From chemistry-request@server.ccl.net Wed Jun 28 15:39:26 2000 Received: from facepe.dqf.ufpe.br (root@facepe.dqf.ufpe.br [150.161.5.2]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id PAA27920 for ; Wed, 28 Jun 2000 15:38:23 -0400 Received: from npd.ufpe.br (danon.dqf.ufpe.br [150.161.5.125]) by facepe.dqf.ufpe.br (8.9.3/8.8.7) with ESMTP id QAA13193 for ; Wed, 28 Jun 2000 16:42:54 -0300 Message-ID: <395A5426.94A0BAFA@npd.ufpe.br> Date: Wed, 28 Jun 2000 16:38:15 -0300 From: Elizete Ventura da Silva X-Mailer: Mozilla 4.7 [en] (Win98; I) X-Accept-Language: en MIME-Version: 1.0 To: chemistry@ccl.net Subject: gauopt Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Hello CCL’s Does anybody knows how to optimize contraction and orbital exponents (simultaneously) in MINI basis set during geometry optimization (using gauopt option from Gaussian 98)? Thanks in advance. Elizete Ventura e-mail: elizete@npd.ufpe.br From chemistry-request@server.ccl.net Wed Jun 28 16:57:01 2000 Received: from tisch.mail.mindspring.net (tisch.mail.mindspring.net [207.69.200.157]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id QAA28244 for ; Wed, 28 Jun 2000 16:57:01 -0400 Received: from quantumnt (user-2injk30.dialup.mindspring.com [165.121.208.96]) by tisch.mail.mindspring.net (8.9.3/8.8.5) with SMTP id QAA02936 for ; Wed, 28 Jun 2000 16:56:59 -0400 (EDT) Message-ID: <000701bfe143$675b7a50$0300a8c0@mindspring> Reply-To: "Jim Kress" From: "Jim Kress" To: References: <157A51F55AAAD3119CD70008C7B1629D6383C0@LVLXCH01> Subject: Re: CCL:AIM code Date: Wed, 28 Jun 2000 16:56:57 -0400 You're out of luck. I've experienced this problem since G94W and Gaussian's response has always been, "We can't help you". At least they're consistent... You can get Baders code directly from his group and run it separately from G98. Get it here: http://www.chemistry.mcmaster.ca/aimpac/ Jim Check out my web site http://www.kressworks.com/ ----- Original Message ----- From: Shobe, Dave To: 'Alejandro Montoya' ; ; Sent: Wednesday, June 28, 2000 1:21 PM Subject: CCL:AIM code > I'd like to ask a more general question. The G98 manual lists errors that > can occur, but says nothing about ways to correct them, giving me the > impression that if the AIM code fails for any reason, you're out of luck. > Is there anything one can do when the AIM calculations fail? > > --David Shobe > Süd-Chemie Inc. > phone (502) 634-7409 > fax (502) 634-7724 > email dshobe@sud-chemieinc.com > > Any opinions herein are not necessarily representative of Süd-Chemie. > > > -----Original Message----- > From: Alejandro Montoya [mailto:alemoe@Carbon.udea.edu.co] > Sent: Wednesday, June 28, 2000 3:52 PM > To: chemistry@ccl.net > Subject: CCL:AIM code > > > > Hello CCLers, > > I would apreciate if someone can point me on what is the best procedure > to obtained the Bond order using the AIM theory when the AIM code fails > in G98 and an error message is obtained at the end of the file. > > NO CONVERGENCE IN SVDSOL FOR SURFACE SHEET 14 > Error termination > > thanks in advance, > > Regards, > > > > > > > > -= This is automatically added to each message by mailing script =- > CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To Admins > MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH > CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net 70 > Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: jkl@ccl.net > > > > From chemistry-request@server.ccl.net Wed Jun 28 19:17:36 2000 Received: from shiva.hunter.cuny.edu (shiva.hunter.cuny.edu [146.95.128.96]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id TAA28751 for ; Wed, 28 Jun 2000 19:17:36 -0400 Received: from shiva (shiva [146.95.128.96]) by shiva.hunter.cuny.edu (8.9.3/8.9.3) with ESMTP id TAA19134; Wed, 28 Jun 2000 19:17:43 -0400 (EDT) Date: Wed, 28 Jun 2000 19:17:42 -0400 (EDT) From: AM To: chemistry@CCL.NET cc: Artem Masunov , adios@picower.edu Subject: SUMMARY: Computational Chemists in Consulting Businesses Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear CCLers, Thank you all for your help! Especial thanks to Wibke Sudholt, Rachelle Bienstock, Lisa Balbes, Mark Stave, Ivan Rossi, John McKelvey, and David Rigby. ============================================================================ The original question was: ACS NY/NJ section, Younger Chemists Committee is looking for a Computational Chemist who is in Consulting business. If anyone knows (or is) such a person, we'd really like to hear from you! ============================================================================ The replies: From: Wibke Sudholt Look at http://www.cosmologic.de, it's some kind of small "Computational Chemistry Consulting" company. From: Bienstock.Rachelle Here are a couple of Computational Chemists with Consulting Businesses that I know of: Dr. Rebecca Rone Rone Biotechnology Consulting Rebecca Rone [rone@mcs.net] David N. Haney, Ph.D. Haney Associates 5455 Westknoll Dr. La Jolla, CA 92037 619-483-1197 Fax: 619-483-1197 EMail: haney@hbond.com Dr. Lisa Balbes webmaster@balbes.com Osiris Consultants 648 Simmons Avenue Kirkwood MO 63122 USA. From: "Lisa M. Balbes, Ph.D." What are you looking for them to do? I might be able to help you out. From: Mark Stave You might consider contacting Ernie Chamot who I believe has worked in the consulting business as a computational chemist for a number of years. Ernest Chamot Chamot Laboratories, Inc. 530 E. Hillside Rd. Naperville, Illinois 60540 (630)637-1559 echamot@chamotlabs.com http://www.chamotlabs.com/cl From: Ivan Rossi I am ... an Italian CompChem consultant Dr. Ivan Rossi - PNACoS Consulting & CIRB Biocomputing Unit PNACoS, Via della Grada 4/F, I-40122 Bologna, ITALY e-mail: ivan@biocomp.unibo.it Web: http://www.biocomp.unibo.it/ivan From: John McKelvey I am a consultant in computational chemistry. I retired from Eastman Kodak after 22 years, during which I was largely responsible for guiding the research labs there in chemical modeling. I was a beta tester for several popular MO codes, such as MOPAC and JAGUAR. I am a principal consultant for SERENA Software, and am co-author of their commercial MMX molecular mechanics code found in PCMODEL. I was responsible for having a principal author of Gaussian, Mike Frisch, at Kodak. Gaussian88 was a Kodak release. My current work is focused in the area of UV-VIS modeling in the pharmaceutical industry. I have given numerous seminars on Applications of Molecular Orbital Theory in Industry in universities, at national ACS meetings, Gordon Conferences [one of which I served as Chair: Computational Chemistry], and at the Sanibel Conference. Regards, John McKelvey, PhD McKelvey Computational Chemistry 861 Bennett Rd Carmel, IN 46032 Phone 317-815-9048 e-mail: JMMcKel@attglobal.net ============================================================================ __ ___________ Artem.Masunov@usa.net;tel 212-650-7792;fax877-280-5146 / \ / __ __ \ Ph.D., Research Associate, Computational Biophys.Chem. / \/\ \ \ \ \ \ Chemistry Department, City College, City U. of NY / /\ \ \ \ \ \ \ \ J-1325, Convent Ave at 138 Str, New York, NY 10038 / ____ \ \ \ \ \ \ \ --------------------------------------------------- /__/\ _/\ _\ \ _\ \ _\ \ _\ FORGIVE MY NONSENSE AS I ALSO FORGIVE THOSE \ _\/ \/__/\ __/\ __/\ __/ WHO THINK THEY TALK SENSE -- ROBERT FROST From chemistry-request@server.ccl.net Wed Jun 28 22:51:29 2000 Received: from shiva.hunter.cuny.edu (shiva.hunter.cuny.edu [146.95.128.96]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id WAA29640 for ; Wed, 28 Jun 2000 22:51:29 -0400 Received: from shiva (shiva [146.95.128.96]) by shiva.hunter.cuny.edu (8.9.3/8.9.3) with ESMTP id WAA27688 for ; Wed, 28 Jun 2000 22:51:34 -0400 (EDT) Date: Wed, 28 Jun 2000 22:51:33 -0400 (EDT) From: AM To: chemistry@CCL.NET Subject: water/cyclohexane partition database Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear all, We need data on solvation