISTINCT RESTART FILES: 1 TEMPERATURE IS CALCULATED ASSUMING EXTENDED BULK BEHAVIOR FICTITIOUS ELECTRON MASS: 400.0000 TIME STEP FOR ELECTRONS: 5.0000 TIME STEP FOR IONS: 5.0000 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 STEPS UNTIL DIIS RESET ON POOR PROGRESS: 10 FULL ELECTRONIC GRADIENT IS USED CONVERGENCE CRITERIA FOR GEOMETRY OPTIMIZATION: 5.000000E-04 GEOMETRY OPTIMIZATION BY GDIIS/BFGS SIZE OF GDIIS MATRIX: 5 EMPIRICAL INITIAL HESSIAN (DISCO PARAMETRISATION) SPLINE INTERPOLATION IN G-SPACE FOR PSEUDOPOTENTIAL FUNCTIONS NUMBER OF SPLINE POINTS: 5000 EXCHANGE CORRELATION FUNCTIONALS LDA EXCHANGE: NONE LDA XC THROUGH PADE APPROXIMATION S.GOEDECKER, J.HUTTER, M.TETER PRB 54 1703 (1996) *** DETSP| THE NEW SIZE OF THE PROGRAM IS 56737 kBYTES *** >>>>>>>> CENTER OF MASS HAS BEEN MOVED TO CENTER OF BOX <<<<<<<< ***************************** ATOMS **************************** NR TYPE X(bohr) Y(bohr) Z(bohr) MBL 1 O 94.235082 94.486299 94.486299 3 2 H 96.480076 94.486299 97.220733 3 3 H 96.480076 94.486299 91.751866 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 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> > OXYGEN LDA-PSEUDOPOTENTIAL (GOEDECKER TYPE) > >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> > HYDROGEN LDA-PSEUDOPOTENTIAL (GOEDECKER TYPE) > >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> **************************************************************** * ATOM MASS RAGGIO NLCC PSEUDOPOTENTIAL * * O 15.9994 1.2000 NO GOEDECKER S NONLOCAL * * P LOCAL * * H 1.0080 1.2000 NO S LOCAL * **************************************************************** PARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARA NCPU NGW NHG PLANES GXRAYS HXRAYS ORBITALS Z-PLANES 0 178076914244941 160 17766 71046 1 320 1 178076114244954 160 17766 71044 1 320 2 178076514244876 160 17766 71044 1 320 3 178077614244766 160 17767 71043 1 320 GORDER| PROGRAMING ERROR. INFORM THE PROGRAMER PROGRAM STOPS IN SUBROUTINE GORDER| ERROR IN G-VEC ORDERING (NGW) [PROC= 0] From Lionel.Guez at idris.fr Wed Jun 18 13:12:05 2003 From: Lionel.Guez at idris.fr (Lionel Guez) Date: Wed, 18 Jun 2003 13:12:05 +0200 Subject: [CPMD-list] Version 3.7.1 does not pass a test from the test suite Message-ID: <200306181312.05978.guez@idris.fr> Hello. I installed version 3.7.1 of CPMD on an IBM Power 4 machine (at IDRIS, http://www.idris.fr). I created the Makefile with : $ configure IBM-SP4-64 Compilation ended successfully. I downloaded "http://www.cpmd.org/CPMD/contrib/cpmd-test.tar.Z". The tests "pert/C6H6/in-4", "std/Si64/inp-1" and "std/Si64/inp-2" ran ok. But the test "pert/C6H6/in-1" failed. I launched it with 4 and 8 processes. In both cases, it stopped reporting the error : PROGRAM STOPS IN SUBROUTINE NMR_CURRENT| dimension mismatch [PROC= 0] PROGRAM STOPS IN SUBROUTINE NMR_CURRENT| dimension mismatch [PROC= 2] (The ranks of the processes differed in the two runs.) This problem did not occur when I tested version 3.7.0 with the same input file. I see that this message was generated from the subroutine STOPGM, called by ? calc_current ? (file ? nmr_current_p.f ?) : if (il_rho .ne. kr1*kr2*kr3 .or. nnr1 .ne. kr1*kr2*kr3) & call STOPGM('NMR_CURRENT','dimension mismatch') I also notice that the file ? nmr_current_p.F ? is one of the files that were modified between versions 3.7.0 et 3.7.1. The only difference between versions 3.7.0 and 3.7.1 of ? nmr_current_p.F ? is that the two lines above were commented in version 3.7.0. Does anyone have any idea about this problem ? Sincerely, Lionel GUEZ From carlospi at caliescali.com Wed Jun 18 17:03:12 2003 From: carlospi at caliescali.com (carlospi at caliescali.com) Date: Wed, 18 Jun 2003 10:03:12 -0500 (COT) Subject: [CPMD-list] cpmd2cube in ifc Message-ID: <1055948592.3ef07f30d3e12@correo.caliescali.com> Hello all I dont know how compiler the cpmd2cube in IFC and processor Intel-Xeon. What changes is necessary in makefile? Thanks Carlos _____________________________________________________ Conoce nuestra nueva imagen http://www.telesat.com.co?cod=tlst-0618 From cur at zurich.ibm.com Thu Jun 19 21:27:29 2003 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Thu, 19 Jun 2003 21:27:29 +0200 Subject: [CPMD-list] cpmd-test calc questions In-Reply-To: Message-ID: Hello. The tests will be better organized in the next release of the code (due in some days) and some of them has been erroneously deleted during the ongoing reorganization. Yes, the problems in 2 and 3 are due to a too small RAM in your hardware. 