From AJS1@phx.cam.ac.uk Tue Dec 7 04:53:44 1993 Received: from ppsw2.cam.ac.uk for AJS1@phx.cam.ac.uk by www.ccl.net (8.6.4/930601.1506) id EAA27430; Tue, 7 Dec 1993 04:34:14 -0500 Received: from fandango.ch.cam.ac.uk by ppsw2.cam.ac.uk with SMTP-CAM (PP-6.0) as ppsw.cam.ac.uk id <14511-0@ppsw2.cam.ac.uk>; Tue, 7 Dec 1993 09:33:08 +0000 Received: by fandango.ch.cam.ac.uk (920330.SGI/MDTG-Vevision: 1.3 fandango.ch.cam.ac.uk) id AA01575; Tue, 7 Dec 93 09:30:43 GMT From: Anthony Stone Message-Id: <9312070930.AA01575@fandango.ch.cam.ac.uk> Subject: Correction To: chemistry@ccl.net Date: Tue, 7 Dec 1993 09:30:43 +0000 (GMT) X-Mailer: ELM [version 2.4 PL13] Content-Type: text Content-Length: 415 Thanks to those who have pointed out that I gave the wrong reference in my recent posting about population analysis and dipole moments. It should have been Stone & Alderton, Mol. Phys. 56 (1985) 1047. Anthony Stone University Chemical Laboratory, Internet: ajs1@phx.cam.ac.uk Lensfield Road, Phone: +44 223 336375 Cambridge CB2 1EW Fax: +44 223 336362 From AJS1@phx.cam.ac.uk Tue Dec 7 05:12:50 1993 Received: from ppsw2.cam.ac.uk for AJS1@phx.cam.ac.uk by www.ccl.net (8.6.4/930601.1506) id EAA27452; Tue, 7 Dec 1993 04:48:45 -0500 Received: from fandango.ch.cam.ac.uk by ppsw2.cam.ac.uk with SMTP-CAM (PP-6.0) as ppsw.cam.ac.uk id <14869-0@ppsw2.cam.ac.uk>; Tue, 7 Dec 1993 09:48:26 +0000 Received: by fandango.ch.cam.ac.uk (920330.SGI/MDTG-Vevision: 1.3 fandango.ch.cam.ac.uk) id AA01599; Tue, 7 Dec 93 09:46:10 GMT From: Anthony Stone Message-Id: <9312070946.AA01599@fandango.ch.cam.ac.uk> Subject: Field gradients To: chemistry@ccl.net Date: Tue, 7 Dec 1993 09:46:09 +0000 (GMT) X-Mailer: ELM [version 2.4 PL13] Content-Type: text Content-Length: 969 Perhaps it is worth spelling out the fact that the field gradient tensor is not normally traceless, because its trace is F_xx + F_yy + F_zz = F_kk = - (del squared)V = (rho/epsilon_0), by Poisson's equation, where rho is the charge density. The `field gradient' tensor that occurs in the interaction with nuclear quadrupole moments is the same thing with the trace subtracted out: F'_ij = F_ij - (1/3)F_kk delta_ij. So the `field gradient' eq is in fact eq = F'_zz = F_zz - (1/3)(F_xx+F_yy+F_zz) (in principal axes), not just F_zz. Most of the textbooks fudge this in various ways, sometimes by just asserting that (del squared)V = 0 `because of Laplace's equation'. Naturally this leads to problems when people discover that it isn't so. Anthony Stone University Chemical Laboratory, Internet: ajs1@phx.cam.ac.uk Lensfield Road, Phone: +44 223 336375 Cambridge CB2 1EW Fax: +44 223 336362 From JEREMYW@num-alg-grp.co.uk Tue Dec 7 05:53:45 1993 Received: from ben.britain.eu.net for JEREMYW@num-alg-grp.co.uk by www.ccl.net (8.6.4/930601.1506) id FAA27642; Tue, 7 Dec 1993 05:44:02 -0500 Received: from num-alg-grp.co.uk by ben.britain.eu.net via JANET with NIFTP (PP) id ; Tue, 7 Dec 1993 10:43:50 +0000 Message-Id: <27114.9312071042@nags2.nag.co.uk> Received: from by nags2.nag.co.uk (4.1/UK-2.1) id AB27114; Tue, 7 Dec 93 10:42:24 GMT Date: 7 Dec 93 10:40:00 WET From: JEREMY WALTON Subject: Distribution of LAPACK and other public domain packages To: CHEMISTRY >> How about EISPACK and LINPACK in FORTRAN? >> Are they freeware? If so where can they be fetched? Thanks! These packages are available from NAG, who operate, on a cost-recovery basis, a distribution service for the following collections of public domain software: 1. LAPACK 2. EISPACK 3. LINPACK 4. MINPACK 5. OTTER 1-4 are all in FORTRAN; an implementation of LAPACK in C is in progress. More information is available from the NAG helpdesk (helpdesk@nag.co.uk, or infodesk@nag.com) or from the NAG gopher server at nags2.nag.co.uk. -------------------------------------------------------------------------------- | Jeremy Walton nagjpw@vax.oxford.ac.uk | | The Numerical Algorithms Group Ltd, Oxford, UK jeremyw@nag.co.uk | | Tel: +44 865 511245 | | Fax: +44 865 310139 | -------------------------------------------------------------------------------- From wieslaw@enstay.ensta.fr Tue Dec 7 06:53:51 1993 Received: from ensta.ensta.fr for wieslaw@enstay.ensta.fr by www.ccl.net (8.6.4/930601.1506) id GAA28010; Tue, 7 Dec 1993 06:47:21 -0500 Received: from enstay.ensta.fr (enstay.ensta.fr [192.44.56.1]) by ensta.ensta.fr (8.6.4/8.6.4) with SMTP id MAA04247 for ; Tue, 7 Dec 1993 12:46:24 +0100 Received: by enstay.ensta.fr, Tue, 7 Dec 93 13:09:11 +0100 Date: Tue, 7 Dec 93 13:09:11 +0100 From: wieslaw@enstay.ensta.fr (wieslaw) Message-Id: <9312071209.AA07434@enstay.ensta.fr> To: chemistry@ccl.net Subject: LAPACK with HP BLAS-question Dear Netters, I would e very grateful if somebody could give me an advice how to install LAPACK routines on HP735 in such a way that these routines use the HP BLAS library insted of BLAS present in the LAPACK package. It seems that functions in these two BLAS libraries have different names. Is changing the names in LAPACK sources the only way to have HP optimized code being implemented? With kind regards, Wieslaw Nowak Lab. of Aplied Optics, ENSTA,Ecole Polytechnique Centre de l'Yvette 91120 Palaiseau, France e-mail: wieslaw@enstay.ensta.fr From shenkin@still3.chem.columbia.edu Tue Dec 7 10:55:03 1993 Received: from mailhub.cc.columbia.edu for shenkin@still3.chem.columbia.edu by www.ccl.net (8.6.4/930601.1506) id KAA00233; Tue, 7 Dec 1993 10:35:23 -0500 Received: from still3.chem.columbia.edu by mailhub.cc.columbia.edu with SMTP id AA22737 (5.65c+CU/IDA-1.4.4/HLK for chemistry@ccl.net); Tue, 7 Dec 1993 10:35:16 -0500 Received: by still3.chem.columbia.edu (930416.SGI/930416.SGI.AUTO) for @cunixf.cc.columbia.edu:chemistry@ccl.net id AA12051; Tue, 7 Dec 93 10:34:44 -0500 Date: Tue, 7 Dec 93 10:34:44 -0500 From: shenkin@still3.chem.columbia.edu (Peter Shenkin) Message-Id: <9312071534.AA12051@still3.chem.columbia.edu> To: chemistry@ccl.net Subject: Acronym collision? I am writing a program that I will want to make available to the chemistry community. Never mind what it does, for now -- that will all be