From kamal@univ-orleans.fr Thu Jun 1 03:45:12 1995 Received: from centre.univ-orleans.fr for kamal@univ-orleans.fr by www.ccl.net (8.6.10/930601.1506) id DAA21983; Thu, 1 Jun 1995 03:39:26 -0400 Received: from lcba2.univ-orleans.fr (lcba2.univ-orleans.fr [193.49.81.2]) by centre.univ-orleans.fr (8.6.9/8.6.9) with SMTP id JAA24219 for ; Thu, 1 Jun 1995 09:44:57 +0200 LD?Resent-Date: Thu, 1 Jun 1995 09:44:57 +0200 Message-Id: <199506010744.JAA24219@centre.univ-orleans.fr> X-Sender: kamal@mailhost.univ-orleans.fr X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 01 Jun 1995 09:37:12 +0100 To: chemistry@ccl.net From: kamal@univ-orleans.fr (Kamal Azzaoui) Subject: H-Bonds and Homo Hi Some days ago I asked this question : I would like to know if somebody had any references about relationships between HOMO or LUMO energy level and H-bond acceptor or donor parameters. I received one answer from Tim Mitchell, his paper is interesting thanks Tim *************************************************************************** Well, you could think of this as H-bonding....... I reported the relationship between pKa and HOMO energy in Eur J Med Chem 1990, 25, 117-120. If a base (imidazolidine in this case) is to accept a proton, it must form the covalent bond by transfering charge from it's lone pair. The easier it can do this, the more likely it is to happen, hence the effect on pKa. Most often the lone pair orbital makes up a large part of the HOMO, hence the relationship between pKa and E HOMO. As least that's what we were proposing.... If you think of H-bonding as some part-way state to protonation, then you may find a relationship between E HOMO and H-bond acceptor ability. Please let me know your findings..... Tim Mitchell SmithKline Beecham, Broockham Park, UK. ********************************************************************* ------------------------------------------------------------------ Kamal AZZAOUI LCBA. Universite d'Orleans. BP 6759. 45067 Orleans cedex 2 France Tel: 38-41-70-42 email: kamal@univ-orleans.fr ------------------------------------------------------------------ From SVKK@tifrvax.tifr.res.in Thu Jun 1 06:15:15 1995 Received: from tifrvax.tifr.res.in for SVKK@tifrvax.tifr.res.in by www.ccl.net (8.6.10/930601.1506) id GAA22998; Thu, 1 Jun 1995 06:02:48 -0400 Received: from tifrvax.tifr.res.in by tifrvax.tifr.res.in (PMDF #12198) id <01HR738NPHTS003QC8@tifrvax.tifr.res.in>; Thu, 1 Jun 1995 15:30 +0530 Date: Thu, 1 Jun 1995 15:30 +0530 From: "DR. S. V. K. KUMAR" Subject: Re: CCL:GAMESS-UK To: chemistry@ccl.net Message-id: <01HR738NPHTS003QC8@tifrvax.tifr.res.in> X-VMS-To: IN%"chemistry@ccl.net" Dear Dr. Beveridge, I learnt about GAMESS - UK from your message in the Chenmistry bulletin board. We have been using Gaussian 92 for some time. I would be thankful if you kindly inform us about the availability of the GAMESS - UK and its comparative features with Gaussian. If there is any anonymous ftp site available for getting the program please send me an e-mail to : S. V. K. Kumar Tata Institute of Fundamental Research Bombay 400 005, India e-mail : svkk@tifrvax.tifr.res.in From moon@kistmail.kist.re.kr Thu Jun 1 08:15:26 1995 Received: from garam.kreonet.re.kr for moon@kistmail.kist.re.kr by www.ccl.net (8.6.10/930601.1506) id IAA24748; Thu, 1 Jun 1995 08:05:50 -0400 Received: from kistmail.kist.re.kr (kistmail.kist.re.kr [161.122.12.3]) by garam.kreonet.re.kr (8.6.9H1/8.6.9) with SMTP id VAA21784 for ; Thu, 1 Jun 1995 21:08:26 +0900 Received: by kistmail.kist.re.kr (5.65/KISTMAIL-0.1) id