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 From carlospi at estudiantes.fisica.unal.edu.co Thu Jun 19 23:43:20 2003 From: carlospi at estudiantes.fisica.unal.edu.co (Pinilla Castellanos Carlos) Date: Thu, 19 Jun 2003 16:43:20 -0500 (COT) Subject: [CPMD-list] cpmd2cube Message-ID: Hello All, I trying compiling cpmd2cube, my makefile for IFC, Intel Xeon, and Ret Hat 7.2 is: ****************************************************************************** SUFFIXES: .o .F90 ## Cray T3E #FC = f90 -f free #FFLAGS = -D__T3E -DFFT_T3E -eI -F # -g # -R abcps -ei -m 0 #LFLAGS = #LIBS = FC = ifc FFLAGS = -c -w -r8 -w90 -w95 -O3 -pc64 -axW -ip -tpp7 -DFFT_DEFAULT -D_IFC LFLAGS = LIBS = -L/opt/intel/mkl/32/ -lmkl_lapack -lmkl_p4 \ -lguide -lpthread -Vaxlib -g2caux ############## EXE = cpmd2cube ************************************************************************* and the error is ************************************************************************* ifc -c -c -w -r8 -w90 -w95 -O3 -pc64 -axW -ip -tpp7 -DFFT_DEFAULT -D_IFC util.F90 module UTIL module subroutine MY_GETARG getarg not defined!!! ^ Error 24 at (25:util.F90) : syntax error getarg not defined!!! ^ Error 24 at (25:util.F90) : syntax error getarg not defined!!! ^ Error 7 at (25:util.F90) : incomplete statement module function FILE_EXISTS module function WSTRLEN module subroutine QUICKSORT Error 355 : In procedure MY_GETARG variable GETARG has not been given a type 4 Errors compilation aborted for util.F90 (code 1) make: *** [util.o] Error 1 ******************************************************************************* thanks for your help Carlos From hutter at pci.unizh.ch Fri Jun 20 09:03:55 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Fri, 20 Jun 2003 09:03:55 +0200 (MEST) Subject: [CPMD-list] Generating Gaussian ECPs from PPs In-Reply-To: <3EEDC68E.2000203@unibas.ch> References: <3EEDC68E.2000203@unibas.ch> Message-ID: Hi in principle you can do this type of fitting, but you should consider that: - the Stuttgart ECP's (and all other ECP's normaly used in LCAO programs) are using a Z/r term whereas the plane wave pseudos have a Z erf(r)/r term. So the fit will never be very good. - A good basis set cannot be optained by just fitting to the pseudo valence wavefunction. However, you could use one of the standard methods used in quantum chemistry to get a basis set, once you have the pseudopotential. As you will never be able to reproduce numbers exactly it is probably best to use a pseudopotential from the quantum chemistry world with the LCAO calculations. You should be able to achieve results close to the plane wave calculations (or the other way around). regards Juerg Hutter ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Mon, 16 Jun 2003, Audrius Alkauskas wrote: > Hello > > I have generated files for TM BLYP pseudopotentials with fhi98PP. I use > TM PPs in CPMD, but I would also like to use them with localized basis. > Would it be anyhow justified to fit the pseudopotential by least squares > fit to a sum of Gaussians (Stuttgart-Dresden type EPCs) in order I could > use these pseudopotentials in Gaussian98? Also, can basis set be > generated from pseudoelectron orbitals also fitting the output to > several contractions of Gaussians? > > Regards > > Audrius Alkauskas > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Fri Jun 20 09:26:42 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Fri, 20 Jun 2003 09:26:42 +0200 (MEST) Subject: [CPMD-list] NPT simulation In-Reply-To: <3EE0B072.32033.28D28511@localhost> References: <3EE0B072.32033.28D28511@localhost> Message-ID: Hi The problem is associated with the way CPMD performs constant presure calculations. The basis set during the calculation is fixed by the cutoff and the cell at the beginning of the calculation. With other words we are performing a constant number of plane wave calculation and not a constnat cutoff calculation. To achieve an almost constant cutoff calculation have a look into the manual (keyword CONSTANT CUTOFF in section &SYSTEM). The reference for this method is P. Focher et al. Europhys. Lett 26 p345 (1994). But now back to your problem. Whenever the cell changes the values for the plane waves change, only the Miller indices are the same. This means we have to recalculate all quantities that depend on the plane wave values, especially the pw representation of the pseudopotentials. To avoid excessive recalculations, these functions are initially calculated on a grid and then interpolated using splines. The last point of the grid is given by the cutoff at the start of the calculation. Now in you calculation the cell is shrinking and therefore the effective cutoff is increased. The interpolation has to calculated values beyond the last grid point and becauese this is numerically instable the code issues a warning. What can you do? 1) Use the REFERENCE CELL keyword in the &SYSTEM section. The reference cell is used to calculate the initial plane waves (default is to use the normal cell). If you choose a reference cell larger than the maximal cell you expect in you simulation you are guaranteed that during the calculation the cutoff stays above the value you have chosen. 