revealed in due time. The question is what to name it. A few of the names that seem appropriate reduce to the following acronyms: SACS EACS Again, never mind what these stand for -- that would be too much of a hint. The question is: are you aware of existing software likely to be in use by chemists which is known by either of these acronyms? Maybe we need an acronym registry; are you listening, Jan? :-) -P. ************************f*u*cn*rd*ths*u*cn*gt*a*gd*jb************************ Peter S. Shenkin, Box 768 Havemeyer Hall, Dept. of Chemistry, Columbia Univ., New York, NY 10027; shenkin@still3.chem.columbia.edu; (212) 854-5143 ********************** Atheist: an evangelical agnostic. ******************** From DSMITH@uoft02.utoledo.edu Tue Dec 7 11:06:28 1993 Received: from uoft02.utoledo.edu for DSMITH@uoft02.utoledo.edu by www.ccl.net (8.6.4/930601.1506) id KAA00248; Tue, 7 Dec 1993 10:36:06 -0500 Received: from UOFT02.UTOLEDO.EDU by UOFT02.UTOLEDO.EDU (PMDF V4.2-10 #3438) id <01H6718VQMAO004RR2@UOFT02.UTOLEDO.EDU>; Tue, 7 Dec 1993 10:35:54 EST Date: Tue, 07 Dec 1993 10:35:54 -0500 (EST) From: "DR. DOUGLAS A. SMITH, UNIVERSITY OF TOLEDO" Subject: Zinc Basis Sets - Summary (edited) To: chemistry@ccl.net Message-id: <01H6718VROVM004RR2@UOFT02.UTOLEDO.EDU> X-Envelope-to: chemistry@ccl.net X-VMS-To: CHEMISTRY MIME-version: 1.0 Content-transfer-encoding: 7BIT Appended is the summary of responses I received to my recent request for basis sets useful in the calculation of ZnR2 compounds. I have also included a general note from Tom Cundari from over a year ago regarding calculations on transition metals. Doug Douglas A. Smith Assistant Professor Department of Chemistry and member, Center for Drug Design and Development The University of Toledo Toledo, OH 43606-3390 voice 419-537-2116 fax 419-537-4033 email dsmith@uoft02.utoledo.edu =========================================================================== From: IN%"scheiner@biosym.com" 1-DEC-1993 20:22:25.73 To: IN%"dsmith@uoft02.utoledo.edu" CC: Subj: basis set for ZnR2 Doug, Take a look at the following publication: A. Schaefer, H. Horn, and R. Ahlrichs, "Fully Optimized Contracted Gaussian Basis Sets for Atoms Li to Kr", J. Chem. Phys. Vol. 97, pp. 2571-2577 (1992). No specific applications are reported, however, the basis sets for the first row transition metals are of valence double-zeta and full double-zeta quality, and should do reasonably well for the organo-zinc compounds of interest. - Andy ================================================================ Andrew C. Scheiner Phone: (619) 546-5346 Biosym Technologies, Inc. scheiner@biosym.com ================================================================ From: IN%"msrge@csv.warwick.ac.uk" "Mr C Wilson" 2-DEC-1993 06:08:00.56 To: IN%"DSMITH@uoft02.utoledo.edu" CC: Subj: RE: basis set for ZnR2? Either of the Los Alamos basis sets : LANL1MB LANL1DZ references for both basis sets are : P.J.Hay and W.R.Wadt J.Chem.Phys,82 pages 270-299 (1985) Craig From: IN%"jamorskc@CHIMCN.UMontreal.CA" 2-DEC-1993 08:43:37.27 To: IN%"dsmith@uoft02.utoledo.edu" CC: Subj: RE: basis set for ZnR2? Hi Doug, we did some calculation on ZnCH3 using two different all electron basis set here they are: MIDI 4 (13s,7p,4d) -> [5s,3p,2d]: plus two p polarization orbitals (alpha=0.176, 0.055) ref: S.Huzinaga edited by Elsevier - Amsterdam- Oxford- New-York- Tokyo 1984 Watcher Basis set (14s,11p,5d) -> [6s,5p,2d] ref: A.J.H.Watchers, J.Chem.Phys. vol. 52, p1033 (1969). with contraction of: L.Gianolo,R.Pavani,E.Clementi, Societa Chimica Italiana p108 (1978) regards, Christine Jamorski From: IN%"young@slater.cem.msu.edu" 2-DEC-1993 09:01:06.06 To: IN%"dsmith@uoft02.utoledo.edu" CC: Subj: Zn basis sets Hi, In response to your question about ZnR2 compounds. First of all, as with any approximation, you must ask yourself what exactly you want to determine and whether you want qualitative or quantitative answers and how accurate. However, as far as ab initio work, you will find that the most commonly used basis sets for that row of the transition series are Wachters basis J. H. Wachters, J. Chem. Phys. vol 53, no 3, 1 Feb (1970), p. 1033 which is a generalized contraction and Hay's segmented contraction set P. J. Hay, J. Chem. Phys., vol 66, no 10, 15 May (1977), p. 4377 These basis sets are often uncontracted further or have additional even tempered exponents added. For a look at some nice ab initio work, look at articles by Bauschlicker or Harrison. A few points of note on this type of calculation. Transition metals are fairly easy to deal with when they are in their preferred coordination. However, the low coordination compounds that are only stable in the gas phase can be tricky to deal with. They have many orbital energies and electronic states that are nearly degenerate. This leads to problems with convergence, converging to the correct state and contaminated wave functions. Convergence can be forced by the use of forced routines, such as the direct minimization in Gaussian. Converging to the desired state can be controlled by the construction of the initial guess (we often construct it by hand). If contaminated wave functions are a problem for the questions you are asking, you can work in a program such as COLUMBUS, where you have complete control over the symmetry constraints, usually at the expense of a terribly complex input. In short, you may find yourself in one of three situations. Your problem may work fine the first time. You may decide it's not worth the trouble. You may have a lot of work and fun gaining an in depth knowledge and mastery of ab initio techniques. Good luck. Dave Young young@slater.cem.msu.edu youngdc@msucem From: IN%"CUNDARIT%MEMSTVX1.BITNET@OHSTVMA.ACS.OHIO-STATE.EDU" 29-OCT-1992 23:17:48.42 To: IN%"chemistry@ccl.net" CC: Subj: More TM calcs. Concerning TM calcs, I'd like to follow up on Thersa Windus' comments on transition metal calcs. We've probably used GAMESS (Iowa State version) as extensively for transition metals as anyone else and from our point of view the big benefit