From owner-chemistry@ccl.net Wed Sep 12 03:30:01 2007 From: "akef afaneh akef_afnh() yahoo.com" To: CCL Subject: CCL:G: G03: Restarting optimization and SCF Message-Id: <-35136-070912032836-11046-pxGZmF0KtZr43cKDWa0zYQ\a/server.ccl.net> X-Original-From: akef afaneh Content-Transfer-Encoding: 8bit Content-Type: multipart/alternative; boundary="0-1392426740-1189582105=:60254" Date: Wed, 12 Sep 2007 00:28:25 -0700 (PDT) MIME-Version: 1.0 Sent to CCL by: akef afaneh [akef_afnh=yahoo.com] --0-1392426740-1189582105=:60254 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Hi; You're not too specific on what you have already tried, so the advice I will give you is very basic stuff. There are two different SCF algorithms available in Gaussian. The default algorithm DIIS is quite fast and works well for most systems. For problematic cases the quadratically convergent (QC) algorithm is much more reliable but also much slower than DIIS. The latter option is used with the SCF=QC keyword and becomes quite usefull in the case of convergence problems with the default DIIS algorithm. In the case of severe convergence problems, a few further strategies exist to approach the problem. As a first point it is important to check the geometry of the system under investigation. Errors in the Z-Matrix definition, which position atomic centers too close together, often result in convergence problems during SCF calculations. Also, it is often found that self consistency is achieved with one method, but not another. It is therefore quite helpful to calculate a wavefunction first with one method and then use the converged wavefunction as the initial guess for another calculation. It is, for example frequently found that Hartree-Fock calculations converge more readily than DFT calculations and initial calculation of the HF wavefunction can therefore aid in getting the DFT calculation started. Ultimately, the convergence behaviour also improves with reduced electron numbers. If, for example, calculations for neutral radicals do not converge, a converged wavefunction can often be obtained for the corresponding cationic systems (lacking one electron). The cation wavefunction can then be used as the initial guess of the radical system. If still no convergence can be achieved: go look for another project! --------------------------------- Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase. --0-1392426740-1189582105=:60254 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: 8bit
Hi;
You're not too specific on what you have already tried, so the advice I will give you is very basic stuff. There are two different SCF algorithms available in Gaussian. The default algorithm DIIS is quite fast and works well for most systems. For problematic cases the quadratically convergent (QC) algorithm is much more reliable but also much slower than DIIS. The latter option is used with the SCF=QC keyword and becomes quite usefull in the case of convergence problems with the default DIIS algorithm.
In the case of severe convergence problems, a few further strategies exist to approach the problem. As a first point it is important to check the geometry of the system under investigation. Errors in the Z-Matrix definition, which position atomic centers too close together, often result in convergence problems during SCF calculations. Also, it is often found that self consistency is achieved with one method, but not another. It is therefore quite helpful to calculate a wavefunction first with one method and then use the converged wavefunction as the initial guess for another calculation. It is, for example frequently found that Hartree-Fock calculations converge more readily than DFT calculations and initial calculation of the HF wavefunction can therefore aid in getting the DFT calculation started. Ultimately, the convergence behaviour also improves with reduced electron numbers. If, for example, calculations for neutral radicals do not converge, a converged wavefunction can often be obtained for the corresponding cationic systems (lacking one electron). The cation wavefunction can then be used as the initial guess of the radical system. If still no convergence can be achieved: go look for another project!

Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase. --0-1392426740-1189582105=:60254-- From owner-chemistry@ccl.net Wed Sep 12 08:30:01 2007 From: "Aggelos Avramopoulos aggavramop||yahoo.gr" To: CCL Subject: CCL:G: Eigenvectors and Eigenvalues of the overlap matrix of G03 Message-Id: <-35137-070912082744-2337-N2g7TlCxQpvcrWnIofRuSg(a)server.ccl.net> X-Original-From: "Aggelos Avramopoulos" Date: Wed, 12 Sep 2007 08:27:40 -0400 Sent to CCL by: "Aggelos Avramopoulos" [aggavramop ~~ yahoo.gr] Dear CCLs I am using G03 and i would like to print out the eigenvectors and eigenvalues of the overlap matrix. I found that by using the option IOP(3/33=1) i can print in the produced rwf file the overlap matrix. Is there any way to print either in the log file or in any produced G03 file,the eigenvectors and the eigenvalues of the diagonilized overlap matrix? Or, alternatively since i know the overlap matrix, although not in a symmetric form, is there anyone who could provide either a code or any relevant software in order to diagonilize the matrix and to take the eigenvectors and the corresponding eigenvalues ? Thanks in advance for any help,hints and suggestions. Aggelos Avramopoulos From owner-chemistry@ccl.net Wed Sep 12 09:05:00 2007 From: "Antonio M. Marquez marquez[-]us.es" To: CCL Subject: CCL:G: G03: Restarting optimization and SCF Message-Id: <-35138-070912033918-15431-ww3Y/YDiW707xdxzrmwppw _ server.ccl.net> X-Original-From: "Antonio M. Marquez" Content-Transfer-Encoding: 7bit Content-Type: text/plain Date: Wed, 12 Sep 2007 08:37:59 +0200 Mime-Version: 1.0 Sent to CCL by: "Antonio M. Marquez" [marquez|*|us.es] Hi, Usually, when a SCF takes more than (say) 50-100 iterations means that something is wrong with your geometry. Another possibility is that a single determinant SCF wavefunction is not adequate to describe your system. In the first case you may try to start with a better initial geometry. To check the SCF convergency you should force gaussian to print the energies of all SCF iterations. In both cases I'll recommend you to do first a single point calculation until a good convergency is obtained. My 0,01 euro (not dollar) contribution. On Tue, 2007-09-11 at 15:18 -0600, Kaci Tizi_Ouzou kaci.tiziouzou-*-gmail.com wrote: > Hi All, > > > I am having a computation issue with Gaussian 03. > > I have a calculation for which BOTH geometry optimization and SCF do > not coonverge. > > Even though I have upped MAXCYC of the SCF to 1024, the convergence is > still not ok. So my question is: > > If the SCF does not converge WITHIN a geometry optimization run, > should I use: > > # blah blah OPT(restart, MAXCYC=100) <-- Restarting only the > Optimization > > OR > > # blah blah OPT( restart, MAXCYC=100) SCF(restart, MAXCYC=1024) <--- > This does not seem to work though!! > > > Any help will be much appreciated. > > > Kass > > From owner-chemistry@ccl.net Wed Sep 12 09:44:00 2007 From: "Aggelos Avramopoulos aggavramop^yahoo.gr" To: CCL Subject: CCL:G: Eigenvectors and Eigenvalues of the overlap matrix of G03 Message-Id: <-35139-070912090811-977-ThcdQnd3Wzx1uaJYddw8lA|,|server.ccl.net> X-Original-From: Aggelos Avramopoulos Content-Transfer-Encoding: 8bit Content-Type: multipart/alternative; boundary="0-704625529-1189598882=:53254" Date: Wed, 12 Sep 2007 13:08:02 +0100 (BST) MIME-Version: 1.0 Sent to CCL by: Aggelos Avramopoulos [aggavramop * yahoo.gr] --0-704625529-1189598882=:53254 Content-Type: text/plain; charset=iso-8859-7 Content-Transfer-Encoding: 8bit Dear CCLs I am using G03 and i would like to print out the eigenvectors and eigenvalues of the overlap matrix. I found that by using the option IOP(3/33=1) i can print in the produced rwf file the overlap matrix. Is there any way to print either in the log file or in any produced G03 file,the eigenvectors and the eigenvalues of the diagonilized overlap matrix? Or, alternatively since i know the overlap matrix, although not in a symmetric form, is there anyone who could provide either a code or any relevant software in order to diagonilize the matrix and to take the eigenvectors and the corresponding eigenvalues ? Thanks in advance for any help,hints and suggestions. Aggelos Avramopoulos --------------------------------- Χρησιμοποιείτε Yahoo! Βαρεθήκατε τα ενοχλητικά μηνύ ματα (spam); Το Yahoo! Mail διαθέτει την καλύτερη δυνατή προστασία κατά των ενοχλητικών μηνυμάτων http://login.yahoo.com/config/mail?.intl=gr --0-704625529-1189598882=:53254 Content-Type: text/html; charset=iso-8859-7 Content-Transfer-Encoding: 8bit
Dear CCLs
I am using G03 and i would like to print out the eigenvectors and eigenvalues
of the overlap matrix. I found that by using the option IOP(3/33=1) i can print
in the produced rwf file the overlap matrix. Is there any way to print either in the
log file or in any produced G03 file,the eigenvectors and the eigenvalues of the
diagonilized overlap matrix?
Or, alternatively since i know the overlap matrix, although not in a symmetric
form, is there anyone who could provide either a code or any relevant software
in order to diagonilize the matrix and to take the eigenvectors and the corresponding eigenvalues ?
Thanks in advance for any help,hints and suggestions.
Aggelos Avramopoulos