2) Use the SPLINE RANGE keyword in section &CPMD to increase the range of the grid for the spline interpolation. A value of 2 - 5 might be needed, depending on the change in volume. You migth also want to increas the number of spline points. regards Juerg Hutter ---------------------------------------------------------- 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 Fri, 6 Jun 2003, Luminita Jitariu wrote: > > Hello > I am trying to run an NPT simulation (Parrinello-Rahman for a liquid at high T and P) > I have done wf. optimization and 10 steps to get the initial velocities and then started the > MD with the following section > &CPMD > ??? RESTART WAVEFUNCTION GEOFILE LATEST > ??? MOLECULAR DYNAMICS PT > ??? PARRINELLO-RAHMAN > ??? NOSE IONS > ??? 523.? 1100. > ??? MAXSTEPS > ??? 10000 > ??? TIMESTEP > ??? 5.0 > ??? EMASS > ??? 1100 > ??? TRAJECTORY SAMPLE > ??? 100 > ??? STORE > ??? 10 > ??? STRESS TENSOR > ??? 1000 > &END > Could anybody tell me what does the following warning message mean? (it comes out at every > single step) > WARNING| SPLINE REGION SMALLER THAN MAXIMUM G-VALUE > ??????? 49? 0.13389? 0.00590? 803.8??? -525.56120??? -525.18114??? -525.04135?? 0.759 3.42 > WARNING| SPLINE REGION SMALLER THAN MAXIMUM G-VALUE > ??????? 50? 0.13175? 0.00623? 783.1??? -525.54972??? -525.17930??? -525.04132?? 0.776 3.42 > Thanks > Luminita > Dr. Luminita C. Jitariu > University of Manchester > Department of Chemistry > Oxford Rd > Manchester > M13 9PL > _______________________________________________ CPMD-list mailing list CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From hutter at pci.unizh.ch Fri Jun 20 09:34:05 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Fri, 20 Jun 2003 09:34:05 +0200 (MEST) Subject: [CPMD-list] Has any of you encountered this bug? In-Reply-To: References: Message-ID: Hi this is a know problem in CPMD. It appears in calculations with high cutoffs and/or large unit cells. It is related with the ordering of g-vectors and the parallel algorithm used. Usually changing the cutoff slightly or changing the cell parameters by small amounts (about 0.0001) helps. However there might be no hope for the size of cells you are trying. The good news is that the check is not 100% accurate. This means by just ignoring the message you will most likely get correct results. The error would only appear in restarts where you could see a small inconsistency in energy in the first step. Final results should not be affected (except for MD if you do restarts). To avoid the stop, comment out the two lines at the end of file lodapa.F of the form CALL STOPGM('GORDER','ERROR IN G-VEC ORDERING (NHG)') regards Juerg Hutter ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Tue, 17 Jun 2003, Henning Heiberg-Andersen wrote: > > Hi, > > The attached files shows an example of a CPMD bug > which appears for large simulation cells. I have > reproduced it with various input at different architectures, > so I hope this information will be notified. The bug is also > present in the 3.7.0 version. > > Best regards, > Henning Heiberg-Andersen > > From sebastia at mpip-mainz.mpg.de Fri Jun 20 10:33:25 2003 From: sebastia at mpip-mainz.mpg.de (Daniel Sebastiani) Date: Fri, 20 Jun 2003 10:33:25 +0200 (CEST) Subject: [CPMD-list] Version 3.7.1 does not pass a test from the test suite In-Reply-To: <200306181312.05978.guez@idris.fr> References: <200306181312.05978.guez@idris.fr> Message-ID: Dear Lionel, you can safely comment out the two lines. The problem occurred only on some machines, therefore it was not detected at the release of v3.7.1. regards Daniel On Wed, 18 Jun 2003, Lionel Guez wrote: > Hello. > > I installed version 3.7.1 of CPMD on an IBM Power 4 machine (at IDRIS, > http://www.idris.fr). I created the Makefile > with : > > $ configure IBM-SP4-64 > > Compilation ended successfully. > > I downloaded "http://www.cpmd.org/CPMD/contrib/cpmd-test.tar.Z". The tests > "pert/C6H6/in-4", "std/Si64/inp-1" and "std/Si64/inp-2" ran ok. But the test > "pert/C6H6/in-1" failed. I launched it with 4 and 8 processes. In both cases, > it stopped reporting the error : > > PROGRAM STOPS IN SUBROUTINE NMR_CURRENT| dimension mismatch [PROC= 0] > PROGRAM STOPS IN SUBROUTINE NMR_CURRENT| dimension mismatch [PROC= 2] > > (The ranks of the processes differed in the two runs.) > This problem did not occur when I tested version 3.7.0 with the same input > file. > > I see that this message was generated from the subroutine STOPGM, called by ? > calc_current ? (file ? nmr_current_p.f ?) : > > if (il_rho .ne. kr1*kr2*kr3 .or. nnr1 .ne. kr1*kr2*kr3) > & call STOPGM('NMR_CURRENT','dimension mismatch') > > I also notice that the file ? nmr_current_p.F ? is