to GAMESS is the built-in ECPs from Walt Stevens and co., the so-called SBKJs (Can. J. Chem. 1992, 70, 612). We have had quite good success in predicting geometries, rotational barriers, reaction barriers and the like using the SBKJs, which are formally triple-zeta for the "valence" nd, (n+1)s and (n+1)p. They also include the (n-1)s and (n-1)p outer core which the concensus seems to indicate are essential. To my knowledge, the SBKJ for the s- and p-block elements are built in Gaussian, but not for the d-block although they can be entered manually. Now that the SBKJs have been published I encourage anyone interested in getting into TM-containing systems to give them a whirl in their favorite program. Folks have, of course, have taken the basic Hay-Wadt scheme and augmented the basis sets and done some excellent TM work; the research of Morokuma, Hall and their collaborators come to my mind, but undoubtedly there are others. The work of Gernot Frenking is the first methodical analysis that I have seen of the various core size and basis set augmentation schemes for the Hay-Wadt. I highly recommend Gernot's article to anyone planning on getting into TM calcs. There are also the ECP schemes of Ross, Ermler, Pitzer, et al.; if my understanding is correct, this group has now extended their scheme to the entire Periodic Table! Doesn't CADPAC include Huzinaga's model potentials for the d-block? Preuss and company have published ECP schemes for the lanthanides, and we have extended the SBKJ scheme to the lanthanides. Thus, the ECP options in ab-initio calcs. are rapidly growing, definitely worth a look for anyone interested in the nether regions of the Periodic Table. Tom Cundari Assistant Professor Department of Chemistry Memphis State University Memphis, TN 38152 From: IN%"kauppm@CHIMCN.UMontreal.CA" 2-DEC-1993 10:14:00.50 To: IN%"dsmith@uoft02.utoledo.edu" CC: Subj: ZnR2 Dear Dr. Smith, somebody passed me your request for information on ZnR2 calculations. In fact, we had the same problem some years ago, as we were interested in some alkyl transfer reactions. So we tested several pseudopotential/ basis set combinations on M(CH3)2 and M(CH3) (M=Zn,Mg). This is published in J. Comp. Chem. 1990,11,1029-1037. The extensive applications we made can be found in J.Am.Chem.Soc. 1991,113,5606. Actually, these studies were part of my diplom thesis. I am presently spending some time in Canada, and my present address is given below. Please feel free to ask for more information. With best regards, Martin Kaupp ************************************************************ * Dr. Martin Kaupp * * Departement de chimie * * Universite de Montreal Fax (514) 343-2468 * * C.P. 6128, Succ. A Tel. (514) 343-6111-3991 * * Montreal, Quebec H3C 3J7 * * Canada * * email kauppm@chims1.chimcn.umontreal.ca * ************************************************************ From: IN%"hommes@organik.uni-erlangen.de" "Nico van Eikema Hommes" 3-DEC-1993 03:10:21.67 To: IN%"DSMITH@uoft02.utoledo.edu" CC: Subj: RE: basis set for ZnR2 Hello Doug! We have good experiences with the basis sets recently developed by the Ahlrichs group in Karlsruhe. See J.Chem.Phys. 97, 2571 (1992). The basis set files are available via anonymous ftp from tchibm3.chemie.uni-karlsruhe.de (129.13.108.8) We have the split-valence and the largest (TZ or DZ) basis sets converted to Gaussian-92 format. Email me if you would like a copy. Also, the newer pseudopotentials by Hay and Wadt as well as the Stuttgart pseudopotentials work fine. Gernot Frenking from Marburg recently published research using these in J.Comp.Chem. Second derivatives may not be available for these (Gaussian 92 does not have them, for instance), but they are very good e.g. for high-level single points. Greetings from Erlangen. Nico -- +=====================================+================================+ | Dr. N.J.R. van Eikema Hommes | hommes@organik.uni-erlangen.de | | Institut fuer Organische Chemie I | Tel. : 49/0 - 9131 - 85 - 4096 | | Henkestr. 42, D-91054 Erlangen, FRG | Fax : - 9132 | +=====================================+================================+ From: IN%"d3e129@cagle.pnl.gov" 3-DEC-1993 10:47:49.91 To: IN%"DSMITH@uoft02.utoledo.edu" CC: Subj: basis set for ZnR2? Doug, I don't know the reference off the top of my head but David Funk and Grzegorz Chalasinski did some work with Zinc when Dave was a grad student at Utah. As I recall the open d-shell on the Zn atoms caused some strange sensitivities in the basis set. The results should have been published post 1985. This ain't much info but hopefully it will help some. Regards, Rick Kendall ============================================================================= === Ricky A. Kendall === === Mail Stop K1-90 === === Molecular Science Software Group === === Theory, Modeling, and Simulation Program === === Molecular Science Research Center === === Pacific Northwest Laboratory === === Richland, WA 99352 === === Phone: (509) 375-2602 === === Fax: (509) 375-6916 === ============================================================================= From herbert.homeier@rchs1.chemie.uni-regensburg.de Tue Dec 7 11:55:04 1993 Received: from rrzs1.rz.uni-regensburg.de for herbert.homeier@rchs1.chemie.uni-regensburg.de by www.ccl.net (8.6.4/930601.1506) id KAA00409; Tue, 7 Dec 1993 10:55:15 -0500 Received: from rchs1.chemie.uni-regensburg.de (rchs2.chemie.uni-regensburg.de) by rrzs1.rz.uni-regensburg.de (4.1/URRZ-relay (1.4)) id AA11784; Tue, 7 Dec 93 16:55:03 +0100 Received: by rchs1.chemie.uni-regensburg.de (4.1/URRZ-sub (1.31)) id AA00567; Tue, 7 Dec 93 16:56:42 +0100 Date: Tue, 7 Dec 93 16:56:42 +0100 From: Herbert Homeier (t4720) Message-Id: <9312071556.AA00567@rchs1.chemie.uni-regensburg.de> To: CHEMISTRY@ccl.net Subject: Re: Distribution of LAPACK and other public domain packages Hi netters, besides from NAG as noted below, eispack, lapack, linpack, minpack, and many other packages are also available from netlib. (anon ftp from netlib.ornl.gov or get xnetlib using archie which is a slowly starting but otherwise