free energies for as many organic compounds as possible, especially for non-polar solvent. Does anyone know of such a database (preferably, on-line, preferably, free)? I will summarize. Regards, Artem Masunov From chemistry-request@server.ccl.net Thu Jun 29 03:18:30 2000 Received: from uvaix7e1.comp.UVic.CA (root@uvaix7e1.comp.UVic.CA [142.104.5.107]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id DAA30936 for ; Thu, 29 Jun 2000 03:18:30 -0400 Received: from p17-109.dialup.UVic.CA (p17-109.dialup.UVic.CA [142.104.17.109]) by uvaix7e1.comp.UVic.CA (8.9.3/8.9.3) with SMTP id AAA143874 for ; Thu, 29 Jun 2000 00:18:18 -0700 From: Roy Jensen To: chemistry@ccl.net Subject: SUMMARY: CCL:Help on transition metal--chlorine optimization Date: Thu, 29 Jun 2000 00:18:20 -0700 Message-ID: <4otlls8h8en6t3mio9j1c2b7i766elsji6@4ax.com> References: In-Reply-To: X-Mailer: Forte Agent 1.8/32.548 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by server.ccl.net id DAA30937 Thanks for the overwhelming assistance. The responses centered around two areas. - modify the scf keyword with 'vshift' and 'qc' - start with a lower level of theory to generate a better initial wavefunction Gaussian, Inc. replied that G98W A.9 worked with the input file and maxcyc=100. Time to get the update... Below is a summary of the emails. Preliminary results look good, I will post an input file that works well once I have done more tests. Roy Jensen ORIGINAL EMAIL ============= >I posted a message to the CCL a few weeks ago; a summary was just >posted. DFT was the choice for geometry optimizations on transition >metal--chlorine clusters (neutral and ionic). > >I am using Gaussian 94W and 98W however I cannot get the simplest job >to run. I have tried several methods and various basis sets. Almost >all fail to converge on the first step. A sample input and output file >are given below. > >Thanks, >Roy Jensen > > >INPUT FILE >========= ># UB3LYP LANL2DZ opt test > >Optimized geometry for FeCl. > >0 2 >Fe >Cl 1 R1 > >R1=2.19 > > >OUTPUT FILE >=========== >>... > of initial guess= .7500 > Requested convergence on RMS density matrix=1.00E-08 within 64 >cycles. > Requested convergence on MAX density matrix=1.00E-06. > Keep R1 and R2 integrals in memory in canonical form, NReq= >727936. > Integral accuracy reduced to 1.0E-05 until final iterations. > Problem detected with inexpensive integrals. > Switching to full accuracy and repeating last cycle. > >>>>>>>>>> Convergence criterion not met. > SCF Done: E(UB+HF-LYP) = -137.386151674 A.U. after 65 cycles > Convg = .4226E-01 -V/T = 2.6548 > S**2 = .8263 > Annihilation of the first spin contaminant: > S**2 before annihilation .8263, after .7500 > Convergence failure -- run terminated. > Error termination via Lnk1e in D:\G94W\l502.exe. > Job cpu time: 0 days 0 hours 11 minutes 38.0 seconds. > File lengths (MBytes): RWF= 5 Int= 0 D2E= 0 Chk= 1 Scr= >1 > RESPONSES ========== "Li Zhenhua" >use SCF=QC but it is very slow. Or you try vshift=400 scfcyc=300 or so. Shahar Keinan >I've done some calculation on [CuCl4]2-, but I've used jaguar, this is because >they have a speciall method for creating initial wave function whcih is better. > >I did this after receiving this email from > "Help at Schrodinger, Inc" : > >>First, I don't really think that 3-21G is an accurate enough basis set >for your system. You will probably actually have better convergence >and better accuracy, without much difference in run time, if you use a >LACVP based basis set. > >Second, both the numerical accuracy of DFT and additional basis >functions (especially diffuse bases) can make SCF convergence more >difficult. So you should try to first converge the SCF wavefunction >with HF LACVP. When that wavefunction converges, you can use the >wavefunction in the restart file as a starting guess for, for >instance, a HF LACVP* calculation. When that converges, restart for >DFT LACVP*, then DFT LACVP*+. [Only after you've successfully >converged the DFT LACVP*+ wavefunction, turn on geometry optimization >in the restart file]. > >This step-by-step process is much more likely to succeed. It also >makes it easier to see which aspect of the calculation (orbital >occupations vs. DFT grids vs. basis set) is causing the convergence >difficulties, which in turn makes it easier to decide which added >flags are expected to help the most: choosing a different convergence >scheme, virtual orbital level shifting, or finer DFT grids. > >The use of and flags for choosing convergence schemes, virtual >shifting, and DFT grids for hard to converge cases are described in >the manual, especially on pp. 124-6. > >Jaguar 3.5 also provides another unique feature to improve SCF >convergence (both HF and DFT) for some TM systems. We realized that >most poor convergence of these systems was due to a really poor >initial guess wavefunction (both orbital shapes and occupations). So >we developed an algorithm based upon ligand-field theory that creates >a high-quality initial guess specifically designed for organometallic >systems. The success rate of this initial guess is phenomenal >compared with the Huckel guess, that is used in Gaussian, for >instance. Use of this feature in Jaguar 3.5 is also described on >pp.124-6 of the User's Manual. The method and some results are >published in: G. Vacek, J.K. Perry and J.-M. Lanlois, >Chem. Phys. Lett., 310 (1999) pp. 189-94. Alexander Hofmann >have you tried > > scf=(vshift=50) or another value to shift energy levels around HOMO or > guess=(alter) to switch symmetry or > scf=(qc) which is another convergency procedure > >in combination with increased maxcyc. Cory Pye >What's happening is that your SCF is not converging. To make it converge, you >must either improve the initial guess orbitals or change the convergence >method. I tried using an HF/STO-3G guess and it works great! > >----------------------------------------- >%chk=FeCl >%mem=8000000 >#N hf/sto-3g SCF=Direct > >FeCl > >0 2 >Fe >Cl Fe ClFe > >ClFe = 2.19 > >--Link1-- >%chk=FeCl >%mem=8000000 >#N ub3lyp LANL2DZ guess=read opt SCF=Direct > >FeCl > >0 2 >Fe >Cl Fe ClFe > >ClFe = 2.19 > >------------------------------ > >This gives me, in the end, ... > > 1\1\GINC-CYGNUS\FOpt\UB3LYP\LANL2DZ\Cl1Fe1(2)\CPYE\27-Jun-2000\0\\#N U > B3LYP LANL2DZ GUESS=READ OPT SCF=DIRECT\\FeCl\\0,2\Fe,0.,0.,-1.0607924 > 833\Cl,0.,0.,1.6223885039\\Version=Sun64-SVR4-Unix-G98RevA.9\HF=-138.0 > 816757\S2=0.75781\S2-1=0.\S2A=0.750006\RMSD=9.406e-09\RMSF=7.930e-06\D > ipole=0.,0.,0.3491222\PG=C*V [C*(Cl1Fe1)]\\@ Darko Babic > I have tinkered with your input and finally got around the >SCF-convergence problem, but you might still have a lot of diffi- >culties with your real tasks. You should try QC instead of the >default SCF algorithm in Gaussian. It takes much more time, but >it usually solves the problem. Otherwise you might try VShift >option (with caution). I tried a different route (which I pre- >fer): first, I did common ROHF instead of DFT with the intention >of using the ROHF wavefunction as a guess in DFT calculation. It >didn't help, so I decreased the number of electrons by 2 (put >the charge 2 instead of 0). This worked. The resulting wave- >function was used as a guess for the charge 0 case and the geo- >metry was optimized. Finally, this geometry was reoptimized >by using DFT with ROHF geometry and wavefunction as the initial >guesses. The combined input looked as: > >--------------start---------------------------------- >%Chk=fecl >#P ROHF LANL2DZ IOp(5/7=100) pop=none > >FeCl - first stage > >2 2 >Fe >Cl 1 R1 > >R1=2.19 > >--Link1-- >%Chk=fecl >#P ROHF LANL2DZ IOp(5/7=100) Geom=checkpoint Guess=read opt pop=none > >FeCl - second stage > >0 2 > >--Link1-- >%Chk=fecl >#P UB3LYP LANL2DZ IOp(5/7=100) Geom=checkpoint Guess=read opt > >FeCl - the final stage > >0 2 > > >-------------end------------------------------------- > >Yes, at the beginning I had to increase the