one of the files that were > modified between versions 3.7.0 et 3.7.1. The only difference between > versions 3.7.0 and 3.7.1 of ? nmr_current_p.F ? is that the two lines above > were commented in version 3.7.0. > > Does anyone have any idea about this problem ? > > Sincerely, > Lionel GUEZ > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From cur at zurich.ibm.com Fri Jun 20 10:36:11 2003 From: cur at zurich.ibm.com (Alessandro Curioni) Date: Fri, 20 Jun 2003 10:36:11 +0200 Subject: [CPMD-list] Difference between IBM-SP3-SMP and IBM-SP3 In-Reply-To: <200306121013.13629.tdeutsch@cea.fr> Message-ID: Hello. The SP3 Makefiles were not updated in the last release. If you want to use the full MPI/OpenMP implementation the -qsmp=omp,noauto flags has to be added in the FFLAGS and LFLAGS. However, in contrast to the power4, mixed implementations on power3 based computers, give better results only on systems with > 1000 atoms. My suggestion is to use the full MPI (IBM-SP3) setting. This , however, is a statement thas is very system dependent , because 2 processors POwer3 based nodes behave in a completely different way than 16 processors Power3+ based nodes. If you are interested to know more or specific flags for a specific architecture please contact us. Best Regards, From Ari.P.Seitsonen at iki.fi Fri Jun 20 11:16:44 2003 From: Ari.P.Seitsonen at iki.fi (Ari.P.Seitsonen at iki.fi) Date: Fri, 20 Jun 2003 11:16:44 +0200 Subject: [CPMD-list] cpmd2cube In-Reply-To: (message from Pinilla Castellanos Carlos on Thu, 19 Jun 2003 16:43:20 -0500 (COT)) References: Message-ID: <200306200916.h5K9Gid04054@magadino.cscs.ch> Dear Carlos, > ************************************************************************* > ifc -c -c -w -r8 -w90 -w95 -O3 -pc64 -axW -ip -tpp7 -DFFT_DEFAULT -D_IFC > util.F90 > module UTIL > module subroutine MY_GETARG > > getarg not defined!!! > ^ > Error 24 at (25:util.F90) : syntax error > > getarg not defined!!! > ^ > Error 24 at (25:util.F90) : syntax error > > getarg not defined!!! > ^ > Error 7 at (25:util.F90) : incomplete statement > module function FILE_EXISTS > module function WSTRLEN > module subroutine QUICKSORT > Error 355 : In procedure MY_GETARG variable GETARG has not been given a > type > > 4 Errors > compilation aborted for util.F90 (code 1) > make: *** [util.o] Error 1 > **************************************************************************** Please add || defined ( __IFC ) to the line #if defined ( __AIX ) || defined ( __SR8000 ) || defined ( __PGI ) || defined ( __IFC ) quite at the beginning of 'util.F'. Greetings, apsi -- -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/ Tel +41 1 635 44 97 / Fax +41 1 635 68 38 / GSM +41 79 719 09 35 Anschrift: Physikalisch Chemisches Institut (PCI), Universit?t Z?rich (UniZh) Indirizzo: Winterthurerstra?e 190, CH-8057 Z?rich Address: Schweiz / Svizzera / Suisse / Svizra / Switzerland From ekasciut at unity.ncsu.edu Fri Jun 20 17:37:34 2003 From: ekasciut at unity.ncsu.edu (Eliana) Date: Fri, 20 Jun 2003 10:37:34 -500 Subject: [CPMD-list] more than one thermostats In-Reply-To: <200306201420.QAA35008@cpmd.org> Message-ID: <200306201437.h5KEbWq28189@uni03mr.unity.ncsu.edu> Dear all, I want to divide my system in two, using different Nose-Hoover thermostats. Does someone know if it is possible and in that case how to do it? Thanks, Eliana Asciutto From carlospi at estudiantes.fisica.unal.edu.co Fri Jun 20 16:23:26 2003 From: carlospi at estudiantes.fisica.unal.edu.co (Pinilla Castellanos Carlos) Date: Fri, 20 Jun 2003 09:23:26 -0500 (COT) Subject: [CPMD-list] cpmd2cube1 Message-ID: Hello Thanks for your help Francesco, Ari and cpmd-list, I compiled the cpmd2cube with your changes. My makefile final is FC = ifc FFLAGS = -c -w -r8 -w90 -w95 -O3 -pc64 -axW -ip -tpp7 -DFFT_DEFAULT -D_IFC -zero LFLAGS = LIBS = -L/opt/intel/mkl/32/ -lmkl_lapack -lmkl_p4 \ -lguide -lpthread -Vaxlib -g2caux In RetHat 7.2. From FLancon at lbl.gov Fri Jun 20 21:46:11 2003 From: FLancon at lbl.gov (Frederic LANCON) Date: Fri, 20 Jun 2003 12:46:11 -0700 Subject: [CPMD-list] IBM-SP3 Message-ID: <16115.25731.718877.916740@silphion.dhcp.lbl.gov> Hello, Are there any reports available on the scalability of CPMD on SP3 parallel computers, i.e. CPU time versus number of nodes (with 16 processors)? Thanks, Fr?d?ric -- | Dr. Frederic LANCON (Fr?d?ric LAN?ON) | CEA-Grenoble, DRFMC/SP2M/L_Sim, France | present address: (@: FLancon?at?lbl.gov) | Lawrence Berkeley National Laboratory | 1 Cyclotron road, mailstop 72-150 | Berkeley, Ca 94720, USA | fax: 510 486 5888 From iivanov at cmm.chem.upenn.edu Fri Jun 20 22:25:14 2003 From: iivanov at cmm.chem.upenn.edu (Ivaylo Ivanov) Date: 20 Jun 2003 16:25:14 -0400 Subject: [CPMD-list] memory problem Message-ID: <1056140714.18723.206.camel@quicksilver> Hello all, I'm having problems using geometry optimization with QM/MM. I was wondering if any of you have seen similar errors. The code dies with the following message: MEMORY| MEMORY REQUIRED: -403304190 WORDS PROGRAM STOPS IN SUBROUTINE MEMORY| NEGATIVE OR TOO BIG VALUE [PROC= 0] Optimizing a wavefunction works without problems. I tried several options: with/without PCG MINIMIZE , BFGS, LBFGS - none of them worked. I also tried two versions of CPMD 3.6.5 and 3.5.0 with similar results. Any suggestions would be greatly appreciated.