nice X11 surface to access netlib). Best regards Herbert ----- Begin Included Message ----- From chemistry-request@ccl.net Tue Dec 7 15:39:57 1993 Subject: Distribution of LAPACK and other public domain packages To: CHEMISTRY Sender: chemistry-request@ccl.net Content-Length: 1345 X-Lines: 30 >> How about EISPACK and LINPACK in FORTRAN? >> Are they freeware? If so where can they be fetched? Thanks! These packages are available from NAG, who operate, on a cost-recovery basis, a distribution service for the following collections of public domain software: 1. LAPACK 2. EISPACK 3. LINPACK 4. MINPACK 5. OTTER 1-4 are all in FORTRAN; an implementation of LAPACK in C is in progress. More information is available from the NAG helpdesk (helpdesk@nag.co.uk, or infodesk@nag.com) or from the NAG gopher server at nags2.nag.co.uk. -------------------------------------------------------------------------------- | Jeremy Walton nagjpw@vax.oxford.ac.uk | | The Numerical Algorithms Group Ltd, Oxford, UK jeremyw@nag.co.uk | | Tel: +44 865 511245 | | Fax: +44 865 310139 | -------------------------------------------------------------------------------- ---Administrivia: This message is automatically appended by the mail exploder: CHEMISTRY@ccl.net -- everyone | CHEMISTRY-REQUEST@ccl.net -- coordinator MAILSERV@ccl.net: HELP CHEMISTRY | Gopher: www.ccl.net (coming soon) Anon. ftp www.ccl.net | CHEMISTRY-SEARCH@ccl.net -- archive search ----- End Included Message ----- From ADAMO@CHEMNA.DICHI.UNINA.IT Tue Dec 7 12:02:13 1993 Received: from CHEMNA.DICHI.UNINA.IT for ADAMO@CHEMNA.DICHI.UNINA.IT by www.ccl.net (8.6.4/930601.1506) id LAA01136; Tue, 7 Dec 1993 11:40:35 -0500 From: Received: from CHEMNA.DICHI.UNINA.IT by CHEMNA.DICHI.UNINA.IT (PMDF V4.2-11 #3559) id <01H67G4SHJU80000J7@CHEMNA.DICHI.UNINA.IT>; Tue, 7 Dec 1993 17:40:06 CET Date: Tue, 07 Dec 1993 17:40:06 +0100 (CET) Subject: THANK for MOLCAS To: chemistry@ccl.net Message-id: <01H67G4SJ5PU0000J7@CHEMNA.DICHI.UNINA.IT> X-VMS-To: IN%"chemistry@ccl.net" X-VMS-Cc: ADAMO MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII Content-transfer-encoding: 7BIT Dear netters, THANK. I received many answers to my question on MOLCAS package. Overall I received them in few hours, saving so a lot of time. I enclose a summary of the answers. ciao carlo ------------------------------------------------------------------------------- Carlo Adamo | tel. +39-81-5476504 Dipartimento di Chimica | fax +39-81-5527771 via Mezzocannone 4 | e-mail ADAMO@CHEMNA.DICHI.UNINA.IT I-80134 Napoli | Italy | ________________________________________________________________________________ ****************************************************************************** MOLCAS contact point molcas@garm.teokem.lu.se -------------------- MOLCAS order form ----------------- To order MOLCAS please fill in the details below and sign in the space provided. To accept your order, please also sign a copy of the MOLCAS copyright agreement and send both forms to : Prof. B. O. Roos, Department of Theoretical Chemistry, University of Lund, Chemical Center, P.O.B. 124, S--22100 Lund, Sweden +------------------------------------------------------------------------------+ ! Name: ! ! ! ! ........................................................................... ! ! Name of organisation/institution: ! ! ! ! ........................................................................... ! ! Address/site: ! ! ! ! ........................................................................... ! ! ! ! ........................................................................... ! ! ! ! ........................................................................... ! ! ! ! ........................................................................... ! ! ! ! ........................................................................... ! ! E-mail address: ! ! ! ! ........................................................................... ! ! Telephone: ! ! ! ! ........................................................................... ! ! Fax: ! ! ! ! ........................................................................... ! +------------------------------------------------------------------------------+ Please note which of the available versions you would like to get: +------------------------------------------------------------------------------+ ! o IBM / VM-XA half inch tape in VMFPLC2 format (9600Bpi) ! ! o IBM RISC-6000 / AIX 3 quarter inch tape in 'tar' format (150MBytes) ! ! o IBM RISC-6000 / AIX 3 8mm video tape in 'tar' format (2.3 GBytes) ! ! o IBM RISC-6000 / AIX 3 HD disketts ! +------------------------------------------------------------------------------+ I/we place an order for MOLCAS: +------------------------------------------------------------------------------+ ! Signature: ! ! ! ! ........................................................................... ! ! Date: ! ! ! ! ........................................................................... ! +------------------------------------------------------------------------------+ (new page) MOLCAS copyright agreement -------------------------- MOLCAS is a general purpose software package for electronic structure calcu- lations with special emphasize on correlated wavefunctions. MOLCAS contains programs and documentations written by the members of the Department of Theoretical Chemistry, University of Lund, Sweden and other suppliers. The programs have been assembled into a common software package in cooperation with IBM Sweden AB. The University of Lund reserves the copyright and all other property rights on this material. Permission is hereby granted to use MOLCAS subject to the terms set below: (i) The software may be used only in the environment and at the site specified on the order form. (ii) The use is restricted to research purposes only. (iii) The user may alter the software as far as it is required for his research purposes. (iv) The user may make copies of the material as are reasonably required for the user's own working and back-up