maximum number of >cycles to 100 and I kept it at later stages too (but I didn't >check if that was necessary). I used G94. > > Generally, it's a good practice in these situations to >decrease the computation level by switching to common SCF (if >DFT is wanted) and to smaller basis sets, and then to use the >result as the guess in the actual job. Also, it's good to use >#P option as it will show you the course of convergence and >help you to decide whether increasing the maximum number of >cycles is appropriate or not. "Stefan Fau" >open shell clusters like the one you describe are a problem. If you >want to do it good, you have to use some sort of multireference >method. Approximate solutions can be obtained by DFT, since the >current implementations include correlation effects in some way. > >The solution to your SCF problem might be a different SCF algorithm, >e.g. quadratic convergence (QC) or (scaled) steepest descent (SSD, >SD). >If your problems persist, do a single point calculation with a minimal >basis set and use the wavefunction as an initial guess in calculations >with your desired basis set. Don't forget to check other possible >wavefunctions. You can use guess=alter to set SCF on the right track. "Shobe, Dave" >Some tricks to aid convergence: > >1. Use a single-zeta basis set first, then use guess=read to import this >into the double-zeta calculation. This simplifies the "vector space" in >which the wave function has to be found. > >2. Sometimes using a shorter than equilibrium bond length as the starting >geometry helps. This produces a stronger interaction between the atoms. > >3. Once you have a decent wavefunction, use vshift to prevent it from >drifting astray. You might also want to make a copy of the .chk file once >you have a good wavefunction, so that you can go back to it later. > >4. Even with the above tricks, 64 cycles is usually far too few. Maxcyc=500 >is more like it. > >One more thing: it's very easy to end up with a low-lying excited state with >this kind of molecule. If you know where the electrons are "supposed" to >be, you can look at the population analysis (you may need pop=full, which I >should warn you generates tons of output) to see if the right orbitals are >occupied. Also, the stable command can check for some kinds of spurious >states. "Tapas Kar" >You may try the following: >1.(a) vshift=500 or more if energy is oscillating, or (b) scf=qc >2. generate the initial guess using UHF calculation and start dft from that >guess. Art >My experience with complexes of metals is not big, but you can try to >reduce or make better the quality of integration grid (via keyword >INT). Also you can modify basis set to expand orbitals. That could be >reasonamble in case of Fe. barbosa@anne.chemie.unibas.ch (Frederique Barbosa) >I already had such problem during the optimization of radicals. >You can bypass it by using the IOP(5/13=1). This iop will allow >the optimization to continue even if the scf did not converge >after 65 cycles. But it will not give more cycles to converge >the scf so you have to be aware of this. John Bushnell > Regarding your problems getting FeCl calcs to converge: >You definitely need to look at the guess that Gaussian starts >with. It is not obvious (to me anyway) what the most likely >iron configuration would be in ground state FeCl. In fact, >it's not even obvious that it is a doublet. Iron is known >as a difficult case, and you may have to try altering the >default guess several times to get any state to converge. >Once you have a converged state it is usually much easier >to think about how to switch things around to verify that >you have found the ground state. That being said, I would >add a couple more tricks to the ones people have suggested >already on CCL: > > > Try an ROHF calculation. Sometimes it takes longer to >converge, but sometimes it will clean the wavefunction up >enough to work with it. The spin contamination is a bit >large in your sample output, even perhaps for an unconverged >function. > > Definitely don't try an optimization right off the bat >(as your input file suggests). Try for a single point >calculation with (default) loose SCF convergence. You're >looking for ANYTHING that looks reasonable to start. As >suggested, always backup the checkpoint files of converged >calculations for further tinkering -- guess=(read,alter). >Nothing makes fellow computer users crankier than running >full blown opt/freq calculations with large basis sets >before looking at what you're doing either. ;-) > > Maybe try a quartet state. > > Sometimes you can 'cheat' and add or subtract a few electrons to >the system in order to reduce the chaos created by all those >low lying states, or alternatively use a neighboring metal atom >to get a better first guess for iron. Be creative! :-) > > >Whatever you do, you can rarely trust an ab-initio calculation >on transition metal systems out-of-the-box. Be careful, examine >the atomic state tables (CE Moore) and try any possibilities you >can think of...good luck! From chemistry-request@server.ccl.net Thu Jun 29 07:01:51 2000 Received: from nets5.rz.rwth-aachen.de (nets5.rz.RWTH-Aachen.DE [137.226.144.13]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id HAA32750 for ; Thu, 29 Jun 2000 07:01:51 -0400 Received: from ac.rwth-aachen.de (IDENT:krzys@krzys.dorf.RWTH-Aachen.DE [134.130.59.215]) by nets5.rz.rwth-aachen.de (8.10.1/8.10.1/5) with ESMTP id e5TB1e323958 for ; Thu, 29 Jun 2000 13:01:41 +0200 (MET DST) Sender: krzys@nets5.rz.rwth-aachen.de Message-ID: <395B2CA1.2ED181CE@ac.rwth-aachen.de> Date: Thu, 29 Jun 2000 13:01:53 +0200 From: Krzysztof Radacki Organization: RWTH Aachen X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0smp i686) X-Accept-Language: en MIME-Version: 1.0 To: CCL Subject: g98 error Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, can some body explain me what is the reason for following error in g98.a7 (and preferably how can I omit it): Cycle 17 Pass 1 IDiag 1: RMSU= 1.58D-07 CP: 9.96D-01 4.39D-02 7.82D-01 2.00D-01 1.05D+00 CP: 2.63D+00 2.77D+00 2.38D+00 2.54D+00 2.62D+00 CP: 2.52D+00 2.56D+00 2.44D+00 2.71D+00 2.63D+00 CP: 1.97D+00 Process number 1 did not complete successfully. Error termination via Lnk1e in /usr/local/g98.a7/l502.exe. Job cpu time: 0 days 0 hours 23 minutes 44.5 seconds. File lengths (MBytes): RWF= 44 Int= 0 D2E= 0 Chk= 15 Scr= 1 regards Krzys Radacki From chemistry-request@server.ccl.net Thu Jun 29 10:42:00 2000 Received: from linus.imcm.cas.cz (Linus.imcm.cas.cz [147.231.77.250]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id KAA00927 for ; Thu, 29 Jun 2000 10:42:00 -0400 Received: from thales (thales.imcm.cas.cz [147.231.77.130]) by linus.imcm.cas.cz (8.9.3/8.9.3) with ESMTP id QAA03481 for ; Thu, 29 Jun 2000 16:41:32 +0200 Message-Id: <200006291441.QAA03481@linus.imcm.cas.cz> From: "Petr Toman" To: chemistry@ccl.net Date: Thu, 29 Jun 2000 16:42:05 +0200 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Subject: Overlap between MO of 2 molecules Reply-to: toman@imc.cas.cz Priority: normal X-mailer: Pegasus Mail for Win32 (v3.12a) Dear colleagues, Could somebody give me an advice how could I calculate the overlap between molecular orbitals of 2 neighbouring molecules (preferably by Gaussian)? 1) Would it be correct to calculate firstly the MO of each molecule by itself to obtain the molecular orbital coefficients C_mi, then perform the calculation of both molecules together to obtain the overlap matrix (between AO) S_mn. Finally compute the overlap between i-th MO of the first molecule and j-th MO of the second molecule as SUM (C_mi * C_nj * S_mn) over m,n where m are the AO of the 1. molecule and n are the AO of the 2. molecule. 2) By the way, how could I force Gaussian 98 to print overlap matrix ??? 3) What method could be reliable (B3LYP, HF,...)