From ilya@lisboa.ks.uiuc.edu Fri Apr 15 02:28:34 1994 Received: from lisboa.ks.uiuc.edu for ilya@lisboa.ks.uiuc.edu by www.ccl.net (8.6.4/930601.1506) id CAA13861; Fri, 15 Apr 1994 02:03:50 -0400 Received: from beograd by lisboa.ks.uiuc.edu (NX5.67d/NX3.0M) id AA08154; Fri, 15 Apr 94 01:03:50 -0500 From: Ilya Logunov Message-Id: <9404150603.AA08154@lisboa.ks.uiuc.edu> Received: by beograd.ks.uiuc.edu (NX5.67d/NX3.0X) id AA03938; Fri, 15 Apr 94 01:03:49 -0500 Date: Fri, 15 Apr 94 01:03:49 -0500 Received: by NeXT.Mailer (1.100) Received: by NeXT Mailer (1.100) To: CHEMISTRY@ccl.net Subject: Matrix element Dear Netters, Here is one more question concerning computation of matrix elements. I want to compute matrix element , where d/dq is a spatial derivative over a particular cartesian (or internal) nuclear coordinate. Is there any program that computes such matrix elements? ********************************************************* Ilya Logunov Theoretical Biophysics Group Department of Chemistry and Beckman Institute University of Illinois at U-C ilya@lisboa.ks.uiuc.edu ********************************************************* From mbdtsmd@hpf.ch.man.ac.uk Fri Apr 15 04:28:35 1994 Received: from nessie.mcc.ac.uk for mbdtsmd@hpf.ch.man.ac.uk by www.ccl.net (8.6.4/930601.1506) id DAA14532; Fri, 15 Apr 1994 03:51:18 -0400 Received: from hpf.ch.man.ac.uk (actually mchhpf.ch.man.ac.uk) by nessie.mcc.ac.uk with SMTP (PP); Fri, 15 Apr 1994 08:51:07 +0100 From: Mark Davidson Message-Id: <19413.9404150751@hpf.ch.man.ac.uk> Subject: Bibtex (style file) To: CHEMISTRY@ccl.net Date: Fri, 15 Apr 94 8:51:03 BST Cc: CHEMISTRY@ccl.net Mailer: Elm [revision: 70.85] To anyone, Has anybody got a bib-style file for the J. Mol. Struc. (THEOCHEM). If you have, could you make it available. Thanks Mark Davidson From oles@kjemi.uio.no Fri Apr 15 04:40:55 1994 Received: from pat.uio.no for oles@kjemi.uio.no by www.ccl.net (8.6.4/930601.1506) id DAA14505; Fri, 15 Apr 1994 03:41:04 -0400 From: Received: from ulrik.uio.no by pat.uio.no with local-SMTP (PP) id <00205-0@pat.uio.no>; Fri, 15 Apr 1994 09:40:46 +0200 Received: by kelvin ; Fri, 15 Apr 1994 09:40:44 +0200 Date: Fri, 15 Apr 1994 09:40:44 +0200 Message-Id: <9404150740.AA26131@kelvin> To: chemistry@ccl.net Subject: Ir basis and applications Satyam, My Email bounced (Seems that you've used up your disk quota!), so I'm replying to the list. There is a plethora of ECP's for 3rd row transition metals around, and our group has made some, too. These are of the Huzinaga type (Bonifacic & Huzinaga, J. C. P. 60, 2779, (1973)), with the nodes kept for the valence orbitals and the subvalence shell described by frozen orbitals. Some applications to iridium-containing systems: O. Gropen, M. Sj\o voll, H. Str\o msnes, E. Karlsen, O. Swang, and K. F\ae gri, ``RECP calculations for reactions of $H_{2}$ with $Pt$, $Os$, $Ir$, and $Re$ --- a systematic comparison.'' Theor. Chim. Acta 87, 373 (1994) O. Swang, K. F\ae gri, jr., and O. Gropen, ``Theoretical study of methane activation by $Re$, $Os$, $Ir$, and $Pt$'' J. Phys. Chem. 98, 3006 (1994) (The \o and \ae are LaTeX codes for some specifically Scandinavian letters) Best regards, Ole. From chp1aa@surrey.ac.uk Fri Apr 15 07:28:41 1994 Received: from mail-server.surrey.ac.uk for chp1aa@surrey.ac.uk by www.ccl.net (8.6.4/930601.1506) id HAA15830; Fri, 15 Apr 1994 07:11:58 -0400 Received: from central.surrey.ac.uk by mail-server.surrey.ac.uk with SMTP using (PP) id <21549-0@mail-server.surrey.ac.uk>; Fri, 15 Apr 1994 12:11:22 +0100 Received: by central.surrey.ac.uk (1.37.109.8/16.2) id AA01468; Fri, 15 Apr 1994 12:11:21 +0100 From: Mr Andrew D Allen Message-Id: <9404151111.AA01468@central.surrey.ac.uk> Subject: SCF-Convergence problems To: chemistry@ccl.net Date: Fri, 15 Apr 94 12:11:21 BST Mailer: Elm [revision: 70.85] My original request detailed problems obtaining an SCF for a partially optimised system. The replys I have got are detailed below. My original Plea for help... I am running AMPAC v2.1 and have run into problems with SCF convergence. I have a large, ~60 atoms, molecule which converges OK if carry out a geometry optimisation on all coordinates. The problem arises when I try to constrain the optimisation by flaging some of the bonds, angle and torsions to not be optimised. Has anyone experienced this problem with Semi-emp, or QM methods in general, and has anyone any possible ways to overcome the problem. I would like to avoid doing a full geometry optimisation for the molecule due to its size. Andy. REPLIES. ############################################################################# Hi, Andy, I would appreciate if you let me know (or summarize to CCL) the answers you get. I started to work on methods to improve convergence in various contexts. I am not familiar with AMPAC, unfortunately. Perhaps you could send me some more details? Best regards Herbert ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Greetings...Most cases of difficulty I've run into in MOPAC were cured by level shifting...doe AMPAC do this as well ? John McKelvey Wastman Kodak ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Does your AMPAC have the "EF" optimiser ? If so and it is like the one o in MOPAC then you can compute the Hessian at the beginning. That helps a lot. If you care to send me your original Z-matrix I could take a shot at it, out of scientific curiosity. John ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Dear Professor Allen, If you could send me trhe data file, I'd be happy to take a look at it and see what the difficulty is. I'll also run it through AMPAC 5.0 and check what the new program does as well. Please send both data files and the output file for the run that failed. Andy Holder ############################################################################## For the record, AMPAC ver2.1 has SHIFT, PULAY and CAMP-KING convergers and neither was sucessful in producing an SCF after 200 iterations. As for EF I have not come across this method. I have passed Andy Holder a copy of my input file and as far as I know he is still looking into it. I have since managed to optimise the the full structure, and have a gradient of 2.7 Kcal/Ang, I am further refining the structure to get the gradient even lower. -- ############################################################################### Structural and Computation Chemistry Group_________chp1aa@uk.ac.surrey - JANET. Department of Chemistry____________________________phone_______+44(483)-259591. University of Surrey_______________________________fax_________+44(483)-300803. Guildford, Surrey, GU2 5XH, UK_____________________ftp____________131.227.110.2 ############################################################################### From steve@chemiris.cc.binghamton.edu Fri Apr 15 08:28:38 1994 Received: from chemiris.cc.binghamton.edu for steve@chemiris.cc.binghamton.edu by www.ccl.net (8.6.4/930601.1506) id IAA16443; Fri, 15 Apr 1994 08:02:18 -0400 Received: by chemiris.cc.binghamton.edu (920330.SGI/920502.SGI) for CHEMISTRY@ccl.net id AA08245; Fri, 15 Apr 94 11:02:25 -0400 Date: Fri, 15 Apr 1994 10:56:30 -0400 (EDT) From: Steve Schafer Subject: Neural Networks To: CHEMISTRY@ccl.net Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII I am need of as much information as possible on the use of Artificial Neural Networks as applied to chemical systems. Any books or papers would be of great help. I have