purposes. The copies will then become subject to these conditions. (v) No part of this material may be redistributed without the specific written permission of the authors of MOLCAS. (vi) The material enclosed is supplied without any warranty of any kind. The authors of MOLCAS, therefore, assume no responsibility of any kind arising from the use of the material. I/we have taken notice of the MOLCAS copyright agreement: +------------------------------------------------------------------------------+ ! Name: ! ! ! ! ........................................................................... ! ! Name of organisation/institution: ! ! ! ! ........................................................................... ! ! Signature: ! ! ! ! ........................................................................... ! ! Date: ! ! ! ! ........................................................................... ! +------------------------------------------------------------------------------+ From noy@tci002.uibk.ac.at Tue Dec 7 12:53:57 1993 Received: from uibk.ac.at for noy@tci002.uibk.ac.at by www.ccl.net (8.6.4/930601.1506) id MAA01987; Tue, 7 Dec 1993 12:20:52 -0500 Received: from tci002.uibk.ac.at by uibk.ac.at with SMTP id AA07490 (5.65c/IDA-1.4.4 for ); Tue, 7 Dec 1993 18:18:29 +0100 Received: by tci002.uibk.ac.at (AIX 3.2/UCB 5.64/CFIBK-2e.AIX) id AA15726; Tue, 7 Dec 1993 18:18:27 +0100 From: noy@tci002.uibk.ac.at (Teerakiat Kerdchaoren) Message-Id: <9312071718.AA15726@tci002.uibk.ac.at> Subject: Summary: scaling velocities in MD To: chemistry@ccl.net, chem-comp@mailbase.ac.uk Date: Tue, 7 Dec 1993 18:18:27 +0100 (NFT) X-Mailer: ELM [version 2.4 PL21] Content-Type: text Content-Length: 11605 Dear computational chemistry mailing-list members, A couple of weeks ago, I post a question about scaling velocities in MD to the net. The original messages and the responses are herewith enclosed. I would like to express my special thanks to them whom sent me those replies. Indeed, I found all of them valuable and helpful. Wish you nice X'mas. warm regards, Teerakiat noy@tci2.uibk.ac.at ---------------------- original messages ----------------------- Dear CCL fellows, I am writing an MD program and here comes some questions. To my experience, the average temperature tends to raise up gradually in a simulation resulting in the increasing of total energy. 1) Is it possible to apply a "scaling temperature" algorithm to control the temperature range ( my MD is NVE ) ? To my idea, scaling the velocities cause the deviation of the p-q coordinate ( generalized momentum-generalized coordinates ) in the phase space. The trajectories are therefore deviated, is it true ? How to scale both velocities and coordinates to keep the correct trajectories and p-q coordinates in the phase space ? 2) Is it possible to apply an "external bath" in which is generally used with NVT ensemble to an NVE ensemble ? Thanks a lot in advance. best wishes, T.K.Noy ------------------------ response ------------------------------- From: "Michael E. Coltrin" We published a method for simulating the effects of a thermal reservoir on lattice atoms. The method ensures the proper flow of energy between the dyanamical atoms in your trajectory study and a surrounding bath (if the bath is at a different temperature). The method also will ensure that the total energy will not "drift" as you described. Briefly, the method periodically resets a portion of the velocity vector of each atom at the edge of the dynamical zone (but doesn't reset the position). The method is very easy to implement in a trajectory code. The reference is M. E. Riley, M. E. Coltrin, and D. J. Diestler, J. Chem. Phys., 88, 5934 (1988). I hope this will be useful to you. Mike Coltrin ------------------------------------------------------------------------- From: burkhart@goodyear.com (Craig W. Burkhart) You may want to try the canonical ensemble dynamics of Nose' and Hoover. Their method is inherently isothermal, whereas the standard methods are isokinetic. That is a subtle, but important, difference. The appropriate refs are: S. Nose', J. Chem. Phys. 81, 511(1984). W.H. Hoover, Phys. Rev. A 31, 1695(1985). Hope this helps. Craig W. Burkhart, Ph.D -------------------------------------------------------------------------- From: tolja@maya.geo.uu.se (Anatoli Belonoshko) I have written quite a few MD programs. To my experience, the average temperature does not tend to rise gradually, unless you 1) have not "equilibrated" your system properly 2) use inappropriate numerical scheme solving Hamilton equation 3) use too large timestep 4) still have some other bugs Usual practice for NVE ensemble is to equilibrate system scaling velocities to maintain the temperature, and then to stop the scaling. NVE is rather famous on inability to provide "exact" temperature. On my experience, the scaling during "averaging" stage does not cause sigificant errors, but it is better still to avoid such a scaling. "external bath" technique is applicable only to NVT ensemble by definition. Constant energy simulation can not be realized when system have an energy external source. Best regards, Anatoly Belonoshko Uppsala University, Sweden e-mail: tolja@maya.geo.uu.se -------------------------------------------------------------------------- From: dam@retina.chem.psu.edu (K.V. Damodaran) Yes, the average temperature tends to rise in the beginning of the simulation. BUT, usually the starting configuration has a very large potential energy and the increase in temperature is due to the strong forces, a consequence of which is the sharp decrease in potential energy. Thus in NEV simulations the total energy will not be conserved very well in this region. But this part of the trajectory is discarded anyway. Velocity scaling applied for this part of the trajectory scales down the velocities by SQRT(T-To), when the temperature goes beyond some tolerance. NEV simulations do not require velocity scaling when you actually collect the data. So you probably don't have to be concerned about the deviations