? Thanks a lot, Petr From chemistry-request@server.ccl.net Thu Jun 29 11:04:22 2000 Received: from Mercury.acs.unt.edu (mercury.acs.unt.edu [129.120.220.1]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id LAA01057 for ; Thu, 29 Jun 2000 11:04:22 -0400 Received: from jove.acs.unt.edu (10759@jove.acs.unt.edu [129.120.220.41]) by Mercury.acs.unt.edu (8.8.8/8.8.8) with ESMTP id KAA29577 for ; Thu, 29 Jun 2000 10:04:08 -0500 (CDT) Received: from localhost (tdp0006@localhost) by jove.acs.unt.edu (8.8.8/8.8.8) with ESMTP id KAA13207 for ; Thu, 29 Jun 2000 10:04:14 -0500 (CDT) Date: Thu, 29 Jun 2000 10:04:14 -0500 (CDT) From: TREVOR D POWER To: chemistry@ccl.net Subject: Generating Enantiomers In-Reply-To: <395A5426.94A0BAFA@npd.ufpe.br> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hey Guys, With computational methods, my student has independently located the enantiomer of a complex that I have the crystal structure for. In order to compare these two, I need to generate the precise enantiomer of one of these. Is there a simple way of doing this with the cartesian coordinates? Many thanks for your assistance, David Power University of North Texas Department of Chemistry NT Station, Box 305070 Denton, TX 76203-5070 tdp0006@unt.edu From chemistry-request@server.ccl.net Thu Jun 29 12:37:28 2000 Received: from havana.ks.uiuc.edu (havana.ks.uiuc.edu [130.126.120.73]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id MAA01666 for ; Thu, 29 Jun 2000 12:37:28 -0400 Received: from geneseo.ks.uiuc.edu (geneseo.ks.uiuc.edu [130.126.120.127]) by havana.ks.uiuc.edu (8.9.1/8.9.1) with ESMTP id LAA25906 for ; Thu, 29 Jun 2000 11:37:15 -0500 (CDT) Received: (from johns@localhost) by geneseo.ks.uiuc.edu (8.9.1b+Sun/8.9.1) id LAA20553 for chemistry@www.ccl.net; Thu, 29 Jun 2000 11:37:14 -0500 (CDT) Date: Thu, 29 Jun 2000 11:37:13 -0500 From: John Stone To: chemistry@server.ccl.net Subject: Announce: VMD 1.5 Released Message-ID: <20000629113713.A20535@geneseo.ks.uiuc.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0i VMD "Visual Molecular Dynamics" 1.5 Announcement ------------------------------------------------ The Theoretical Biophysics group at the Beckman Institute For Advanced Science and Technology, the University of Illinois (U-C), is proud to announce the public release of VMD 1.5. VMD is a package for the visualization and analysis of biomolecular systems. This software is distributed free of charge and includes source code, documentation, and precompiled binaries for Compaq, IBM, HP, Linux, Sun, and SGI systems systems running Unix, as well as Microsoft Windows 95/98/NT/2000. The VMD documentation includes an installation guide, a users guide, and a programmers guide for interested researchers. VMD also provides on-line help through the use of an external HTML viewer. VMD development is supported by the NIH National Center for Research Resources. A full description of VMD is available via the VMD WWW home page: http://www.ks.uiuc.edu/Research/vmd/ We are eager to hear from you, and thank you for using our software! John Stone vmd@ks.uiuc.edu June, 2000 README file for VMD 1.5 --------------------------------------------------------------------------- What is VMD? See also http://www.ks.uiuc.edu/Research/vmd/ --------------------------------------------------------------------- VMD is designed for the visualization and analysis of biological systems such as proteins, nucleic acids, lipid bilayer assemblies, etc. It may be used to view more general molecules, as VMD can read standard Protein Data Bank (PDB) files and display the contained structure. VMD provides a wide variety of methods for rendering and coloring a molecule: simple points and lines, CPK spheres and cylinders, licorice bonds, backbone tubes and ribbons, and others. VMD can be used to animate and analyze the trajectory of a molecular dynamics (MD) simulation. In particular, VMD can act as a graphical front end for an external MD program by displaying and animating a molecule undergoing simulation on a remote computer. The program has many features, which include: o No limit on the number of molecules, atoms, residues or number of animation frames, except available memory. o Many molecular rendering and coloring methods. o Stereoscopic display capability. o Extensive atom selection syntax for choosing subsets of atoms for display (includes boolean operators, regular expressions, and more). o Integration with the program 'Babel' which allows VMD to read many molecular data file formats. Even without the use of Babel, VMD can read PDB files, as well as CHARMM- and X-PLOR compatible binary DCD files and X-PLOR compatible PSF files. o Ability to write the current image to a file which may be processed by a number of popular raytracing and image rendering packages, including POV-Ray, Radiance, Raster3D, Rayshade, and Tachyon. o Extensive graphical and text-based user interfaces, which use the Tcl package to provide full scripting capabilities. o Extensions to the Tcl language which enable researchers to write their own routines for molecular analysis. o Modular, extensible source code using an object-oriented design in C++, with a programmer's guide outlining the source code structure. o Integration with the program NAMD, a fast, parallel, and scalable molecular dynamics program developed in conjunction with VMD in the Theoretical Biophysics Group at the University of Illinois. See the NAMD WWW home page for more info: http://www.ks.uiuc.edu/Research/namd VMD can be used to interactively display and control an MD simulation using NAMD. What's new in VMD 1.5? ---------------------- New Features o Support for several Gromacs structure and trajectory file formats: - Can read .gro for structure (concatenated multiple-frame .gro files are read automatically) (variable precision ASCII format) - Can read .g96 for structure (concatenated multiple-frame .g96 files are read automatically) (fixed high precision ASCII format) - Can read .trr for trajectory (full precision, portable binary format) - Can read .xtc for trajectory (variable precision, compressed binary format) Note: Doesn't support .trj files, or combined run-input files yet, but Gromacs users can run trjconv to convert these into one of the formats that VMD can read. o Windows versions of VMD now write their snapshot images as 24-bit color Windows Bitmaps (.bmp) files directly. o Added the ability to modify the name, type, resname, resid chain, and segname of a selection of atoms. o Entirely new rendering capabilities implementing "material" properties, for user control of ambient, diffuse, and specular reflectivity, as well as surface shininess. General Improvements and Bug Fixes o Improved the rendering output for Raster3D and Tachyon o General improvements to all renderer output. o Major revision of IMD (Interactive MD) features in VMD. o The Unix versions of VMD are now built with FLTK instead of XForms, this brings both the Unix and Windows versions of VMD to using exactly the same GUI, and decreases VMD's code/binary size noticably. o New OpenGL code to allow SGI O2s to support stereoscopic display with 16-bit depth buffering and and 12-bit color. o Fixed quite a few of the outstanding problem reports (bugs) that were sent in on previous versions of VMD. PRs fixed: 5, 7, 10, 19, 23, 29, 58, 64, 68, 75, 76, 78, 79, 81, 83, 87, 88, 89, 90, 94, 95, 96, 97, 98, 100, 104, 105 o Rewrote large parts of the VMD display command system, yielding some significant performance increases. o Improvements to VMD's built-in help for various commands, better help topic listings. User Interface Changes o New GUI color scheme, using lighter colors which are easier on most people's eyes, particularly with laptops, projectors, etc. o New "Mouse" form which supercedes the old "popup menu" found in all of the previous Unix versions of VMD. The Mouse menu is identical on both Windows and Unix, unlike the previous versions of VMD where the Windows version had no analog to