AA05587; Thu, 1 Jun 95 21:05:49 +0900 Date: Thu, 1 Jun 95 21:05:49 +0900 From: moon@kistmail.kist.re.kr (Moon Tae Sung) Message-Id: <9506011205.AA05587@kistmail.kist.re.kr> To: chemistry@ccl.net Subject: Hessian matrix calculation for cartesian coordinate I want to calculate Hessian matrix for cartesian coordinate. Let me know the source code of Hessian matrix calculation for cartesian coordinate. From Tae-Sung Moon From DAREK@ichn.ch.pwr.wroc.pl Thu Jun 1 09:30:17 1995 Received: from sun1000.ci.pwr.wroc.pl for DAREK@ichn.ch.pwr.wroc.pl by www.ccl.net (8.6.10/930601.1506) id JAA27219; Thu, 1 Jun 1995 09:17:14 -0400 Received: from ichn.ch.pwr.wroc.pl (ichn.ch.pwr.wroc.pl [156.17.55.29]) by sun1000.ci.pwr.wroc.pl (8.6.10/8.6.10) with SMTP id PAA01956 for ; Thu, 1 Jun 1995 15:18:22 +0200 Received: from ICHN_1/MAILQUEUE by ichn.ch.pwr.wroc.pl (Mercury 1.13); Thu, 1 Jun 95 15:18:01 +100 Received: from MAILQUEUE by ICHN_1 (Mercury 1.13); Thu, 1 Jun 95 15:17:32 +100 From: "Darek C. Bienko" To: chemistry@ccl.net Date: Thu, 1 Jun 1995 15:17:30 +100 Subject: frequencies for compounds with NO2 group Priority: urgent X-mailer: Pegasus Mail/Windows (v1.22) Message-ID: <32C487232B@ichn.ch.pwr.wroc.pl> hi netters, I am looking for any references about calculated frequencies of compounds with nitro group. I mean ab initio calculations or better DFT ones (especially using blyp functional but not only) Thanks in advance your sincerely D.C. _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ _/ _/ _/ Darek C. Bienko _/ _/ (home) _/ _/Technical University of Wroclaw 10/2 Ladna Str. _/ _/Institute of Inorganic Chemistry 50-353 Wroclaw _/ _/23 Smoluchowskiego Str. _/ _/ _/ _/ _/50-370 Wroclaw; Poland _/ _/ _/ _/ _/mail: darek@ichn.ch.pwr.wroc.pl _/_/_/_/ _/ _/ _/ bienko@indigo.ch.pwr.wroc.pl _/ _/ _/ _/ _/ _/ _/ _/ _/ _/------------------------------------------------------------_/ _/ When all else fails look at the Schroedinger equation _/ _/ R.T.Pack _/ _/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/ From sw3q+@andrew.cmu.edu Thu Jun 1 10:30:48 1995 Received: from po4.andrew.cmu.edu for sw3q+@andrew.cmu.edu by www.ccl.net (8.6.10/930601.1506) id KAA29998; Thu, 1 Jun 1995 10:26:18 -0400 Received: (from postman@localhost) by po4.andrew.cmu.edu (8.6.12/8.6.12) id KAA01297 for chemistry@ccl.net; Thu, 1 Jun 1995 10:26:14 -0400 Received: via switchmail; Thu, 1 Jun 1995 10:26:13 -0400 (EDT) Received: from pcs31.andrew.cmu.edu via qmail ID ; Thu, 1 Jun 1995 10:25:50 -0400 (EDT) Received: from pcs31.andrew.cmu.edu via qmail ID ; Thu, 1 Jun 1995 10:25:48 -0400 (EDT) Received: from mms.4.40.Nov..4.1993.10.34.00.sun4c.411.MacMail.0.9.CUILIB.3.45.SNAP.NOT.LINKED.pcs31.andrew.cmu.edu.sun4c.411 via MS.5.6.pcs31.andrew.cmu.edu.sun4c_411; Thu, 1 Jun 1995 10:25:48 -0400 (EDT) Message-ID: Date: Thu, 1 Jun 1995 10:25:48 -0400 (EDT) From: Stanley Wong To: chemistry@ccl.net Subject: Zinc Chloride solution I am trying to make a solution of zinc chloride in water. All the reference books say zinc chloride has a high solubility in water, but when I made a solution I couldn't get it all to dissolve. I tried heating the solution and still not all the zinc chloride would dissolve. If you have any suggestions I would greatly appreciate it. Thanks. From jpnort@mail.wm.edu Thu Jun 1 10:45:22 1995 Received: from mail.wm.edu for jpnort@mail.wm.edu by www.ccl.net (8.6.10/930601.1506) id KAA00825; Thu, 1 Jun 1995 10:45:10 -0400 Received: from gopher.cc.wm.edu by mail.wm.edu (AIX 3.2/UCB 5.64/4.03) id AA24510; Thu, 1 Jun 1995 10:35:53 -0400 