of the p-q coordinate in the phase space. Scaling the coordinates will amount to using an external pressure bath. If you apply an external temp. bath to an NVE ensemble, it will become an NVT ensemble!! See the book by Allen & Tildesley for details. damodaran dam@retina.chem.psu.edu ------------------------------------------------------------------------ From: doherty@msc.edu (David C. Doherty) Subject: Re: Molecular Dynamics questions Status: RO > 2) Is it possible to apply an "external bath" in which is >generally used with NVT ensemble to an NVE ensemble ? A method that many use is found in Berendsen et al., J. Chem. Phys., 81, 3684-3690 (1984). This includes a pressure bath also. --- David C. Doherty Minnesota Supercomputer Center, Inc. doherty@msc.edu ------------------------------------------------------------------------- If you are running an NVE simulation, then the temperature should not rise. This indicates problems with your simulations. Check if the forces are being calculated correctly. Also, if there are no other problems, your time step may be too large. See if this systematic change in the energy persists even if you reduce the time step. You could add a temperature controller but you should first make sure your simulation runs correctly in NVE. Hope this helps. Dilip Paithankar Chemical Engineering, Purdue University, West Lafayette, IN 47907 Telephone: 317/495-6910 (home), 317/494-4055 (office) 317/494-0805 (FAX). ------------------------------------------------------------------- I am writing a MD-program myself and I encountered the same problem. The only solution I found was to restrict the run to a sensible length (say: energy conservation 1 : 1.0d(6-7)). I would be interested in the other suggestions you get. Please report to the net. Yours sincerely, Ferenc Molnar EMail (SMTP): ferenc@rchsg8.chemie.uni-regensburg.de c5071@rchs1.chemie.uni-regensburg.de ferenc@rchnw2.ngate.uni-regensburg.de --------------------------------------------------------------------------- From: Jim_Bareman.XRCC@xerox.com One of the really nice features of the molecular dynamics technique is that there is usually a "conserved quantity" (such as the Hamiltonian) which makes it relatively easy to detect if you have an error in your code. In your case, if you are carrying out an NVE calculation, then the total energy MUST be conserved. The fact that you do not observe this suggests one of two things to me: (1) You have a bug in your code. This may be something as trivial as an actual coding error, or something more subtle, such as having too great a distance between points in a potential look-up table. (2) The system you are trying to simulate is in an unfavorable configuration, resulting in large forces on the particles. If this is the case, you will need very small time steps in order to obtain total energy conservation. The situation described in (2) should disappear if the system is allowed to run on for a while. From your problem description, I suspect (1) is likely to be your problem. I suggest you use the conservation of total energy to ensure that your code is really correct, rather than to try cover up the problem with some sort of velocity scaling. Happy Bug Hunting! Dr. James P. Bareman Member of Research Staff Xerox Research Centre of Canada. ------------------------------------------------------------------------- From: JEREMY WALTON The traditional way to get the NVE simulation running at (close to) the temperature you want is to have an equilibration period where the velocities are all uniformly scaled so the system is forced to have the kinetic energy appropriate to the required temperature. Following this period (typically a few thousand timesteps), scaling is switched off and the averaging over configurations can be started. The temperature should be monitored during this averaging period (it's one of the things you calculate as an average); if it drifts too much, then maybe the equilibration period wasn't long enough, or maybe your timestep is too big. There are a number of algorithms for this. Two which I recall are that of Berendsen and that of Nose. Check out Allen & Tildesley's "Computer Simulation of Liquids" (OUP, 1987), where all this is described in some detail. Hope this helps, Jeremy ---------------------------------------------------------------------------- From: I have a comment that is a "tangental" answer. In fitting several variable temperature NMR simulations where you have an equilbrium situation between several species of various populations, I noted that I could not obtain satisfactory fits of my spectra by simply increasing the rate of exchange between the various species (as dictated by the Eyring equation) until I began to increase the populations of the minor species with increasing temperature. Further, the best fits were obtained when I could track the popolations of each species at several temperatures and generate a Van't Hoff plot that could be used to "predict" the populations for spectra where overlap prevented population elucidation. If your MD model is varying in temperature, a more realistic model would include an increasing of minor conformation populations with increasing temperature (as dictated by the Van't Hoff equation). What do others think. Jay Brown ---------------------------------------------------------------------- From: "Gerald Loeffler" there is a 'classical' paper by by Berendsen et. al. that served as a guideline for me, when I implemented temperature-scaling in my MD-program: Berendsen,Postma,van Gunsteren,DiNola,Haak:"Molecular dynamics with coupling to an external bath",J.Chem.Phys.,Vol.81,No.8,1984 good luck, Gerald ----------------------------------------------------------------------- From: Rob Hooft In general, if your E rises in a NVE you should try to 1) reduce the time step. 2) If a cutoff radius is used, increase it. Rising energy in NVE is due to problems with integration accuracy, so the only real way to decrease the rising is to improve the integration