the popup menu. o New "Materials" form which provides controls over rendering properties for representations. (controls for shininess, diffuse, ambient and specular reflectivity etc) o Renamed the "Mol" and "Sim" forms to their longer names for increased intuitiveness to new users. o New "Cancel" button on the Molecule form providing the ability to cancel loading of very large trajectory files. User Documentation Updates o Improved HTML version of documentation (new rev of latex2html) o Improved PDF version of documentation, now uses better scalable fonts o Updated the VMD tutorial to cover new GUI features o The user guide does more to cover issues specific to the Windows versions of VMD in better detail. o Updated docs on rendering to snapshots and to external renderers o Many updates to sections explaining the VMD GUI, particularly the new GUI features. Known bugs ---------- Please visit the VMD web site for information on known bugs, workarounds, and fixes: http://www.ks.uiuc.edu/Research/vmd/ Cost and Availability --------------------- VMD, NAMD, and BioCoRE represent the broad efforts of the Theoretical Biophysics group, an NIH Resource for Macromolecular Modeling and Bioinformatics, designed to develop and distribute free, effective tools (with source code) for molecular dynamics studies in structural biology. For more information, see: http://www.ks.uiuc.edu/Research/VMD/ http://www.ks.uiuc.edu/Research/NAMD/ http://www.ks.uiuc.edu/Research/biocore/ The VMD project is funded by the National Institutes of Health (grant number PHS 5 P41 RR05969). Disclaimer and Copyright ------------------------ VMD is Copyright (c) 1995-2000 the Board of Trustees of the University of Illinois and others. The terms for using, copying, modifying, and distributing VMD are specified in the file LICENSE. If you use VMD in a way you think is interesting or novel, we would like to know about it. The authors request that any published work which utilizes VMD includes a reference to the VMD web page: http://www.ks.uiuc.edu/Research/vmd/ and/or the following reference: Humphrey, W., Dalke, A. and Schulten, K., "VMD - Visual Molecular Dynamics", J. Molec. Graphics, 1996, vol. 14, pp. 33-38. Documentation ------------- Three VMD manuals are available which describe how to install, use, and modify VMD. The VMD installation guide, is contained in the VMD distribution in the file "doc/ig.ps". The User's Guide and Programmer's Guide are available separately (due to size) from the VMD web site. Quick help may be accessed by pressing the "Help" button on the main VMD form, or by typing help in the VMD command window. This will bring up the VMD quick help page, and will lead you to several other VMD help files and manuals. Quick Installation Instructions ------------------------------- Detailed instructions for compiling this version of VMD can be found in the installation guide, ig.ps. For quick installation of the binary distribution for Unix do the following: 1) uncompress and untar the distribution into a working directory. In this working directory, there are several subdirectories such as bin, src, doc, data, as well as this README and a configure script. Change to this working directory after the unpacking is complete. 2) Edit the file 'configure'; change the values for the $install_library_dir and $install_bin_dir to a directory in which vmd data files and executables should be installed: $install_bin_dir is the location of the startup script 'vmd'. It should be located in the path of users interested in running VMD. $install_library_dir is the location of all other VMD files. This included the binary and helper scripts. It should not be in the path. 3) A Makefile must be generated based on these configuration variables by running "./configure". 4) After configuration is complete, cd to the src directory, and type "make install". This will install VMD in the two directories listed above. Note that running "make install" twice will print error messages because you are attempting to overwrite some read-only files. This should be fine. 5) When installed, type 'vmd' to start (make sure the $install_bin_dir directory is in your path). Required Libraries ------------------ VMD requires several libraries and programs for various of its functions. In particular, it uses GL or OpenGL based 3-D rendering, and will require that you have the appropriate GL or OpenGL libraries on your system. Other programs are required by some of VMD's optional features. Please visit the VMD web site for more information: http://www.ks.uiuc.edu/Research/vmd/ For problems, questions, or suggestions, send e-mail to 'vmd@ks.uiuc.edu'. VMD Development Team Theoretical Biophysics Group University of Illinois and Beckman Institute 405 N. Matthews Urbana, IL 61801 TBG: http://www.ks.uiuc.edu/ VMD: http://www.ks.uiuc.edu/Research/vmd/ README for VMD; last modified June 28, 2000 by John Stone -- Theoretical Biophysics Group Email: johns@ks.uiuc.edu Beckman Institute http://www.ks.uiuc.edu/~johns/ University of Illinois Phone: (217) 244-3349 405 N. Mathews Ave FAX: (217) 244-6078 Urbana, IL 61801, USA Unix Is Good For You!!! From chemistry-request@server.ccl.net Wed Jun 28 23:34:21 2000 Received: from tomts2-srv.bellnexxia.net (tomts2.bellnexxia.net [209.226.175.140]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id XAA29791 for ; Wed, 28 Jun 2000 23:34:21 -0400 Received: from atoma.f2s.com ([206.172.253.32]) by tomts2-srv.bellnexxia.net (InterMail vM.4.01.03.00 201-229-121) with ESMTP id <20000629033408.KRYK22603.tomts2-srv.bellnexxia.net@atoma.f2s.com>; Wed, 28 Jun 2000 23:34:08 -0400 Message-ID: <395AC357.38805999@atoma.f2s.com> Date: Wed, 28 Jun 2000 23:32:39 -0400 From: Jonathan Desp Organization: Atoma ______________________________________ http://www.atoma.f2s.com X-Mailer: Mozilla 4.73 [en] (Win95; I) X-Accept-Language: en MIME-Version: 1.0 To: chemistry@ccl.net Subject: An IBM supercomputer designed for simulating nuclear explosions has turned out to Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit An IBM supercomputer designed for simulating nuclear explosions has turned out to be 23 percent faster than anticipated when the project began, IBM will announce tomorrow. The $110 million machine, called ASCI White, was assembled at IBM's test facility in Poughkeepsie, N.Y., and is being reassembled at its final home, Lawrence Livermore National Laboratory (LLNL) in Livermore, Calif. The machine takes up two basketball courts worth of floor space, weighs 106 tons, and has 8,192 CPUs. http://news.cnet.com/news/0-1003-200-2167700.html?tag=st.ne.1002.thed.ni -- Very truly yours, <><><><><><><><><><><><><><> Jonathan Desp Atoma Matter will become Software http://www.atoma.f2s.com <><><><><><><><><><><><><><> From chemistry-request@server.ccl.net Thu Jun 29 00:32:24 2000 Received: from linux2.ipc.pku.edu.cn (mao@linux2.ipc.pku.edu.cn [162.105.177.40]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id AAA30120 for ; Thu, 29 Jun 2000 00:32:23 -0400 Received: from localhost (mao@localhost) by linux2.ipc.pku.edu.cn (8.9.3/8.9.3) with ESMTP id MAA20859 for ; Thu, 29 Jun 2000 12:38:44 +0800 Date: Thu, 29 Jun 2000 12:38:44 +0800 (CST) From: Fenglou Mao To: "CHEMISTRY@www.ccl.net" Subject: About molsee. Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hi, all, If you did not download molsee, this mail is no use for you, just delete it. IF you downloaded molsee, and you are using a computer diffrent from SGI IRIX and Linux, and the buttons and text position and size looks not so good in your screen, please go to http://mdl.ipc.pku.edu.cn/software/molsee.html to download modified version for SUN and DEC. Sincerely Yours, FengLou Mao ******************************* ADD:Mr. FengLou Mao Institute of Physical Chemistry Peking University BeiJing P.R.China Tel:86-10-62756833 Fax:86-10-62751725 From chemistry-request@server.ccl.net Thu Jun 29 09:34:29 2000 Received: from cns.FTA-Berlin.de (root@cns.FTA-Berlin.de [141.16.244.4]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id JAA00710 for ; Thu, 29 Jun 2000 09:34:28 -0400 Received: from cs1.FTA-Berlin.de (root@cs1.FTA-Berlin.de [194.25.98.134]) by cns.FTA-Berlin.de (8.9.3/8.9.3) with ESMTP id PAA16959 for ; Thu, 29 Jun 2000 15:33:08 +0200 Received: from langmuir.aca-berlin.de (hofmann@langmuir.ACA-Berlin.de [141.16.16.120]) by cs1.FTA-Berlin.de (8.9.3/8.9.3) with ESMTP id PAA13118 for ; Thu, 29 Jun 2000 15:34:03 +0200 Date: Thu, 29 Jun 2000 15:33:51 +0200 (CEST) From: Alexander Hofmann To: CHEMISTRY@ccl.net Subject: Summary: metallic alloy database Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hello together, some weeks ago I've posted a question concerning bandstructures of metallic alloys: > > I'm searching for a database containing electronic properties of alloys > like bandstructure and denstiy of states plots. > I know Landolt-Boernstein, but the content is not extensive enough. > > Both electronic or printed version would be great. The answers: 1. Books Moruzzi, V.L., Sommers, C.B. Calculated electronic properties of ordered alloys World Scientific 1995 ISBN: 981-02-1918-0 Moruzzi, V.L. , Janak, J. F., Williams A.R. Calculated electronic properties of metals Pergamon 1978 ISBN: 0-08-022705-8 (This one is part of the newer one) 2. Electronic nothing ready to use in the sense of the question, _but_ there are companies like SciCo USA (http://www.SciCoUSA.com/) and MaterialsDesign France (http://www.materialsdesign.com/) which are developing and selling software solving this question. There is one structural database (CRYSTMET, http://www.tothcanada.com/toth/crystmet/overview.html) within this software, which provides the necessary geometrical information for calculating bands. You can choose between different physical methods to simulate the band structure. Of course you can take the CRYSTMET database, for example, and a desired computer program and start computation. One of these programs may be the TB-LMTO-ASA-Code suggested by Dr. Jepsen. It is fast and precise enough for this purpose. Thanks to Dr. Hermann Schier (schier@and.mpi-stuttgart.mpg.de) and Dr. Ove Jepsen (jepsen@and.mpi-stuttgart.mpg.de) Max-Planck-Institut fuer Festkoerperforschung John Harris, SciCo USA and FZ-Juelich (J.Harris@fz-juelich.de) Dr. Alexander Mavromaras Materials Design amavromaras@materialsdesign.com Yours Alex --- Dr. Alexander Hofmann Institut fuer Angewandte Chemie Berlin-Adlershof e.V. Richard-Willstaetter-Str. 12 D-12489 Berlin hofmann@aca-berlin.de Tel.: 030/6392-4408 Fax.: 030/6392-4350 http://www.aca-berlin.de PGP-Private key: http://www.uni-leipzig.de/~quant/hofmann/alexander.hofmann.pubkey.asc From chemistry-request@server.ccl.net Thu Jun 29 13:41:39 2000 Received: from nano.chem.nwu.edu (nano.chem.nwu.edu [129.105.14.190]) by server.ccl.net (8.8.7/8.8.7) with SMTP id NAA02458 for ; Thu, 29 Jun 2000 13:41:39 -0400 Received: from nano.chem.nwu.edu ([129.105.14.190]) by nano.chem.nwu.edu (WESMTP 2.2 [Dec 1 1999]) with SMTP id 853008140; Thu, 29 Jun 2000 17:50:52 GMT Date: Thu, 29 Jun 2000 12:50:49 -0500 From: Greg Rechtsteiner Subject: Residual Forces Summary To: CHEMISTRY@ccl.net Message-Id: <4.3.2.7.2.20000629124629.00b06db8@nano.chem.nwu.edu> X-Sender: greg@nano.chem.nwu.edu X-Mailer: QUALCOMM Windows Eudora Version 4.3.2 Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="=====================_853007500==_.ALT" --=====================_853007500==_.ALT Content-Type: text/plain; charset="us-ascii"; format=flowed Hello: I would like to thank everyone for the prompt reponses to my inquiry concerning residual forces from a geometry optimization. A summary is included below: Original Question: > Hello: > > I have been performing geometry optimizations using Gaussian 94 and 98, > and I have the following question(s). > > An optimization is considered complete when it converges > according to the specified convergence criteria, in my case, > the default values for G9x are max. component of force 0.00045, > rms forces 0.0003, displacement 0.0018 and rms displacement 0.0012. > > I am interested in what may be learned from the values of the "residual" > force components, ie max. component of the force and rms forces > that are provided at the final step in the optimization process. > > For example, I have a structure with residual values of 0.000187 for the max. > force component and 0.000051 for the rms force. > > (Please note that no imaginary frequencies were found from a freq. analysis). > > Do these values tell me anything about the stability of the structure? > That is, do larger values (but still within the convergence criteria) suggest > a possible metastable state? Responses: 1. The force data tell you only that you are at or approaching a stationary point and nothing about the stability of the structure. Large RMS and MAX gradient values indicate only that you have not converged the stationary point to high precision. Only the Hessian eigenvalues indicate the curvature of the PES and therefore the (kinetic) stability of the given stationary geometry. NB, however, that for very flat potential surfaces (e.g., those often found for weakly-bound complexes), it may be very difficult to tightly converge the forces. This is a numerical problem, however, and not a general indicator of stability. -Daniel -- T. Daniel Crawford Department of Chemistry crawdad@vt.edu Virginia Polytechnic Institute and http://zopyros.ccqc.uga.edu/~crawdad/ State University 2. I don't think that the residual forces tell anything about the stability of the structure or - more precisely - about the depth of the potential minimum. They just describe the slopes of the potential surface in respect to geometry changes. A purpose for quoting them is only to indicate how far the geometry is from the ideal minimum at which all the forces are zero. The second derivatives are (in frame of the harmonic approximation) related to a depth of the minimum, but they are equally informative as the frequencies calculated from the second derivatives, which are (the frequencies) more common for esti- mating of what you call the stability. Sincerely, Darko Babic Institute "Rudjer Boskovic" HR-10002 Zagreb, P.O.B. 180 Croatia 3. In almost all cases I don't think they give any useful information: The figures just reflect the path that the optimization followed: not any intrinsic property of the molecule itself. The sole exception that I can think of is where the optimization stops with very small forces but large displacements thus indicating that the minima has low curvature that is is very flat and this may indicate that the molecule is quite soft and floppy All the best Larry Cuffe 4. The stability of a structure depends on the second derivative not the first. The residual force does indicate that the structure lies slightly off the point of zero force but except in the case of a very flat potential energy surface the small displacements suggest that the difference in structure is small compared to the other uncertainties. The FREQ calculation can tell you if the point is a maximum on the potential energy surface along some displacement, this should break the symmetry of the molecule if the prior optimization is properly done. It will also give you a better estimate of the displacement to find the minimum by improving the force constants and estimates of the displacement. Note that there can also be instabilities in the wavefunction at any point on the potential energy surface. Most often FREQ will also uncover instabilities in the wavefunction but not in a concrete fashion, rather if the wavefunction is unstable you get nonsense for frequencies. The STABLE or STABLE=OPT options are better if you have concerns here. Doug Fox (Gaussian Support) --=====================_853007500==_.ALT Content-Type: text/html; charset="us-ascii" Hello:

I would like to thank everyone for the prompt reponses to my
inquiry concerning residual forces from a geometry optimization.
A summary is included below:

Original Question:

> Hello:
>
> I have been performing geometry optimizations using Gaussian 94 and 98,
> and I have the following question(s).
>
> An optimization is considered complete when it converges
> according to the specified convergence criteria, in my case,
> the default values for G9x are max. component of force 0.00045,
> rms forces 0.0003, displacement 0.0018 and rms displacement 0.0012.
>
> I am interested in what may be learned from the values of the "residual"
> force components, ie max. component of the force and rms forces
> that are provided at the final step in the optimization process.
>
> For example, I have a structure with residual values of 0.000187 for the max.
> force component and 0.000051 for the rms force.
>
> (Please note that no imaginary frequencies were found from a freq. analysis).
>
> Do these values tell me anything about the stability of the structure?
> That is, do larger values (but still within the convergence criteria) suggest
> a possible metastable state?

Responses:

1.

The force data tell you only that you are at or approaching a
stationary point and nothing about the stability of the structure. Large
RMS and MAX gradient values indicate only that you have not converged the
stationary point to high precision. Only the Hessian eigenvalues indicate
the curvature of the PES and therefore the (kinetic) stability of the given
stationary geometry. NB, however, that for very flat potential surfaces
(e.g., those often found for weakly-bound complexes), it may be very
difficult to tightly converge the forces. This is a numerical problem,
however, and not a general indicator of stability.
-Daniel
--
T. Daniel Crawford Department of Chemistry
crawdad@vt.edu Virginia Polytechnic Institute and
http://zopyros.ccqc.uga.edu/~crawdad/ State University