Χρησιμοποιείτε Yahoo!
Βαρεθήκατε τα ενοχλητικά μηνύ ματα (spam); Το Yahoo! Mail διαθέτει την καλύτερη δυνατή προστασία κατά των ενοχλητικών μηνυμάτων
http://login.yahoo.com/config/mail?.intl=gr --0-704625529-1189598882=:53254-- From owner-chemistry@ccl.net Wed Sep 12 10:14:00 2007 From: "Gustavo L.C. Moura gustavo a mercury.chem.pitt.edu" To: CCL Subject: CCL: Semiempirical Electron Affinities 2 Message-Id: <-35140-070912094903-10090-YsFspdSk0tyOl+WKMfqfyA/./server.ccl.net> X-Original-From: "Gustavo L.C. Moura" Date: Wed, 12 Sep 2007 09:48:59 -0400 Sent to CCL by: "Gustavo L.C. Moura" [gustavo##mercury.chem.pitt.edu] Dear CCL readers, Last week I asked to the list: > I have been asking myself how I can obtain reliable values for the > electron affinities of organic molecules employing semiempirical methods > like AM1, PM3 or RM1. What I mean is that these semiempirical methods > were parameterized to yield good values for the ionization potentials (it > is part of their parameterization process). But, what about electron > affinity? Should I also use Koopmans like in the case of ionization > potentials? What are your opinions? I have received only one answer from Geoff Hutchison that said: > My experience is that AM1 and PM3 have very poor electron affinity > calculations. As you say, it's not in their parameterization, so they > are very unreliable. I only tend to believe EA from higher level > calculations like B3LYP or better. > Supposedly, the new PM6 method, part of MOPAC 2007 has electron > affinities included in the parameterization. I would like to thank Prof. Hutchison for his answer and I will keep looking for ways of calculating reliable electron affinities using MOPAC. Sincerely yours, Gustavo L.C. Moura From owner-chemistry@ccl.net Wed Sep 12 10:51:01 2007 From: "Serge Gorelsky gorelsky*gmail.com" To: CCL Subject: CCL:G: Eigenvectors and Eigenvalues of the overlap matrix of G03 Message-Id: <-35141-070912094650-8313-LQoVmJ5b9DZxez7mGkUSXw(~)server.ccl.net> X-Original-From: "Serge Gorelsky" Content-Type: multipart/alternative; boundary="----=_Part_1156_4979875.1189601188859" Date: Wed, 12 Sep 2007 08:46:28 -0400 MIME-Version: 1.0 Sent to CCL by: "Serge Gorelsky" [gorelsky::gmail.com] ------=_Part_1156_4979875.1189601188859 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Content-Disposition: inline Hello, AOMix-L (check http://www.sg-chem.net) calculates and prints eigenvalues and eigenvectors of the overlap matrix from Gaussian and the other packages. Please refer to the AOMix manual for the details. Regards, Serge Gorelsky University of Ottawa On 9/12/07, Aggelos Avramopoulos aggavramop||yahoo.gr < owner-chemistry()ccl.net> wrote: > > > Sent to CCL by: "Aggelos Avramopoulos" [aggavramop ~~ yahoo.gr] > Dear CCLs > I am using G03 and i would like to print out the eigenvectors and > eigenvalues > of the overlap matrix. I found that by using the option IOP(3/33=1) i can > print > in the produced rwf file the overlap matrix. Is there any way to print > either in the > log file or in any produced G03 file,the eigenvectors and the eigenvalues > of the > diagonilized overlap matrix? > Or, alternatively since i know the overlap matrix, although not in a > symmetric > form, is there anyone who could provide either a code or any relevant > software > in order to diagonilize the matrix and to take the eigenvectors and the > corresponding eigenvalues ? > Thanks in advance for any help,hints and suggestions. > Aggelos Avramopoulos> > > > -- Best regards, Serge Gorelsky ------=_Part_1156_4979875.1189601188859 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: 7bit Content-Disposition: inline
Hello,
 