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 hutter at pci.unizh.ch Sat Jun 21 16:36:47 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Sat, 21 Jun 2003 16:36:47 +0200 (MEST) Subject: [CPMD-list] more than one thermostats In-Reply-To: <200306201437.h5KEbWq28189@uni03mr.unity.ncsu.edu> References: <200306201437.h5KEbWq28189@uni03mr.unity.ncsu.edu> Message-ID: Hi there are currently three modes available for Nose thermostats. Normal, Ultra, and Massive. Normal results in one thermostat for the complete system. Ultra has a thermostat per atomic species. Massive has a thermostat per degree of freedom. You can use the Ultra version to have different thermostats for different parts of the system. This is achieved by artificially dividing atomic species in two sets, one for system part A and one for part B. However, what is currently not possible is to have different temperatures and thermostat parameters for different thermostats. regards Juerg Hutter ---------------------------------------------------------- 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 Fri, 20 Jun 2003, Eliana wrote: > > Dear all, > > I want to divide my system in two, using different Nose-Hoover thermostats. > Does someone know if it is possible and in that case how to do it? > > Thanks, > > Eliana Asciutto > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Sat Jun 21 17:31:15 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Sat, 21 Jun 2003 17:31:15 +0200 Subject: [CPMD-list] CPMD compile option -r8 In-Reply-To: Message-ID: <200306211531.h5LFVFs10152@yello.theochem.ruhr-uni-bochum.de> On Thu, 12 Jun 2003 13:01:22 -0500 (CDT) B Barnes wrote: hello brian and everybody else on the cpmd mailing list! i have some additional data on brian's findings. remembering that one of the changes between cpmd 3.5 and 3.7 was the increase of the default number of spline points, i have performed some additional tests with varying numbers of spline points. as it turns out, i always got a converged wavefunction and almost always within less then 60 steps if i used 500, 1000, 2000 or 4000 spline points, regardles of compiler flags and cpmd version (3.5 or 3.7). but with 5000 spline points it usually took _much_ longer (i usually stopped the jobs after 150 steps and several 'ODIIS| Insufficient progress; reset!' messages). generally for the std/Si64 example i found: - cpmd v3.5 seems to be 'better' than cpmd v.3.7 - ifc v6.0.1 seems to be 'better' than ifc 7.1 - without '-r8' seems to be 'better' than with '-r8' btw: the manual still claims that the default is 3000 (its 5000 now) and hints that low cutoff runs should use a number of 500-1000 spline points. if this assertion is correct, than we might want to have cpmd print out a BIG FAT WARNING message if spline points and cutoff do not match or have a default value computed from the cutoff (probably not as helpful for the 'seasoned users' but more 'beginner friendly'). perhaps one of the real cpmd gurus might want to shed some more light on the topic of how this spline points parameter affects cpmd runs. all the best, axel kohlmeyer. > Hello everyone. I am a new user of CPMD, and have been in e-mail contact > with Axel Kohlmeyer, who has been of great assistance in compiling the > program (I am using his libatlas.a for cpmd's BLAS/LAPACK on our Redhat > 7.3 AMD 2200+ MP systems). He recommended that I forward this information > to the mailing list. > > I have been testing various compiles of CPMD over the past few days. One > particular ifc flag, -r8, makes a large difference in runtime and a > slight difference in accuracy. All my tests were with the cpmd-test > archive of input/output files which is available for download. > > using cpmd 3.7.1 and the following options in my makefile (a trivial > modification of the default PC-IFC make): > > FFLAGS = -c -w90 -w95 -O3 -pc64 -axM -ip -tpp6 > LFLAGS = -L. -latlas -Vaxlib > CFLAGS = -c -O2 -Wall > CPP = /lib/cpp -P -C -traditional > CPPFLAGS = -D__Linux -D__PGI -DLAPACK -DFFT_DEFAULT -DLINUX_IFC > NOOPT_FLAG = > CC = gcc > FC = ifc > LD = ifc > > gcc is 2.96 (redhat 7.3) and ifc is 7.1 > > I can perform the cpmd-test Si64/inp-1 calculation in approx. 