From chemistry-request@ccl.net Thu Jul 23 09:42:25 1992 Date: 23 Jul 92 14:10:41-2300 From: Yves chapron To: chemistry@ccl.net Subject: genetic algoritm (Merz, Le Grand) Status: R Yves CHAPRON Biophysique moleculaire & cellulaire Grenoble 23 jul 92 C.E.N.G. 85 X 38041 GRENOBLE cedex FRANCE - Ph (33)76884212 Fx (33)76885487 E.M. chapron@drac.ceng.cea.fr I am looking for paper and software about "Genetic Algorithm" and more specificly with application to minimization of potential energy functions. I should like to know the E-mail of K. Merz, Jr.,or S. Le Grand. Thanks Yves --- Administrivia: The rest of this message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chemistry-request@ccl.net Thu Jul 23 18:38:42 1992 Date: Thu, 23 Jul 92 16:49:47 EDT From: slawek@rutile.rutgers.edu (A) To: chemistry@ccl.net Subject: van der Waals & CI Status: R I just found an interesting paper: "van der Waals functional forms for molecular simulations" by J.R. Hart and A.K. Rappe, J.Chem.Phys., Vol.97, No. 2, 15 July 1992, Page 1109. The authors show that, for the van der Waals interaction, the Morse potential offers better approximation than exponential-6 and Lennard-Jones functions. They write: "The long range attractive part of a van der Waals interaction EL is described in perturbation theory as a second order effect [here there is a reference to classical papers from 1930s - S.B.]. We obtain the distance dependence of this effect by considering a simple configuration interaction (CI) wave function." Is such an approach correct ? I know that CI is able to take into account correlations resulting from Coulomb interactions between electrons at short (bonding) distances. Can it be used to calculate SECOND ORDER effects at long (non-bonding) distances ? Slawomir Blonski slawek@rutile.rutgers.edu --- Administrivia: The rest of this message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chemistry-request@ccl.net Thu Jul 23 20:56:36 1992 Date: Thu, 23 Jul 92 15:44:01 -0700 From: avslists@mercury.aichem.arizona.edu (LISTS) To: chemistry@ccl.net Subject: Space Groups From Coordinates.... Status: R Is any one aware of programs that would derive space group of a molecule from its Cartesian coordinates? I realize that this is not a trivial task. If there is sufficient interest, I will summerize the responses and send it to the net. Thanks, Ajay Shah Chemistry Dept AI Chemistry Lab University of Arizona Tucson, AZ 85721 e-mail: ajay@mercury.aichem.arizona.edu --- Administrivia: The rest of this message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chemistry-request@ccl.net Thu Jul 23 21:42:32 1992 Date: Thu, 23 Jul 1992 20:08 EST From: EWING@JCVAXA.JCU.EDU Subject: Summary of responses to transition state question To: chemistry@ccl.net Status: R Here's a summary of answers to a question I posed recently. Thanks to all who replied. (Dave Ewing ewing@jcvaxa.jcu.edu) The question: One of our transition state structures is quite symmetric (D3h) but the symmetry of the reaction path we're considering is lower (C3v). We can find a saddle point for this transition state using C3v symmetry, but when we try using D3h symmetry the saddle point is not found. Rather, the structure falls apart into a minimum well. The C3v saddle point structure is very symmetric, D3h for all intents and purposes. Why can't we find the same saddle point under D3h? We're using Gaussian 90 for this work. -------------------------------------------------------- From: IN%"roberson@hydroxide.chem.utah.edu" 12-JUL-1992 16:34:51.43 ... it sounds as if it could be related to one of two common problems in transition state searches: i) SYMMETRY BREAKING The numerical techniques used in solving the SCF equations contain certain inherent instabilities. Usually they don't cause problems < thank God > but in high-accuracy work the increase in variational flexibility due to the lifting of a constraint can result in energy relaxations of magnitudes greater than those of the physical effects under investigation. For details you can see Davidson's review article in JCP '83 87 4783-4790; I believe that Ed Earl has a recent paper out on the subject, but I don't have a citation handy... ii) HIGHER ORDER SADDLE POINTS Have you calculated the vibrational frequencies at the C3v stationary point? A true saddle point will have one imaginary frequency, but PES wlaking algo- rithms tend to be greedy and readily walk to the nearest stationary point regardless of local curvature; you may well be at a second-order saddle point. A more-subtle possibility is that your C3v saddle is a true transition state which corresponds to a D3h minimum; that is, it is as low as you can get in energy without distorting along the three-fold