Received: by gopher.cc.wm.edu (AIX 3.2/UCB 5.64/4.03) id AA27915; Thu, 1 Jun 1995 10:35:52 -0400 Date: Thu, 1 Jun 1995 10:35:52 -0400 (EDT) From: "J.P. Northrop" Subject: Plotting Programs To: chemistry@ccl.net Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII I am looking for a programs that will take 3D coordinates and plot them. The input file is an ASCII text of a series of numbers. Does anyone know of a good program and where I could find it? Thanks, J.P. The College of William and Mary jpnort@mail.wm.edu From lecante@cemes.cemes.fr Thu Jun 1 10:49:10 1995 Received: from cix.cict.fr for lecante@cemes.cemes.fr by www.ccl.net (8.6.10/930601.1506) id KAA00464; Thu, 1 Jun 1995 10:37:01 -0400 Received: from cemes.cemes.fr by cix.cict.fr; Thu, 1 Jun 95 16:39:09 +0200 Received: by cemes.cemes.fr; Thu, 1 Jun 95 16:35:35 EST From: lecante@cemes.cemes.fr (Pierre Lecante) Message-Id: <9506011535.AA02886@cemes.cemes.fr> Subject: CCL:Madelung To: chemistry@ccl.net Date: Thu, 1 Jun 1995 16:35:33 +0100 (GMT+0100) X-Mailer: ELM [version 2.4 PL24] Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 833 Dear netters, I'm looking for values of Madelung's constant for different apatitic compounds, and preferably for some software to calculate them. It looks like one was available several years ago from A.G Piken and W. Van Gool but I have currently no path to explore. Archie gave no results and I'd appreciate any informations. I'll summerize the answers. -- +----------------------------------------------------------------+ | Pierre LECANTE | | CNRS/CEMES-LOE Tel : +33 62 25 78 51 | | 29 , rue Jeanne Marvig Fax : +33 62 25 79 99 | | 31055 Toulouse Cedex Email: lecante@cemes.fr | | France | +----------------------------------------------------------------+ From yu@infinity.wavefun.com Thu Jun 1 12:00:19 1995 Received: from volvo.wavefun.com for yu@infinity.wavefun.com by www.ccl.net (8.6.10/930601.1506) id LAA03616; Thu, 1 Jun 1995 11:52:47 -0400 Received: by volvo.wavefun.com (931110.SGI/930416.SGI) for chemistry@ccl.net id AA24717; Thu, 1 Jun 95 08:52:40 -0700 Received: from infinity.wavefun.com(198.147.95.8) by volvo via smap (V1.3) id sma024715; Thu Jun 1 08:52:30 1995 Received: by infinity.wavefun.com (931110.SGI/930416.SGI) for @volvo.wavefun.com:chemistry@ccl.net id AA14927; Thu, 1 Jun 95 08:52:27 -0700 From: "yu" Message-Id: <9506010852.ZM14925@infinity.wavefun.com> Date: Thu, 1 Jun 1995 08:52:27 -0700 Reply-To: yu@volvo.wavefun.com X-Mailer: Z-Mail (3.1.0 22feb94 MediaMail) To: chemistry@ccl.net Subject: C-13 chemical shifts calculations Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 Dear Netters, I am looking for the emperical or semi-emperical methods to calculate C-13 chemical shifts. Can anyone direct me the recently published papers, research or review, on this topics? If therer are enough interests, I will summary it. My email address is yu@wavefun.com Thank you! Jianguo Yu From M.J.C.Crabbe@reading.ac.uk Thu Jun 1 12:04:31 1995 Received: from sums1.rdg.ac.uk for M.J.C.Crabbe@reading.ac.uk by www.ccl.net (8.6.10/930601.1506) id LAA03558; Thu, 1 Jun 1995 11:50:37 -0400 Received: from sksscsc1.rdg.ac.uk by sums1.rdg.ac.uk with SMTP - Local (PP); Thu, 1 Jun 1995 16:50:00 +0100 Received: by sksscsc1.rdg.ac.uk (8.6.12/8.6.12) id QAA09864; Thu, 1 Jun 1995 16:49:55 +0100 Date: Thu, 1 Jun 1995 16:49:54 +0100 (BST) From: James Crabbe To: "J.P. Northrop" cc: chemistry@ccl.net Subject: Re: CCL:Plotting Programs In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII DestTopMolecular Modeller (DTMM) available