AOMix-L (check http://www.sg-chem.net) calculates and prints eigenvalues and eigenvectors of the overlap matrix from Gaussian and the other packages. Please refer to the AOMix manual for the details.
 
Regards,
Serge Gorelsky
University of Ottawa
 
On 9/12/07, Aggelos Avramopoulos aggavramop||yahoo.gr <owner-chemistry()ccl.net> wrote:

Sent to CCL by: "Aggelos  Avramopoulos" [aggavramop ~~ yahoo.gr ]
Dear CCLs
I am using G03 and i would like to print out the eigenvectors and eigenvalues
of the overlap matrix. I found that by using the option IOP(3/33=1) i can print
in the produced rwf file the overlap matrix. Is there any way to print either in the
log file or in any produced G03 file,the eigenvectors and the eigenvalues of the
diagonilized overlap matrix?
Or, alternatively since i know the overlap matrix, although not in a symmetric
form, is there anyone who could provide either a code or any relevant software
in order to diagonilize the matrix and to take the eigenvectors and the corresponding eigenvalues ?
Thanks in advance for any help,hints and suggestions.
Aggelos Avramopoulos



E-mail to subscribers: CHEMISTRY()ccl.net or use:
     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST()ccl.net or use
     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

Subscribe/Unsubscribe:
    

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemistry/announcements/conferences/

Search Messages: http://www.ccl.net/htdig   (login: ccl, Password: search)
     http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/