accuracy. ------------------------------------------------------------------------ From: Tom Connor Bishop i'm sure you've had more than one person tell you that something must be wrong if you're total energy increases. this is the standard test to see if the simlation is accuarate ie the law of conservation of energy must hold. often times temperatures will increase but this must be coupled with a decrease in potential energy!!! either something is wrong with your code, or you are not running the simulation with parameters appropriate to the system. (note that for non-equilib systems the time step must be shorter in order to obtain numberical stability) just my ------------------------------------------------------------------------- From noy@tci002.uibk.ac.at Tue Dec 7 13:53:57 1993 Received: from uibk.ac.at for noy@tci002.uibk.ac.at by www.ccl.net (8.6.4/930601.1506) id NAA02550; Tue, 7 Dec 1993 13:29:04 -0500 Received: from tci002.uibk.ac.at by uibk.ac.at with SMTP id AA08146 (5.65c/IDA-1.4.4 for ); Tue, 7 Dec 1993 19:28:21 +0100 Received: by tci002.uibk.ac.at (AIX 3.2/UCB 5.64/CFIBK-2e.AIX) id AA26398; Tue, 7 Dec 1993 19:28:19 +0100 From: noy@tci002.uibk.ac.at (Teerakiat Kerdchaoren) Message-Id: <9312071828.AA26398@tci002.uibk.ac.at> Subject: rigid model in H-bond system To: chemistry@ccl.net, chem-comp@mailbase.ac.uk Date: Tue, 7 Dec 1993 19:28:19 +0100 (NFT) X-Mailer: ELM [version 2.4 PL21] Content-Type: text Content-Length: 935 Dear computational chemistry mailing-list members, I am working with an MD of ion in liquid ammonia system. The rigid model is employed in the simulation. There have been some doubts with the simulation of H-bond system using the rigid models in that the geometry associated with H-bond is measurably different from the geometry in the gas phase. Hence, Stillinger and Rahman ( J.Chem.Phys. 60, 1545(1974) ) included the intramolecular vibration into their ST2 rigid model in an average sense by changing O-H bondlength slightly. Robert and coworkers ( J.Chem.Phys. 82(11),5289 (1985) ) tried to investigate the effect of intramolecular vibrations on the intermolecular structure of water, and found that the effect is quite small. Does anybody know there are such kind of works on other H-bond systems such as NH3 and HF ? Any references and comments are appreciated. Thanks a lot in advance. sincerely, Teerakiat From ravishan@swan.wcc.wesleyan.edu Tue Dec 7 14:53:54 1993 Received: from swan.wcc.wesleyan.edu for ravishan@swan.wcc.wesleyan.edu by www.ccl.net (8.6.4/930601.1506) id OAA03291; Tue, 7 Dec 1993 14:36:39 -0500 Received: by swan.wcc.wesleyan.edu (AIX 3.2/UCB 5.64/4.03) id AA15182; Tue, 7 Dec 1993 14:33:53 -0500 Date: Tue, 7 Dec 1993 14:33:53 -0500 Message-Id: <9312071933.AA15182@swan.wcc.wesleyan.edu> From: "G. Ravishanker" To: chemistry@ccl.net Subject: Averaging question.... Hi I have a question regarding averaging properties over an MD run - especially those that fluctuate a lot (specifically interproton distances in MD simulations). For example, if one is monitoring interproton distances for NMR calculations, typically there is a cutoff associated with it, so that we do not keep information on an unusual number of "unwanted" pairs. The question is, if the cutoff is 10A and during the course of the MD, the molecule distorted so that some pairs that did not qualify initially are within 10A. How does (or should) one deal with this? 1. Use a large enough cutoff that this would not matter. (The programs will be larger and the problem to sort things out will get worse). 2. Update the interproton pairlist once every N steps. I prefer the second one. However, how does one average in the second case. Let us say, there is a pair that came within 10A in the last 10 ps of a 100ps trajectory. Do I use the average distance to be 1. sum of distances / 10ps or 2. sum of distances / 100ps Any comments or helpful insight will be appreciated. Ravi **************************************************************************** * Ganesan Ravishanker Ph: (203) 344-8544 Ext. 3110 * * Coordinator of Scientific Computing, Fax:(203) 344-7960 * * Adjunct Associate Professor(Dept. of Chem.) * * Wesleyan University e-mail:ravishan@swan.wcc.wesleyan.edu * * Middletown, CT 06457. * **************************************************************************** From stoutepf@chemsci1.es.dupont.com Tue Dec 7 16:53:59 1993 Received: from gatekeeper.es.dupont.com for stoutepf@chemsci1.es.dupont.com by www.ccl.net (8.6.4/930601.1506) id QAA05162; Tue, 7 Dec 1993 16:29:46 -0500 Received: by gatekeeper.es.dupont.com (5.65/ULTRIX-mjr-062991); id AA26568; Tue, 7 Dec 93 16:29:37 -0500 Received: from [138.196.64.95] (chem95.es.dupont.com) by chemsci1.es.dupont.com. chemsci1 chemsci1.es.dupont.com (4.1/SMI-4.1) id AA08665; Tue, 7 Dec 93 16:28:05 EST Message-Id: <9312072128.AA08665@chemsci1.es.dupont.com. chemsci1 chemsci1.es.dupont.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Reply-To: stoutepf@chemsci1.es.dupont.com X-Organization: The Du Pont Merck Pharmaceutical Company X-Mailer: Eudora 1.4 Date: Tue, 7 Dec 1993 16:28:45 -0500 To: chemistry@ccl.net, dibug@comp.bioz.unibas.ch From: stoutepf@chemsci1.es.dupont.com (Pieter Stouten) Subject: DMol or DGauss ? Cc: hodgecn@lldmpc.dnet.dupont.com (Nick Hodge), wasserzr@lldmpc.dnet.dupont.com, stoutepf@chemsci1.es.dupont.com We will purchase a (DFT) ab initio package, and are considering DMol (sold by Biosym) and DGauss (sold by MSI, part of Unichem). Our situation is: * We already have a Biosym infrastructure * We will be running the DFT program on SGI machines * We would have to purchase a graphical interface to the DGauss program, if we go with that program I sollict your opinions with respect to the following issues: * What is the better program in terms of reliability and speed ? * What "useful" features are present in one and not in the other ? * How good are the interfaces of these programs ? Example: how easy would it be to retrieve charges and use those in molecular mechanics ? Or: how easy is it to display orbitals, reactive sites, etc. ? * And most important, given