from Oxford University Press, will do this and a lot more. It is avaialable at about $200; contact: Janet Caldwell, Oxford University Press, Oxford, Walton Street, OX2 6AW, tel : 01865 56767. James Crabbe. On Thu, 1 Jun 1995, J.P. Northrop wrote: > I am looking for a programs that will take 3D coordinates and > plot them. The input file is an ASCII text of a series of numbers. Does > anyone know of a good program and where I could find it? > Thanks, > > J.P. > The College of William and Mary > jpnort@mail.wm.edu > > -------This is added Automatically by the Software-------- > -- Original Sender Envelope Address: jpnort@mail.wm.edu > -- Original Sender From: Address: jpnort@mail.wm.edu > CHEMISTRY@ccl.net -- everyone | CHEMISTRY-REQUEST@ccl.net -- coordinator > MAILSERV@ccl.net: HELP CHEMISTRY | Gopher: www.ccl.net 73 > Anon. ftp www.ccl.net | CHEMISTRY-SEARCH@ccl.net -- archive search > http://www.ccl.net/chemistry.html | for info send: HELP SEARCH to MAILSERV > > From craig@hobbes.gh.wits.ac.za Thu Jun 1 12:16:42 1995 Received: from hobbes.gh.wits.ac.za for craig@hobbes.gh.wits.ac.za by www.ccl.net (8.6.10/930601.1506) id MAA04134; Thu, 1 Jun 1995 12:11:12 -0400 Received: (from craig@localhost) by hobbes.gh.wits.ac.za (8.6.11/8.6.9) id SAA09762; Thu, 1 Jun 1995 18:08:24 +0200 Date: Thu, 1 Jun 1995 18:08:23 +0200 (GMT+0200) From: Craig Taverner To: "J.P. Northrop" cc: chemistry@ccl.net Subject: Re: CCL:Plotting Programs In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII > I am looking for a programs that will take 3D coordinates and > plot them. The input file is an ASCII text of a series of numbers. Does > anyone know of a good program and where I could find it? > Thanks, If your data is molecular then a good program would be SCHAKAL from Egbert Keller (kell@ruf.uni-freiburg.de). It is very good at plotting molecules using various ball-and-stick and space filling models. If your data is inorganic crystal data, then ATOMS is good (don't know the source off-hand). If your data is organic crystal data, then the standard is ORTEP, available in hundreds of versions/formats for various machines from many sources. There's also PLUTO and many others. Since I'm a crystallographer, I'm sure I don't know of many other purely molecular packages out there. I'd guess there are VERY MANY. Cheers, Craig "If God had meant us to be naked, we would have been born that way." Craig Taverner Structural Chemistry, University of the Witwatersrand, South Africa From sahmed@beavis.chemse.gatech.edu Thu Jun 1 13:30:20 1995 Received: from beavis.chemse.gatech.edu for sahmed@beavis.chemse.gatech.edu by www.ccl.net (8.6.10/930601.1506) id NAA06622; Thu, 1 Jun 1995 13:26:21 -0400 Received: by beavis.chemse.gatech.edu (940816.SGI.8.6.9/940406.SGI) for chemistry@ccl.net id NAA14287; Thu, 1 Jun 1995 13:26:49 -0400 From: "Savant Ahmed" Message-Id: <9506011326.ZM14285@beavis.chemse.gatech.edu> Date: Thu, 1 Jun 1995 13:26:48 -0400 X-Mailer: Z-Mail (3.2.0 26oct94 MediaMail) To: chemistry@ccl.net Subject: Gaussian94 Advantages? Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hi, We are running Gaussian92 on a SGI indigo**2 with 1GB scratch space and 200Meg swap. Our jobs are big (MP2 freq) and we keep running out of memory. Gaussian claims to have made memory management/requirements more efficient, especially for MP2 and higher routines, in G94. We would like to upgrade to G94 if this is the case. Would be grateful for suggestions/comments. Savant From jle@toyota.wavefun.com Thu Jun 1 14:15:30 1995 Received: from volvo.wavefun.com for jle@toyota.wavefun.com by www.ccl.net (8.6.10/930601.1506) id OAA08038; Thu, 1 Jun 1995 14:00:27 -0400 Received: by volvo.wavefun.com (931110.SGI/930416.SGI) for chemistry@ccl.net id AA25018; Thu, 1 Jun 95 10:42:48 -0700 Received: from toyota.wavefun.com(198.147.95.4) by volvo via smap (V1.3) id sma025014; Thu Jun 1 10:42:35 1995 Received: by toyota.wavefun.com (931110.SGI/930416.SGI) for @volvo.wavefun.com:chemistry@ccl.net id AA14400; Thu, 1 Jun 95 10:42:32 -0700 From: "Joe Leonard" Message-Id: <9506011042.ZM14398@toyota.wavefun.com> Date: Thu, 1 Jun 1995 10:42:32 -0700 Reply-To: jle@wavefun.com X-Mailer: Z-Mail (3.1.0 22feb94 MediaMail) To: chemistry@ccl.net Subject: Seeking Boron params for PM3 Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 Folks, Has anybody published PM3 parameters for Boron? If so, are they on-line somewhere? Also, does anybody have any experience whether the AM1 or MNDO Boron parameters are useful when combined with the rest of the PM3 set? I know all about the philosophical/religious reasons why one should NOT do this, but I am interested in hearing whether anybody has tried - and whether the combination was useful, useable or garbage... Thanks in advance, Joe -- ------------------------------------------------------------------------ Joe Leonard Wavefunction Inc. 18401 Von Karman, Suite 370 Irvine, CA 92715 I am a professional... 714-955-2120 do not attempt this at home. 714-955-2118 fax jle@wavefun.com From CHEM8@VAX.YORK.AC.UK Thu Jun 1 15:00:25 1995 Received: from sun2.nsfnet-relay.ac.uk for CHEM8@VAX.YORK.AC.UK by www.ccl.net (8.6.10/930601.1506) id OAA09137; Thu, 1 Jun 1995 14:54:29 -0400 Message-Id: <199506011854.OAA09137@www.ccl.net> Via: uk.ac.york.vax; Thu, 1 Jun 1995 19:32:15 +0100 Date: Thu, 1 Jun 95 19:31 BST From: "John Waite, Tel: ++30-1-7238958, N.H.R.F., Organic Chemistry Institute, Vas. Konstantinou 48, Athens 116-35" To: CHEMISTRY Subject: Di-carbonyl Di-anions. refs. Hi Netters, I would be grateful if any CCLer would send me any references concerning the multiplicity (singlet or triplet?) of organic, particularly aromatic, di-carbonyl di-anions. In particular di- (or poly-) benzo-phenone derivatives. Hopefully, John Waite From qocruces@usc.es Thu Jun 1 15:30:22 1995 Received: from uscmail.usc.es for qocruces@usc.es by www.ccl.net (8.6.10/930601.1506) id PAA10009; Thu, 1 Jun 1995 15:28:21 -0400 Received: from qobrue.usc.es by uscmail.usc.es (5.67b/1.34) id AA07262; Thu, 1 Jun 1995 21:32:18 +0200 Date: Thu, 1 Jun 1995 21:32:18 +0200 Message-Id: <199506011932.AA07262@uscmail.usc.es> X-Sender: qocruces@uscmail.usc.es X-Mailer: Windows Eudora Version 1.4.4 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: CHEMISTRY@ccl.net From: qocruces@usc.es (Jacobo Cruces) Subject: CCL: What's TNDO? Dear netters: In the demo of Hyper NMR they talk about something called Typed Neglect of Differential Overlap. Can somebody tell me what is this new method? Semiempirical, parametrized with what? All answers will be welcomed. Jacobo Cruces Colado Department of Organic Chemistry University of Santiago de Compostela Avda. de las Ciencias s/n 15706 SPAIN e-mail: qocruces@uscmail.usc.es jcruces@sfdzuma.usc.es From frisch@lorentzian.com Thu Jun 1 15:45:26 1995 Received: from relay3.UU.NET for frisch@lorentzian.com by www.ccl.net (8.6.10/930601.1506) id PAA10296; Thu, 1 Jun 1995 15:41:31 -0400 Received: from uucp1.UU.NET by relay3.UU.NET with SMTP id QQysha11133; Thu, 1 Jun 1995 15:41:26 -0400 Received: from m10.UUCP by uucp1.UU.NET with UUCP/RMAIL ; Thu, 1 