axis. If so, starting a TS walk in D3h symmetry will lead you away from the desired point! Perhaps these comments will be of some help... Mark Roberson *********************************************************************** From: IN%"br1@inel.gov" 13-JUL-1992 12:12:42.18 I believe the electronic state in the two symmetries do not correlate. What is the cluster you are working with? If the cluster is a open-shell system (or has large open shell contribution) then you could be running into the so-called 'doublet instability problem'. ************************************************************************ From: IN%"PUDZIANOWSKI@bms.com" 13-JUL-1992 13:28:32.75 I think you're seeing one of the possible pitfalls of imposing symmetry constraints during optimizations. When you do this, you're effectively restricting yourself to a sub-surface, or cross section, of the full energy surface. Thus, a critical point on a given cross section need not carry over to the full surface or to another cross section. For an analogy, think of all the possible conic sections and their relation to the cone in three dimensions. It's too bad, since symmetry constraints can save a lot of computational effort; however, one has to be careful to demonstrate that a given critical point is invariant. That is, a given critical point must actually have C3V symmetry, e.g., on the full 3N-6 (or 3N in Cartesian coordinates) dimensional energy sur- face if one expects to locate it with a geometry optimization restricted to C3V symmetry, i.e., by searching on a cross sec- tional surface of C3V symmetry. Otherwise, the critical point located on that cross section is just an artifact of the symmetry restrictions. I hope this helps, rather than confusing the issue! Andy P. *********************************************************************** From: IN%"PA13808%UTKVM1.BITNET@OHSTVMA.ACS.OHIO-STATE.EDU" 13-JUL-1992 18:12:44.16 It looks like another case of the need for extreme accuracy in inputting coordinates in this program. TRY using the NOSYMM option and see what it gives. This is a common problem formolecules with tetrahedral symmetry. Using bond dist and angles as input it is common for the program to end in errors because of apparent loss in symm.This problem was discussed a few months ago on the network. John E. Bloor(PA13808 at UTKVM1) *********************************************************************** From: IN%"fink@gnl2.ucdavis.edu" 14-JUL-1992 13:43:57.15 Your inability to locate a D3h transition state may well be related to the general problem of symmetry breaking of the electronic wave function that Doug McLean first investigated and Rich Martin was interested in for a while and others from time to time. Imposition of the higher symmetry point group constraints commonly leads to a solution that may be discontinuously removed from the potential energy surface at the lower point group, unless a multideterminant form for the wavefunction is used. Some people would call this non-dynamical correlation. If you are approaching the transition state from the C3v pathway, your trial vectors are likely to predispose the wavefunction to break symmetry. Gaussians keyword STABLE can help to identify these situations if you can get a converged result that rigorously maintains D3h. Design a z-matrix with parameters for optimization that always maintain the D3h point group, and my guess is you will converge to an energy significantly higher than your pathway values for C3v. Bill Fink --- Administrivia: The rest of this message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chemistry-request@ccl.net Thu Jul 23 22:56:35 1992 Date: Thu, 23 Jul 92 20:35:04 -0400 From: jle@world.std.com (Joe M Leonard) To: chemistry@ccl.net Subject: Re: MM3 Status: R Thanks to those who responded with MM3 atom types - they seem to be a superset of the MM2 types I'm familiar with, and I have the list up through type #72. If there are more than this, please let me (and others?) know... More MM3 questions - the first few papers do not list the functional form of the out-of-plane bending term. Am I correct in assuming that is is present in the same form as MM2? The MM3/Alkenes article includes terms labeled Sslope, Tslope and Tlquad4. Are these something new with MM3, or are they something only related to the J Comp Chem article (though I didn't see them mentioned)? The table of MM3 parameters for Alkenes does not list Stretch-Bend or Bend-Bend terms - are they the same as those for Alkanes? Finally, has the entire MM3 parameter list been published (or is it available by US or electronic mail)? Thanks in advance, Joe --- Administrivia: The rest of this message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net ---