experience in using feed forward networks, but am also interested in any of the new networks (fuzzy nets .etc). Thank you in advance, Steven Schafer S.U.N.Y. Binghamton Chemistry Department From av@ocirs4.unizh.ch Fri Apr 15 09:28:50 1994 Received: from ocirs4.unizh.ch for av@ocirs4.unizh.ch by www.ccl.net (8.6.4/930601.1506) id IAA16873; Fri, 15 Apr 1994 08:49:00 -0400 Received: by ocirs4.unizh.ch (AIX 3.2/UCB 5.64/4.03) id AA14588; Fri, 15 Apr 1994 13:44:28 +0100 Date: Fri, 15 Apr 1994 13:44:28 +0100 From: av@ocirs4.unizh.ch (Alexander Voityuk) Message-Id: <9404151244.AA14588@ocirs4.unizh.ch> To: chemistry@ccl.net, voityuk@ocirs4.unizh.ch Subject: Semiempirical computations on Hg containing molecules Hello netters, If you are interested in semiempirical study of molecules containing mercury (there is an interest to semiempirical calculations on heavy elements in CCL), you are welcome to try MNDO/d method. Here is brief summary of heats of formation. I think no hype is necessary - mean errors speak for themselves. Calculated and observed heats of formation (kcal/mol) MOLECULE Exp. MNDO/d MNDO AM1 PM3 Hg+ 256.8 256.9 222.5 225.5 271.5 Hg++ 690.8 690.8 679.3 685.4 681.2 Hg(CH3) cation 222.0 226.8 218.8 220.8 251.5 Hg(CH3) radical 40.0 35.4 28.9 38.2 38.0 Hg(CH3)2 22.3 22.3 10.2 27.4 28.4 Hg(C2H5)2 17.8 10.9 11.8 19.7 15.8 Hg(i-C3H7)2 9.7 4.1 22.5 15.8 2.7 Hg(C3H7)2 8.2 2.2 6.9 7.8 3.0 Hg(C6H5)2 93.8 94.6 101.6 101.2 98.4 Hg(CH3)Cl -12.3 -10.6 -18.0 -12.5 -3.4 Hg(C2H5)Cl -15.0 -15.8 -16.7 -16.4 -9.8 C3H7HgCl -18.6 -23.1 -20.2 -23.1 -18.9 i-C3H7HgCl -20.1 -18.6 -10.9 -18.2 -16.3 Hg(C6H5)Cl 24.6 27.4 29.8 26.4 32.4 Hg(CH3)Br -4.4 -3.5 2.8 -11.0 -1.1 Hg(C2H5)Br -7.2 -8.9 4.0 -14.9 -7.7 C3H7HgBr -12.7 -16.0 .5 -21.2 -16.9 i-C3H7HgBr -12.6 -11.9 9.7 -16.8 -14.5 Hg(C6H5)Br 32.6 34.2 50.4 27.3 34.5 Hg(CH3)I 5.2 8.7 13.7 21.2 9.5 Hg(C2H5)I 3.3 3.1 14.6 17.1 2.9 C3H7HgI -1.2 -3.7 11.1 10.8 -5.6 i-C3H7HgI -1.1 -.2 20.2 15.0 -3.9 Hg(C6H5)I 42.9 46.4 60.5 59.1 45.1 Hg(CN)2 91.0 84.7 69.1 84.3 93.1 Hg(Acac)2 -141.2 -140.0 -104.2 -124.8 -176.8 HgCl2 -35.0 -35.1 -36.9 -44.8 -32.7 HgClBr -28.0 -29.0 -17.0 -44.8 -30.5 HgClI -19.6 -18.1 -8.8 -13.9 -21.6 HgBr2 -20.4 -22.6 3.1 -43.9 -26.9 HgBrI -12.1 -11.5 11.8 -12.6 -16.0 HgI2 -4.1 -.2 21.5 19.1 -4.2 HgH radical 57.2 66.2 37.8 55.9 48.1 HgCl radical 20.1 19.8 -6.3 -2.7 4.9 HgBr radical 24.9 24.3 14.8 -3.5 10.7 HgI radical 31.9 30.2 24.6 25.7 -32.2 Hg2 26.0 22.6 30.2 28.0 25.9 MEAN ABSOLUTE ERROR (37 comparisons) 2.45 13.22 9.20 7.85 Computations of molecules containing Hg are supported by SIBIQ program, an IBM PC version of the program is available with anonymous ftp from ocisgi7.unizh.ch as /pub/sibiq/sibiq100.zip. If you have problems with retriving SIBIQ, you are welcome to write to ps@ocisgi7.unizh.ch (Serge Pachkovsky) for the detailed instructions. With best regards, Dr. A.Voityuk From markus@EROS.CCC.Uni-Erlangen.DE Fri Apr 15 10:28:43 1994 Received: from enar.organik.uni-erlangen.de for markus@EROS.CCC.Uni-Erlangen.DE by www.ccl.net (8.6.4/930601.1506) id KAA18206; Fri, 15 Apr 1994 10:07:20 -0400 Received: from kochab.erosnet (kochab.organik.uni-erlangen.de) by enar.organik.uni-erlangen.de with SMTP id AA05961 for CHEMISTRY@ccl.net (5.65c/IDA-1.4.4/bs-02); Fri, 15 Apr 1994 16:06:30 +0200 From: Markus Wagener Received: from beteigeuze.erosnet by kochab.erosnet (5.0/SMI-4.1/client) id AA13102; Fri, 15 Apr 1994 16:06:36 --100 Date: Fri, 15 Apr 1994 16:06:36 --100 Message-Id: <9404151406.AA13102@kochab.erosnet> To: steve@chemiris.cc.binghamton.edu Subject: Re: CCL:Neural Networks Cc: CHEMISTRY@ccl.net Content-Length: 997 Hi Steve, the following could be interessting for you: Jure Zupan, Johann Gasteiger, "Neural Networks for