Jun 1995 15:41:27 -0400 Received: from mjf by m10.lorentzian.com; Thu, 1 Jun 1995 15:22:25 -0400 Received: by mjf (AIX 3.2/UCB 5.64/4.03) id AA20919; Thu, 1 Jun 1995 15:24:22 -0400 From: frisch@lorentzian.com (Mike Frisch) Message-Id: <9506011924.AA20919@mjf> Subject: Re: CCL:Gaussian94 Advantages? To: chemistry@ccl.net Date: Thu, 1 Jun 1995 15:24:22 -0400 (EDT) In-Reply-To: <9506011326.ZM14285@beavis.chemse.gatech.edu> from "" at Jun 1, 95 01:26:48 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1065 Savant Ahmed writes: > > Hi, > > We are running Gaussian92 on a SGI indigo**2 with 1GB scratch space and 200Meg > swap. Our jobs are big (MP2 freq) and we keep running out of memory. Gaussian > claims to have made memory management/requirements more efficient, especially > for MP2 and higher routines, in G94. We would like to upgrade to G94 if this is > the case. Would be grateful for suggestions/comments. > > Savant > The improvements in Gaussian 94 with respect to MP2 frequencies are: 1. Better performance. 2. Moderate reduction in disk usage. 3. Since the rwf can span filesystems, the ability to use several scratch disks and go beyond 2 GBytes of scratch data. So if you're running out of disk in 1 GByte, things will be somewhat better in G94, but you'll still run out of disk before you run out of CPU cycles. The best solution is to get another disk. Fortunetly, 2 GByte and 4 GByte SCSI disks are pretty cheap these these days, and adding third party SCSI disks to SGI machines is straightforward. Mike Frisch frisch@lorentzian.com From Karl.F.Moschner@urlus.sprint.com Thu Jun 1 21:00:25 1995 Received: from sprintf.merit.edu for Karl.F.Moschner@urlus.sprint.com by www.ccl.net (8.6.10/930601.1506) id UAA14703; Thu, 1 Jun 1995 20:47:09 -0400 From: X400-Received: by /PRMD=LANGATE/ADMD=TELEMAIL/C=GB/; Relayed; Thu, 1 Jun 1995 20:43:00 -0400 X400-Received: by /PRMD=LANGATE/ADMD=TELEMAIL/C=GB/; Relayed; Thu, 1 Jun 1995 20:43:00 -0400 X400-Received: by /PRMD=LANGATE/ADMD=TELEMAIL/C=GB/; Relayed; Thu, 1 Jun 1995 20:43:00 -0400 X400-Received: by /PRMD=LANGATE/ADMD=TELEMAIL/C=GB/; Relayed; Thu, 1 Jun 1995 20:43:00 -0400 Date: Thu, 1 Jun 1995 20:43:00 -0400 X400-Originator: Karl.F.Moschner@urlus.sprint.com X400-Recipients: non-disclosure:; X400-MTS-Identifier: [/PRMD=LANGATE/ADMD=TELEMAIL/C=GB/;Thu Jun 1 20:43:18 199501] X400-Content-Type: P2-1984 (2) Content-Identifier: Gaussian94 Advan Message-ID: <"Thu Jun 1 20:43:18 199501*/I=F/G=Karl/S=Moschner/OU=4267EDUR/O=TMUS.URL/PRMD=LANGATE/ADMD=TELEMAIL/C=GB/"@MHS> To: CHEMISTRY@ccl.net Subject: Gaussian94 Advantages vs. Gaussian 92 Since questions regarding Gaussian have resurfaced I thought this commentary might be helful. We carry out Gaussian 92 and now 94 MP2/6-31G(d) calculations regularly and have run into most of the problems many CCL participants have been raising. Because of resource demands and our desire to work on more realistic systems, we restrict higher level Gaussian calculations to more robust systems: SGI Power Challenge 4xR8000, 512 MB RAM, 16 GB scratch space SGI Power Indigo2 R8000, 128 MB RAM, 4 GB scratch space SGI Power Indigo2 R8000, 96 MB RAM, 2 GB scratch space And I have run calculations up to 128 basis functions restricted to as little as 2 MB with differing degrees of parallelism on the Power Challenge to evaluate factors impacting run times. Gaussian 92/94 will try to select the optimum procedures within the specified memory limitation by trading off RAM and scratch storage. But there are absolute minimum requirements before you start