--
Best regards,
  Serge Gorelsky ------=_Part_1156_4979875.1189601188859-- From owner-chemistry@ccl.net Wed Sep 12 12:11:01 2007 From: "Ron Cook cookrl__tda.com" To: CCL Subject: CCL: Semiempirical Electron Affinities 2 Message-Id: <-35142-070912115800-10925-gG2tyRiavE0YdQF3M/CbWw],[server.ccl.net> X-Original-From: "Ron Cook" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset="us-ascii" Date: Wed, 12 Sep 2007 09:59:58 -0600 MIME-Version: 1.0 Sent to CCL by: "Ron Cook" [cookrl^_^tda.com] I agree with Prof Hitchison, the semi-empirical methods are not reliable for electron affinities. I have calculated a large number of IP and EA for a range of molecules in order to evaluate models involving chemical hardness and the electrophilicity indices, and the semiempirical methods are OK for defining general trends for a set of homologous molecules. To really get the IP and EA right you need to go to the G2 level of computation and calculate the neutral, anion and cation thermo energies and then do the subtractions to get IP and EA Ronald Cook -----Original Message----- > From: owner-chemistry*o*ccl.net [mailto:owner-chemistry*o*ccl.net] Sent: Wednesday, September 12, 2007 7:49 AM To: Cook, Ronald L Subject: CCL: Semiempirical Electron Affinities 2 Sent to CCL by: "Gustavo L.C. Moura" [gustavo##mercury.chem.pitt.edu] Dear CCL readers, Last week I asked to the list: > I have been asking myself how I can obtain reliable values for the > electron affinities of organic molecules employing semiempirical methods > like AM1, PM3 or RM1. What I mean is that these semiempirical methods > were parameterized to yield good values for the ionization potentials (it > is part of their parameterization process). But, what about electron > affinity? Should I also use Koopmans like in the case of ionization > potentials? What are your opinions? I have received only one answer from Geoff Hutchison that said: > My experience is that AM1 and PM3 have very poor electron affinity > calculations. As you say, it's not in their parameterization, so they > are very unreliable. I only tend to believe EA from higher level > calculations like B3LYP or better. > Supposedly, the new PM6 method, part of MOPAC 2007 has electron > affinities included in the parameterization. I would like to thank Prof. Hutchison for his answer and I will keep looking for ways of calculating reliable electron affinities using MOPAC. Sincerely yours, Gustavo L.C. Mourahttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt From owner-chemistry@ccl.net Wed Sep 12 22:05:01 2007 From: "Tanja van Mourik tanja.vanmourik(_)st-andrews.ac.uk" To: CCL Subject: CCL: Semiempirical Electron Affinities 2 Message-Id: <-35143-070912220038-31536-212JmnH0TzufYRRwBEyN/A:server.ccl.net> X-Original-From: Tanja van Mourik Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1 Date: Thu, 13 Sep 2007 02:31:18 +0100 MIME-Version: 1.0 Sent to CCL by: Tanja van Mourik [tanja.vanmourik]*[st-andrews.ac.uk] Hi All, The following paper may also be of interest to those interested in calculating electron affinities: Thom H. Dunning, Jr., Kirk A. Peterson, Tanja van Mourik; "Calculation of Electron Affinities. A Roadmap. in The Dissociative Recombination of Molecules with Electrons"; New York; S.L. Guberman (ed); Kluwer Academic/Plenum Publishers, 2002. Tanja -- ================================================================= Tanja van Mourik Royal Society University Research Fellow School of Chemistry, University of St. Andrews North Haugh, St. Andrews Fife KY16 9ST, Scotland (UK) email: tanja.vanmourik : st-andrews.ac.uk web: http://chemistry.st-and.ac.uk/staffmember.php?id=tvm ================================================================= > I agree with Prof Hitchison, the semi-empirical methods are not reliable for > electron affinities. I have calculated a large number of IP and EA for a > range of molecules in order to evaluate models involving chemical hardness > and the electrophilicity indices, and the semiempirical methods are OK for > defining general trends for a set of homologous molecules. To really get > the IP and EA right you need to go to the G2 level of computation and > calculate the neutral, anion and cation thermo energies and then do the > subtractions to get IP and EA > > Ronald Cook > > -----Original Message----- > > From: owner-chemistry[-]ccl.net [mailto:owner-chemistry[-]ccl.net] > Sent: Wednesday, September 12, 2007 7:49 AM > To: Cook, Ronald L > Subject: CCL: Semiempirical Electron Affinities 2 > > > Sent to CCL by: "Gustavo L.C. Moura" [gustavo##mercury.chem.pitt.edu] > Dear CCL readers, > Last week I asked to the list: > > > I have been asking myself how I can obtain reliable values for the > > electron affinities of organic molecules employing semiempirical methods > > like AM1, PM3 or RM1. What I mean is that these semiempirical methods > > were parameterized to yield good values for the ionization potentials (it > > is part of their parameterization process). But, what about electron > > affinity? Should I also use Koopmans like in the case of ionization > > potentials? What are your opinions? > > I have received only one answer from Geoff Hutchison that said: > > > My experience is that AM1 and PM3 have very poor electron affinity > > calculations. As you say, it's not in their parameterization, so they > > are very unreliable. I only tend to believe EA from higher level > > calculations like B3LYP or better. > > Supposedly, the new PM6 method, part of MOPAC 2007 has electron > > affinities included in the parameterization. > > I would like to thank Prof. Hutchison for his answer and > I will keep looking for ways of calculating reliable electron affinities > using MOPAC. > Sincerely yours, > Gustavo L.C. > Mourahttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt> > > > ------------------------------------------------------------------ University of St Andrews Webmail: https://webmail.st-andrews.ac.uk