10 minutes > on a single AMD Athlon. This is an 'acceptable' amount of time -- > the same benchmark completes in ~4 minutes using cpmd 3.5.4 and its > default PC-IFC (with trivial modifications for libraries etc). > > However, if I use the -r8 flag with 3.7.1, which is present by default in > PC-IFC makefile, the calculation ends after 798 iterations (terminated by > the 1 hour runtime limit)! > > using cpmd 3.5.4, with the -r8 flag, the calculation time increases from 4 > to 7 minutes (53 to 90 iterations needed for energy convergence). > > for reference, from the ifc .pdf manual: -r{8|16} defaults to OFF, and > defines the KIND for real variables to be 8, or 16 bytes. essentially, > this is forcing double precision for all real variables. double precision > should be used in any quantum code anyway, right, so why is this messing > things up so badly? > > At any rate, since -r8 is used by default in the 3.7.1 makefile, but seems > to cause serious problems, I thought the list should know. > > -Brian > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From wangyong at dicp.ac.cn Sun Jun 22 03:35:20 2003 From: wangyong at dicp.ac.cn (Yong Wang) Date: Sun, 22 Jun 2003 9:35:20 +0800 Subject: [CPMD-list] How to Install CPMD3.7 with Redhat 7.3 and PGF90 compiler? Message-ID: <200306220134.h5M1YKB21909@smtp.dicp.ac.cn> Dear all, I want to install CPMD3.7 with Redhat 7.3 and PGF90 compiler. Could you do me a favor for giving me an introduction, Thanks in advance! With Best Wishes! Yours, Sincerely Yong Wang ---------------------------------------------------------- Postbox: Yong Wang State Key Lab. of Molecular Reaction Dynamics Dalian Institute of Chemical Physics The Chinese Academy of Sciences Dalian 116023, P. R. China Tel: (+86) 0411-4379317 Fax: (+86) 0411-4676961 Email: wangyong at dicp.ac.cn Our Webpage: http://beam.dicp.ac.cn ---------------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030622/b8ca0d7c/attachment.html From axel.kohlmeyer at theochem.ruhr-uni-bochum.de Sun Jun 22 12:48:32 2003 From: axel.kohlmeyer at theochem.ruhr-uni-bochum.de (Axel Kohlmeyer) Date: Sun, 22 Jun 2003 12:48:32 +0200 Subject: [CPMD-list] How to Install CPMD3.7 with Redhat 7.3 and PGF90 compiler? In-Reply-To: <200306220134.h5M1YKB21909@smtp.dicp.ac.cn> Message-ID: <200306221048.h5MAmWO26878@yello.theochem.ruhr-uni-bochum.de> On Sun, 22 Jun 2003 9:35:20 +0800 "Yong Wang" wrote: > > This is a multi-part message in MIME format. > > --=====002_Dragon085053463075_===== > Content-Type: text/plain; > charset="GB2312" > Content-Transfer-Encoding: 7bit > > Dear all, dear yong wang, > I want to install CPMD3.7 with Redhat 7.3 and PGF90 compiler. > Could you do me a favor for giving me an introduction, Thanks in advance! > With Best Wishes! would you mind telling what _specific_ problems you have? otherwise, - there is the manual - there are some examples in the cpmd-tests tar archive - there are some pointers to introductions to car-parrinello simulations with plane waves and pseudopotentials on the cpmd website under regards, axel kohlmeyer. > > Yours, Sincerely > Yong Wang > ---------------------------------------------------------- > Postbox: > Yong Wang > State Key Lab. of Molecular Reaction Dynamics > Dalian Institute of Chemical Physics > The Chinese Academy of Sciences > Dalian 116023, P. R. China > Tel: (+86) 0411-4379317 > Fax: (+86) 0411-4676961 > Email: wangyong at dicp.ac.cn > Our Webpage: http://beam.dicp.ac.cn > ---------------------------------------------------------- > [tons of html junk deleted] > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > -- ======================================================================= Axel Kohlmeyer e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de Lehrstuhl fuer Theoretische Chemie Phone: ++49 (0)234/32-26673 Ruhr-Universitaet Bochum - NC 03/53 Fax: ++49 (0)234/32-14045 D-44780 Bochum http://www.theochem.ruhr-uni-bochum.de ======================================================================= If you make something idiot-proof, the universe creates a better idiot. From jciezak at syr.edu Mon Jun 23 15:46:55 2003 From: jciezak at syr.edu (Jenn) Date: Mon, 23 Jun 2003 09:46:55 -0400 Subject: [CPMD-list] HCTH functionals Message-ID: <00dd01c3398d$e9644360$6f1de680@spyder> Hi everyone... I am interested in using the HCTH functional on several organic radicals. However, these are not included in the library of functionals that I have, but I have seen in the manual that CPMD does support these functionals. I was wondering if anyone knew how to get these functionals and if anyone had had much success with them? Thanks, Jenn Ciezak Syracuse University Dept of Chemistry Syracuse, NY 13244 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://cpmd.org/pipermail/cpmd-list/attachments/20030623/e30d9e5e/attachment.html From M.A.Zwijnenburg at tnw.tudelft.nl Tue Jun 24 11:23:37 2003 