to swap, at which point your system grinds to a crawl, or you run out of disk space and the job crashes. Keep in mind that the estimates offered are rough approximations based on information in the "Gaussian 92 User's Guide". Storage estimates are in good agreement with my experience in using Gaussian 92. Gaussian 94 may have implemented some storage reductions but I haven't noticed dramatic reductions in the past 1-2 months and have not conducted direct comparisons. Memory ------ In principle, the approximate minimum free memory requirements are: Basis Functions Units 100 150 200 250 300 ----------------------------------------------- SCF In-core, HF GB 0.1 0.5 2 4 8 UHF/ROHF GB 0.2 1.0 3.2 8 16 MP2 Direct, RAM MB 2 6 16 32 64 MP2 Semi-direct MB 3 9 28 48 80 ----------------------------------------------- Direct SCF calculations require even less direct MP2. While 16-32 MB of RAM beyond the system needs is adequate for most jobs, we find that additional memory about double the minimum offers a 10-100% speedup in CPU time and helps minimize I/O. Scratch Space ------------- Generally, scratch storage is more critical than memory, especially for MP2 calculations, in determining the absolute job size limit. To minimize scratch files, you can try the following options which successively reduce your job limit within the 2.0 GB single file size and 8-16 GB file system size limitation of many popular workstation operating systems: Options Approximate Maximum Number of Basis Functions Within 2.0 GB Limit --------------------------------------------------------------------- SCF=Direct 220 ...MP2 [single point] 220 ...Freq 180 ...MP2=Direct Opt=CalcHFFC 180 ...MP2=(Direct,Stingy) Opt=CalcHFFC Freq=NoRaman 165 ...MP2=(Direct,Stingy) Opt=CalcHFFC Freq 150 ...MP2=(Direct) Freq 135 --------------------------------------------------------------------- Currently, DEC along with Cray, Convex and possibly others I am less familiar with, offer UNIX operating systems which do not have these file and file system size limitation. Presumably SGI, IBM, HP, SUN and other UNIX system vendors are working on implementing extended file systems which will overcome these restrictions but user demand has obviously gotten ahead of the suppliers. In response, Gaussian 94 has implemented user definable temporary 'rwf', 'int' and 'd2e' scratch file names. In addition, 'rwf' output, usually the largest, may be parsed over multiple files. This capability greatly increases the job size that may be tackled by allowing you to more effectively distribute output over multiple files and multiple file systems but the amount of output, to my knowledge, is about the same. You still need the free disk space! So, to look at the problem another way: Approximate Disk Space Required (MB) Basis - SCF - -------- MP2 --------- Functions Freq. Direct Freq. STINGY --------- ------- ---------------------- 60 26 12 45 80 82 36 145 --------- ------- ---------------------- 100 200 90 350 120 415 185 725 140 770 340 1350 160 1300 580 2300 180 2100 1000 3700 --------- ------- ---------------------- 200 3200 1400 5600 220 4700 2100 8200 240 6600 2900 12000 260 9000 4100 16000 280 12000 5500 22000 300 16000 7200 28400 --------- ------- ---------------------- Direct SCF storage requirements are similar to direct MP2. Gaussian 94 vs. 92 ------------------ While there may be some reduction in the scratch output, you will probably still have to contend with the scratch space requirements. However, Gaussian 94 allows you to distribute the load over multiple files and