Chemists, An Introduction", VCH Verlagsgesellschaft, Weinheim, 1993. ca. 300 pages The authors not only discuss the basics of neural networks, but also give 11 different examples of their application to chemical problems. Markus -- +-----------------------------------------------------------------------------+ | Dr. Markus Wagener | | Universitaet Erlangen-Nuernberg | | Computer-Chemie-Centrum Tel: +49-9131-85-6569 | | Naegelsbachstrasse 25 FAX: not yet available | | D-91052 Erlangen Email: | | Germany Markus.Wagener@EROS.CCC.Uni-Erlangen.DE | +-----------------------------------------------------------------------------+ From dwelrod@intnet.upj.com Fri Apr 15 14:28:44 1994 Received: from intnet.upj.com for dwelrod@intnet.upj.com by www.ccl.net (8.6.4/930601.1506) id NAA23137; Fri, 15 Apr 1994 13:50:08 -0400 From: Received: by intnet.upj.com (5.67/2.25) id AA13110; Fri, 15 Apr 94 17:46:50 GMT Received: by intnet.upj.com (5.67/2.25) id AA12595; Fri, 15 Apr 94 17:46:48 GMT Message-Id: <9404151746.AA12595@intnet.upj.com> Date: Fri, 15 Apr 94 17:46:48 GMT To: chemistry@ccl.net Subject: RE: Neural Networks in Chemistry Regarding Steven Schafer's request for info on neural nets in chemistry: The number of papers discussing chemical applications of neural networks is approaching 1000. The Zupan and Gasteiger book suggested by Dr. Wagener is an excellent source. Three recent reviews may be faster to acquire than a book, their refs are given below. I would suggest a search of CAS-online for a more complete list. Good Luck. ************************************************* TI Neural networks in chemistry AU Gasteiger, Johann; Zupan, Jure CS Org.-Chem. Inst., Tech. Univ. Muenchen LO Garching W-8046, Germany SO Angew. Chem., 105(4), 510-36 (See also Angew. Chem., Int. Ed. Engl., 1993, 105(4), 503-27) PY 1993 LA Ger AB A review with 68 refs. Models that simulate partial functions of the human brain were compared computer models designed to also simulate brain functions. The role of artificial neurons for classification of objects, modeling of functional relationships and storage and processing of data was discussed and related to storage and processing of chem. information. The applications in anal. of spectroscopic data, prediction of reactions, and control of chem. processes were discussed. KW neuron computer process chem review; artificial intelligence nerve process chem; simulation model intelligence nerve process chem ************************************************** TI Neural Networks: a new computational paradigm with applications in chemistry AU Elrod, David W.; Maggiora, Gerald M. CS Upjohn Lab. LO Kalamazoo, MI, USA SO Proc. Montreux Int. Chem. Inf. Conf., 138-62. Ed by: Collier, Harry. Infonortics: Calne, UK. PY 1993 LA Eng AB A review with 136 references. Computational neural networks are a new computational paradigm, inspired by the brain's massively parallel network of highly interconnected neurons. The interest in computational neural networks (CNN) derives mainly because they offer a new general paradigm for model building and function approximation, and not because of their relationship to models of brain function. Neural networks are considerably different from other computational methods in that functional relationships are not programmed in but are learned directly from data. This ability to adapt to new data, along with the "model- free" nature of neural networks allows them to excel at constructing very general, complex, nonlinear mappings between sets of input-output pairs, even with noisy data. In fact, CNNs can construct mappings of arbitrary accuracy for all reasonably well- behaved functions. In addition to their mapping ability, neural networks are also