From: M.A.Zwijnenburg at tnw.tudelft.nl (Martijn Zwijnenburg) Date: Tue, 24 Jun 2003 11:23:37 +0200 Subject: [CPMD-list] campos pp's Message-ID: <3EF834B9.20580.32269F@localhost> Hi, I've found some Vanderbilt pp's on the campos site: http://www.fysik.dtu.dk/CAMPOS/Documentation/Dacapo/PseudoPotentialOve rView/ Which I would like to use in cpmd v3.7. Is there a program/script arround to convert this pp's in a cpmd format? Greeting from Delft, 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 cavallo at chemistry.unina.it Tue Jun 24 18:52:58 2003 From: cavallo at chemistry.unina.it (Luigi Cavallo) Date: Tue, 24 Jun 2003 16:52:58 +0000 (GMT) Subject: [CPMD-list] VDB implementation in CPMD Message-ID: Hi, I've a few questions on the Vanderbilt USPP implementation in CPMD. According to the Laasonen et al paper, Phys. Rev. B 1993, 47, 10142, when dealing with USPP is better to have 2 independent energy cutoffs. Ec_soft, for the soft part of the electron density, and a much harder one, Ec_hard, for the augmentation functions Qmn. Also, they suggest to replace the Qmn by pseudocounterparts, since this allows to have a good description of all the Qmn using Ec_hard of about 200 Ry. 1) While I think that Ec_soft can be controlled in CPMD with the DUAL keyword, what about Ec_hard ? Is it used at all in CPMD ? There's a way to control it from input ? Is it fixed to somewhat value ? 2) Am I correct that the TPSEU keyword turns on the replacing of the Qmn by pseudocounterparts ? Incidentally, the only reference to TPSE in the manual says: \item[] {\bf TPSEU} asks for an external pseudization of the localized charge. This needs some care and additional input files. I used TPSEU without any additional input file (beside the USPP itself) and everything seems to work properly. Any comment ? ----------------------------------------------------------------------- At page 10149, Laasonen et al say that in order to describe the Qmn, a grid determined by the Ec_hard (about 200 Ry) is required. Efficient implementation thus requires 2 grids, determined by Ec_soft and Ec_hard, and the introduction of a small box around each ion, where the dense grid is used for the Qmn. This box moves with the ion during MD. 1) Is this the CPMD implementation ? In the case, which routines do the job ? Regards, Luigi From hutter at pci.unizh.ch Tue Jun 24 18:28:51 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 24 Jun 2003 18:28:51 +0200 (MEST) Subject: [CPMD-list] HCTH functionals In-Reply-To: <00dd01c3398d$e9644360$6f1de680@spyder> References: <00dd01c3398d$e9644360$6f1de680@spyder> Message-ID: Hi the HCTH/120 functional is implemented in CPMD 3.7 an can be accessed through FUNCTIONAL HCTH in the &DFT ... &END section. There are no pseudopotentials generated for the HCTH functional included in the CPMD distribution. If you list the elements you are looking for, maybe somebody on the list has them ready and can send them to you. The other possibility is that somebody has adapted one of the pseudopotential generating atomic codes and is willing to share it on the list. Juerg ---------------------------------------------------------- Juerg Hutter Phone : ++41 1 635 4491 Physical Chemistry Institute FAX : ++41 1 635 6838 University of Zurich E-mail: hutter at pci.unizh.ch Winterthurerstrasse 190 CH-8057 Zurich, Switzerland ---------------------------------------------------------- On Mon, 23 Jun 2003, Jenn wrote: > Hi everyone... > I am interested in using the HCTH functional on several organic radicals. However, these are not included in the library of functionals that I have, but I have seen in the manual that CPMD does support these functionals. I was wondering if anyone knew how to get these functionals and if anyone had had much success with them? > > Thanks, > Jenn Ciezak > Syracuse University > Dept of Chemistry > Syracuse, NY 13244 > From eesantis at unity.ncsu.edu Tue Jun 24 19:35:54 2003 From: eesantis at unity.ncsu.edu (Erik Santiso) Date: Tue, 24 Jun 2003 12:35:54 -500 Subject: [CPMD-list] HCTH functionals In-Reply-To: Message-ID: <200306241635.h5OGZsT24106@uni00mr.unity.ncsu.edu> Hi! I have adapted Dr. Vanderbilt's code to run with the OPTX, OLYP and HCTH-120 functionals. Is it OK to send it to the list as an attachment? (Size is about 140KB in total). Erik. > Hi > > the HCTH/120 functional is implemented in CPMD 3.7 > an can be accessed through > > FUNCTIONAL HCTH > > in the &DFT ... &END section. > > There are no pseudopotentials generated > for the HCTH functional included in the CPMD > distribution. > If you list the elements you are looking for, maybe > somebody on the list has them ready and can send them > to you. > The other possibility is that somebody has adapted > one of the pseudopotential generating atomic codes > and is willing to share