our present situation would DGauss be that much faster, more reliable and useful to warrant purchasing an entirely new software package ? Thanks for your help. I will summarize responses if there is enough interest. Cheers, Pieter. Pieter Stouten, Senior Research Scientist || Computer Aided Drug Design Group || The Du Pont Merck Pharmaceutical Company || Adventures get spoiled P.O. Box 80353, Wilmington, DE 19880-0353 || by being reduced to data Phone: +1 (302) 695 3515 || -- ARA/Fax: +1 (302) 695 4324 || Poul Anderson E-mail: stoutepf@chemsci1.es.dupont.com || Internet Shogi Server: kzinti || From mjf@iris120.biosym.com Tue Dec 7 20:53:55 1993 Received: from bioc1.biosym.com for mjf@iris120.biosym.com by www.ccl.net (8.6.4/930601.1506) id TAA07062; Tue, 7 Dec 1993 19:56:49 -0500 Received: from iris120.biosym.com by bioc1.biosym.com (5.64/0.0) id AA00279; Tue, 7 Dec 93 16:56:18 -0800 Received: by iris120.biosym.com (931110.SGI.ANONFTP/920502.SGI) for @bioc1.biosym.com:chemistry@ccl.net id AA09407; Tue, 7 Dec 93 16:56:17 -0800 Date: Tue, 7 Dec 93 16:56:17 -0800 From: mjf@biosym.com (Mark J Forster ) Message-Id: <9312080056.AA09407@iris120.biosym.com> To: chemistry@ccl.net Subject: Interproton distance averages Dear Comp. Chemists, G. Ravishanker recently asked: > I have a question regarding averaging properties over an MD run - > especially those that fluctuate a lot (specifically interproton distances > in MD simulations). Here is my $0.02 The correct way to average distances for comparison to NMR data is a little complicated. Distance estimates derived from NMR are most often derived from nuclear Overhauser effect (nOe) data. This is a problematic area because of the transfer of magnetization in a molecule that gives rise to nOe data is not simply a function of the separation of the two spins. In reality the direct nOe transfer pathway only dominates at very short mixing times, as the mixing time increases multiple step pathways become increasingly important - this is 'spin diffusion'. One (of many) approaches to dealing with spin diffusion is to record nOe data at multiple mixing times and try to deduce the initial slope of the nOe buildup. This leads to the cross relaxation rate constant and it is this that is from which we may derive 'NMR distances' for comparisons with MD or other simulation methods. The appropriate form of averaging is stongly dependent upon the timescale of the molecular motions involved. For biological macromolecules the timescale of molecular reorientation is of the order of nanoseconds. Thus a molecular dynamics run of (say) 100ps is only sampling motions that are much slower than the molecular reorientation. Under such conditions the equations of J.Tropp (J.Chem.Phys., vol 72, p6035 1980) show the correct form of averaging to be used. In essence it is inverse r cubed averaging with an additional weighting due to angular functions (spherical harmonics) of the orientation of the H-H vector. If the timescale of the conformational interchange is very much slower than molecular reorientation but faster than the timescale for the nOe buildup itself (which is of the order of seconds) then an inverse r sixth average will be appropriate. Such a timescale would not be probed by MD but perhaps monte carlo or other stochastic schemes would be appropriate. If the conformational interchange is much slower than the nOe buildup then one will be observing nOes that are the average of those in each conformation, but it is likely this situation that distinct resonances will be observed for occurences of the two conformers. Some of these issues are discussed in that excellent text 'Neuhaus and Williamson: The nuclear Overhauser effect in structural and conformational analysis, VCH (1989)' I also have a paper appearing in J.Comp.Chem. (should be out in February) that discusses these issues. Finally I should mention that the NMRchitect product >from Biosym Technologies (my employers) allows the computation of various types of distance average for a family of molecular conformation derived by MD or other means. Hope this helps. Best Wishes __________________________________________________________ Mark / ___ / / /| / Mark J Forster / / | / Biosym Technologies Inc, /__/__|/ \__O_/ 9685 Scranton Rd, /__/ | San Diego, CA 92121, USA. / | /| TEL: (619) 546 9990 / o |<-------- / | FAX: (619) 458 0136 /__ __| e-mail: mjf@biosym.com / Follower of MUFC: The Champions. / 'Ryan Giggs is God's gift to football' / / /_________________________________________________________ From apa@pop.cc.adfa.oz.au Tue Dec 7 21:53:55 1993 Received: from sserve.cc.adfa.oz.au for apa@pop.cc.adfa.oz.au by www.ccl.net (8.6.4/930601.1506) id VAA07473; Tue, 7 Dec 1993 21:11:04 -0500 Received: from [131.236.60.11] (apamac.ch.adfa.oz.au) by sserve.cc.adfa.oz.au (5.67a/1.34) id AA04397; Wed, 8 Dec 1993 13:10:43 +1100 Date: Wed, 8 Dec 1993 13:10:43 +1100 Message-Id: <199312080210.AA04397@sserve.cc.adfa.oz.au> To: chemistry@ccl.net From: apa@pop.cc.adfa.oz.au (Alan Arnold) Subject: zindo parameters for Selenium? Seasons Greetings from down-under to comp-chem listeners... I have been doing some zindo/1 calculations with Hyperchem on Co(III) complexes of macrocyclic and macrobicyclic (cage) ligands containing N and S donor atoms. I would like to use zindo to predict the geometries and relative energies of various conformers of some selenium analogs that we've recently synthesised but Hyperchem's zindo doesn't support Se. Can anyone point me in the direction of zindo parameters for Se, or failing that, any other semi-empirical Se params for comparison with sulphur. If necessary I'm prepared to attempt a zindo parameterisation for Se myself - any suggestions for well-characterised Se-containing molecules as a reference set? eg H2Se, MeSeMe, MeSeSeMe ..... I'll summarise responses for the group in the New Year. ---- Alan Arnold | e-mail: apa@pop.cc.adfa.oz.au Chem. Department,University College | voice : +61 6 268 8080 Australian Defence Force Academy | fax : +61 6 268 8002 CANBERRA ACT 2601 Australia |