multiple file systems. So stock up on FAST disks (see commentary below). They're still much cheaper than memory or more CPU's. There is some CPU speedup particularly for paralell SCF and parallel MP2. No, Gaussian 94 isn't 2x faster than Gausisan 92 but I have gotten 20% on some calculations. Gaussian 94 implements several new features. The contiuum reaction-fields for SCF and DFT optimizations (MP2 energies only) may be of interest and there are significant improvements to the built-in basis sets. Lastly, Gaussian 94 installation and compilation did procede run much more smoothly than the last Gaussin 92 release. Commentary ---------- The large scratch file space requirements of Gaussin 92/94 for larger jobs easily makes I/O a significant portion of the total run time particyularly with the newer faster processors. Combined with the memory demand, I/O performance can significantly impact the throughput. For this reason, it is essential, in most sitiuations, that the scratch file space be local - NFS over standard Ethernet is a killer! Experience with our Power Challenge running a large number MP2 optimizations and frequency calculations generating 1-2 GB of scratch files helps to illustrate the problem. An external striped array of single-ended fast-wide SCSI-2 disks on a single SCSI-channel limited data transfer to 10 MBs with actual rates of about 8 MBs. Analysis of run times with this configuration suggested that I/O accounted for 30-50% of the total run times. Implementation of dual striped internal differential fast-wide SCSI-2 scratch file system increased transfer rates to 16-18 MBs and decraesed I/O to 10-20% of the total run time. Recent installation of an external dual-channel differential fast-wide 4 disk array boosted transfer rates to about 36 MBs with I/O now consistently <<10% of the total run time. The latest I/O configuration ensures efficient performance in tackling larger problems. Note that such heavy I/O easily swamps caches, so RAID controllers with supplemental memory offer no benefits. I/O requirements will grow exponentially with the problem size. Therefore others can similarly expect to find that scratch storage and data transfer rates may be limiting factors in dealing with larger problems especially with PC systems, either under Windows or Linux (no, I don't know of a port). Users would be well advised to consider systems offering fast-wide differential SCSI interfaces and/or multiple channel I/O capabilities for tackling more substatntial calculations. Other more expensive, high-performance options such as HIPPI are also available but generally unwarranted for desktop systems. The prognosis for the near future is actually rosey. HP has announced a new disk controller capable of supporting >=100 MBs. So, with a bit of luck, improved I/O options may appear on the desktop in time to match the next generation of UNIX processors and improved operating systems which support unlimited file sizes. Hope this helps! For specific information, try contacting Gaussian Inc. directly through info@gaussian.com. Also, the "Gaussian 94 Users's reference" may prove very helpful. Karl _______________________________________________________________________ / \ | Comments are those of the author and not Unilever Research U. S. | | | | Karl F. Moschner, Ph. D. | | | | Unilever Research U. S. e-mail: Karl.F.Moschner@urlus.sprint.com | | 45 River Road Phone: (201) 943-7100 x2629 | | Edgewater, NJ 07020 FAX: (201) 943-5653 | \_______________________________________________________________________/