it on the list. > > Juerg > > ---------------------------------------------------------- > Juerg Hutter Phone : ++41 1 635 4491 > Physical Chemistry Institute FAX : ++41 1 635 6838 > University of Zurich E-mail: hutter at pci.unizh.ch > Winterthurerstrasse 190 > CH-8057 Zurich, Switzerland > ---------------------------------------------------------- > > > On Mon, 23 Jun 2003, Jenn wrote: > > > Hi everyone... > > I am interested in using the HCTH functional on several organic radicals. However, these are not included in the library of functionals that I have, but I have seen in the manual that CPMD does support these functionals. I was wondering if anyone knew how to get these functionals and if anyone had had much success with them? > > > > Thanks, > > Jenn Ciezak > > Syracuse University > > Dept of Chemistry > > Syracuse, NY 13244 > > > _______________________________________________ > CPMD-list mailing list > CPMD-list at cpmd.org > http://www.cpmd.org/mailman/listinfo/cpmd-list > --------------------------------------- Illiterate? Write for free information. From hutter at pci.unizh.ch Tue Jun 24 18:38:16 2003 From: hutter at pci.unizh.ch (Juerg Hutter) Date: Tue, 24 Jun 2003 18:38:16 +0200 (MEST) Subject: [CPMD-list] VDB implementation in CPMD In-Reply-To: References: Message-ID: Hi > > I've a few questions on the Vanderbilt USPP implementation in CPMD. > > According to the Laasonen et al paper, Phys. Rev. B 1993, 47, 10142, > when dealing with USPP is better to have 2 independent energy cutoffs. > Ec_soft, for the soft part of the electron density, and a much harder > one, Ec_hard, for the augmentation functions Qmn. Also, they suggest > to replace the Qmn by pseudocounterparts, since this allows to have > a good description of all the Qmn using Ec_hard of about 200 Ry. > > 1) While I think that Ec_soft can be controlled in CPMD with the DUAL > keyword, what about Ec_hard ? Is it used at all in CPMD ? There's > a way to control it from input ? Is it fixed to somewhat value ? In CPMD we have Ec_hard = Ec_soft By using a DUAL > 4 you can increase the value to higher cutoffs for the density without increasing the cutoff of the wavefunctions. However, this is just a workaround and not as efficient as the implementation described in the Laasonen paper. > > 2) Am I correct that the TPSEU keyword turns on the replacing of the > Qmn by pseudocounterparts ? Incidentally, the only reference to TPSE in > the manual says: > > \item[] {\bf TPSEU} asks for an external pseudization of the localized > charge. This needs some care and additional input files. > > I used TPSEU without any additional input file (beside the USPP itself) > and everything seems to work properly. Any comment ? > All modern USPP's have pseudized Q-functions. The pseudization is automatically turned on within the code if detected in the USPP file. The keyword was necessary for (very) old USPP's where the pseudization was done after the PP generation. > ----------------------------------------------------------------------- > > At page 10149, Laasonen et al say that in order to describe the Qmn, a > grid determined by the Ec_hard (about 200 Ry) is required. Efficient > implementation thus requires 2 grids, determined by Ec_soft and Ec_hard, > and the introduction of a small box around each ion, where the dense > grid is used for the Qmn. This box moves with the ion during MD. > > 1) Is this the CPMD implementation ? In the case, which routines do > the job ? > This is also not implemented in CPMD. If you want to implement these features in CPMD you can contact me. greetings Juerg From veronique.vanspeybroeck at ugent.be Wed Jun 25 12:20:53 2003 From: veronique.vanspeybroeck at ugent.be (Veronique Van Speybroeck) Date: Wed, 25 Jun 2003 12:20:53 +0200 Subject: [CPMD-list] READING VELOCITIES Message-ID: <3EF97785.4060901@ugent.be> Hi, can anybody help me with the right format to read in velocities. I have the following inputfile : &CPMD MOLECULAR DYNAMICS CP RESTART LATEST WAVEFUNCTIONS TIMESTEP 5 MOVIE SAMPLE 1 MAXSTEP 1000 LSD PRINT 10 &END &SYSTEM ANGSTROM SYMMETRY 12 CELL 10.17399 1.15717 1.0733 0.0 -0.374493 0.0 CUTOFF 25. MULTIPLICITY 2 &END &ATOMS *C_VDB_U BINARY NEWF LMAX=P 32 0.246379602 1.22685146 5.68702235 4.04793917 7.34643205 0.892963881 4.17329858 10.4576469 -1.95385739 0.348750932 4.7098615 2.92416293 0.150265103 1.25602759 4.18859541 4.06219341 7.19049497 2.33473408 4.16377506 10.4049092 -0.437916163 0.283322521 4.56801506 1.4413751 0.321233738 1.21421457 11.2032098 4.11472361 7.20730506 6.3759174 4.10847777 10.3268882 3.57163531 0.242527677 4.63028694 8.48576077 0.210360666 1.32849529 9.71984393 4.22658102 7.20542156 7.88712334 4.15408117 10.3604427 5.06834126 0.207293197 4.46467993 6.99608433 5.2206214 1.2130572 3.50625331 9.01628486 7.18245953 -1.312820