2.  I don't think that the residual forces tell anything about the
stability of the structure or - more precisely - about the depth of
the potential minimum. They just describe the slopes of the potential
surface in respect to geometry changes. A purpose for quoting them
is only to indicate how far the geometry is from the ideal minimum at
which all the forces are zero. The second derivatives are (in frame
of the harmonic approximation) related to a depth of the minimum, but
they are equally informative as the frequencies calculated from the
second derivatives, which are (the frequencies) more common for esti-
mating of what you call the stability.

Sincerely,
Darko Babic

Institute "Rudjer Boskovic"
HR-10002 Zagreb, P.O.B. 180
Croatia

3.  In almost all cases I don't think they give any useful information:
The figures just reflect the path that the optimization followed: not
any intrinsic property of the molecule itself. The sole exception
that I can think of is where the optimization stops with very small
forces but large displacements thus indicating that the minima has
low curvature that is is very flat and this may indicate that the
molecule is quite soft and floppy

All the best Larry Cuffe

4. The stability of a structure depends on the second derivative not
the first. The residual force does indicate that the structure lies
slightly off the point of zero force but except in the case of a very
flat potential energy surface the small displacements suggest that
the difference in structure is small compared to the other uncertainties.

The FREQ calculation can tell you if the point is a maximum on the
potential energy surface along some displacement, this should break the
symmetry of the molecule if the prior optimization is properly done. It
will also give you a better estimate of the displacement to find the
minimum by improving the force constants and estimates of the displacement.

Note that there can also be instabilities in the wavefunction at any
point on the potential energy surface. Most often FREQ will also uncover
instabilities in the wavefunction but not in a concrete fashion, rather if
the wavefunction is unstable you get nonsense for frequencies. The STABLE
or STABLE=OPT options are better if you have concerns here.

Doug Fox (Gaussian Support)
--=====================_853007500==_.ALT-- From chemistry-request@server.ccl.net Thu Jun 29 18:55:14 2000 Received: from Mercury.acs.unt.edu (mercury.acs.unt.edu [129.120.220.1]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id SAA03684 for ; Thu, 29 Jun 2000 18:55:13 -0400 Received: from jove.acs.unt.edu (10759@jove.acs.unt.edu [129.120.220.41]) by Mercury.acs.unt.edu (8.8.8/8.8.8) with ESMTP id RAA15866 for ; Thu, 29 Jun 2000 17:55:11 -0500 (CDT) Received: from localhost (tdp0006@localhost) by jove.acs.unt.edu (8.8.8/8.8.8) with ESMTP id RAA28791 for ; Thu, 29 Jun 2000 17:55:18 -0500 (CDT) Date: Thu, 29 Jun 2000 17:55:18 -0500 (CDT) From: TREVOR D POWER To: chemistry@ccl.net Subject: Summary: Generating Enantiomers In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hi, Wow! I quickly received many responses to my question: > With computational methods, my student has independently located >the enantiomer of a complex that I have the crystal structure for. In >order to compare these two, I need to generate the precise enantiomer of >one of these. Is there a simple way of doing this with the cartesian >coordinates? I summarize the different answers below: 1) --- Given a set of cartesian coordinates for one of the enantiomers, all you have to do to obtain the other enantiomer is to keep two coordinates constant for all atoms (e.g. x and y), and change only the sign of the third coordinate (e.g. z). --- You could also change every cartesian coordinate to its negative (corresponding to inversion about the origin). The following people gave answers similar to this: Per-Ola Norrby Maricel Torrent Cory C. Pye William Johnson Darko Babic Keny Lipkowitz Larry Cuffe Igor Shanovsky Laurent Chiche Stefan Fau Yvonne Martin Bob Pearlman William T Winter 2) "Molecule" (http://www.ccc.uni-erlangen.de/molecule/) will do what you want. Cheers, Theis Soelling 3) To reflect in the XY plane change the sign of all the Z coordinates To reflect in the XZ plane change the sign of all the Y coordinates To reflect in the YZ plane change the sign of all the X coordinates To invert about the origin change the sign of ALL the coordinates Any good modelling package should provide these facilities. INTERCHEM does see: http://www.interprobe.co.uk Yours sincerely Peter Bladon 4) Depends on what you want to compare. If you want to compare whether they are exact enantiomers, you can do the following. (a) You can calculate the energies of both. Any method, MM, QM will suffice. If they have the same energy, chances are that they have the same "internal" structure. (b) Calculate the interatomic distances of both. If they have the same distance matrix, then they are either the same or they are enantiomers. (c) To "generate" an exact enantiomer, all you want to do is to invert one of the x, y, z coordinates of all atoms. After you have done this, you can repeat (a) and (b). regards. Daquan Gao ----------------------------------------- Thank you all very much, David Power University of North Texas Department of Chemistry NT Station, Box 305070 Denton, TX 76203-5070 tdp0006@unt.edu