useful for pattern recognition/classification and self-organized clustering. A survey of the rapidly increasing number of applications of neural networks in chemistry will highlight some of their properties and limitations, and show what their impact may be in computer-aided chemistry. KW Computational neural networks, artificial neural networks, review, chemical applications. ***************************************************** TI Feed-Forward Neural Networks in Chemistry: Mathematical Systems for Classification and Pattern Recognition AU Burns, John A.; Whitesides, George M. CS Dept. Chem. Harvard U. LO Cambridge, MA. 02138 SO Chem. Rev., 93(8), 2583-2601 PY 1993 AB A review with 123 refs. ========================================================================== David W. Elrod, Ph.D. Computer-Aided Drug Discovery Computational Chemistry Upjohn Laboratories 301 Henrietta St. Kalamazoo, MI 49007-4940 Phone (616) 385-7772 Fax (616) 385-8488 Internet: dwelrod@upj.com ========================================================================== From mckelvey@Kodak.COM Fri Apr 15 16:28:44 1994 Received: from Kodak.COM for mckelvey@Kodak.COM by www.ccl.net (8.6.4/930601.1506) id QAA25930; Fri, 15 Apr 1994 16:13:04 -0400 From: Received: from NTPROF.DECnet MAIL11D_V3 by Kodak.COM (5.61+/2.1-Eastman Kodak) id AA09409; Fri, 15 Apr 94 16:15:21 -0400 Date: Fri, 15 Apr 94 16:15:19 -0400 Reply-To: mckelvey@Kodak.COM Message-Id: <9404152015.AA09409@Kodak.COM> To: osc@Kodak.COM Subject: "d" functions and the MNDO hamiltonian 15 April 94 Dear Netters: Given the recent discussions concerning semi-empirical hamiltonians, and with Walter Thiel's permission, I pass along the response to an inquiry I made to him in early March concerning the status of his paramerisations for MNDO/D. The work already published by him on MNDO/D speaks for itself, and I feel it raises a question as to duplication by other researchers. So....... __________________________________________________________________ Dear John, 11 March 94 Our distributed MNDO93 (UniChem) version contains MNDO/d parameters for Na,Mg,Al,Si,P,S,Cl,Zn,Ge,As,Se,Br,Cd,Sn,Sb,Te,I,Hg (18 elements, but only trivial metals). I still have to write up most of the corresponding papers - only the Cl,Br,I paper has appeared, and the Si paper is in press. We have preliminary parameters for Ti,Cr,Mn,Fe,Ni,Cu,Zr,Hf and plan to decide on the final formalism and parametrization within the next 6 months. The transition metals have been more difficult than the main-group elements (no surprise), and we just want to be sure to do a good job (hence extensive tests). Best Regards, Walter Thiel ________________________________________ Comments? John McKelvey Research Labs Eastman Kodak Co. Rochester, N.Y. 14650-2216 Phone 716-477-3335 Fax 716-722-2327 E-Mail McKelvey@Kodak.Com From gallion@mmd.com Fri Apr 15 17:29:58 1994 Received: from gate.mmd.com for gallion@mmd.com by www.ccl.net (8.6.4/930601.1506) id RAA26436; Fri, 15 Apr 1994 17:00:49 -0400 From: Received: by gate.mmd.com id AA04841 (InterLock SMTP Gateway 1.1 for CHEMISTRY@ccl.net); Fri, 15 Apr 1994 16:00:46 -0500 Date: Fri, 15 Apr 1994 17:01:32 -0400 (EDT) Message-Id: <940415170132.202034cf@cires.cin.mmd.com> Subject: Request for Ca++Channel Models To: CHEMISTRY@ccl.net X-Vmsmail-To: SMTP%"CHEMISTRY@ccl.net" Greetings Fellow Netters: Does anyone have (or know of) available molecular models for Ca++ channels? Thanks in Advance Steve --------------------------------------------------------- Steven L. Gallion Mail: gallion@mmd.com Computer Assisted Drug Design Phone: (513) 948-6581 Marion Merrell Dow FAX: (513) 948-7432 Cincinnati, OH 45215 ---------------------------------------------------------