From chemistry-request@server.ccl.net Wed Sep 1 17:48:19 1999 Received: from helix.nih.gov (helix.nih.gov [128.231.2.3]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id RAA08332 for ; Wed, 1 Sep 1999 17:48:19 -0400 Received: (from mn1@localhost) by helix.nih.gov (8.9.1a/8.9.1) id RAA6473222; Wed, 1 Sep 1999 17:44:29 -0400 (EDT) Date: Wed, 1 Sep 1999 17:44:29 -0400 From: "M. Nicklaus" To: CHEMISTRY@ccl.net cc: "M. Nicklaus" Subject: Backup Experiences Wanted Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear all, This question is only indirectly Comp.Chem. related. But since I'm sure other Comp.Chem. groups are facing similar situations as far as backup (needs) are concerned, here we go. We have a setup with about 20 computers with a variety of Unix variants. These range from nearly 10 Linux/AXP systems, through a few Linux/x86 machines, to several SGI workstations, one DEC AlphaServer, and even the odd OpenStep system. OS's used are Linux/AXP and Linux/x86, from RedHat and Debian, with kernel versions from 2.0.xx to 2.2.xx, IRIX's 6.3-6.5, OSF/1 3.2c, OpenStep 4.2. Total disk space on these systems is currently on the order of 300 GB. We expect this to double within the next 12-18 months, and the number of machines is also likely to go up substantially. What I'm looking for is a unified backup solution for all these systems. In fact, our needs range all the way from mirroring through backup to archiving; but currently, the need for backup is the most urgent one. So here's my QUESTION: What kind of unified backup solution have people found to actually *work* for a setup like this? I'm asking both about the types of hardware used (AIT, DLT, Mammoth, tape changers, RAID farms, CD-R, DVD-RAM, whatever...), as well the software tools (tar, cpio, dump, taper, ftape, afio, Amanda, KBackup, Burt, Arkeia, whatever...) that people are actually using and are happy with, both for backing up and *restoring*. I'm by no means wed to tape technology (we've actually had some bad experiences with previous tape backups when it came to actually *restoring* data), but looking at what's available on the market, it seems that we're most likely to end up with a tape changer of some sort, hooked up to a dedicated machine -- hopefully a Linux box. This system would back up all hard drives (as specified) of all other machines over the network. I'm open to any interesting alternatives to tapes that people might report on. If the (few) Windows 9x PCs we have could be tied into this solution, that would of course not be detrimental. We're also thinking about adding, or even entirely going to, RAID storage, or other NAS/SAN setups. But these are obviously only good in case of disk crashes, not against user stupidity of the kind "I accidentally deleted my entire project yesterday, can you please restore it?" unless they're coupled with some kind of backup mechanism. I'm looking for a solution that's easy to set up yet flexible, reliable, expandable in the future, affordable on the hardware side, and, if possible, free or low-cost on the software side. Licence managers, per-client licenses and the like are quite unwelcome. I'd prefer a solution that's also as little a security risk as possible; i.e., it should not require root privileges all over the place. If there's general interest, I'll be happy to post a summary. Thanks in advance for all responses, Marc ------------------------------------------------------------------------ Marc C. Nicklaus National Institutes of Health E-mail: mn1@helix.nih.gov Bldg 37, Rm 5B29 Phone: (301) 402-3111 37 Convent Dr, MSC 4255 Fax: (301) 496-5839 BETHESDA, MD 20892-4255 USA http://rex.nci.nih.gov/RESEARCH/basic/medchem/mcnbio.htm Laboratory of Medicinal Chemistry, National Cancer Institute ------------------------------------------------------------------------ From chemistry-request@server.ccl.net Wed Sep 1 18:11:35 1999 Received: from mercury.chem.nwu.edu (root@mercury.chem.nwu.edu [129.105.116.2]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id SAA08423 for ; Wed, 1 Sep 1999 18:11:35 -0400 Received: from chem.nwu.edu (kedziora@mordecai.chem.nwu.edu [129.105.14.32]) by mercury.chem.nwu.edu (8.9.1/8.9.1) with SMTP id RAA19428 for ; Wed, 1 Sep 1999 17:11:02 -0500 (CDT) Date: Wed, 1 Sep 1999 17:11:02 -0500 (CDT) From: Gary Kedziora Message-Id: <199909012211.RAA19428@mercury.chem.nwu.edu> To: chemistry@ccl.net Subject: Xeon processor summary X-Status: N X-Mailer: Applixware 4.41 (1021.211.13) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; name="Body" Content-Transfer-Encoding: 7bit I sent this message a while back, but it was automatically rejected by the antispam grepper (I think because I used the word v-a-c-a-t-i-o-n). Here is another try. ORIGINAL QUESTION: Has anyone seen a comparison of the performance of a Xeon processor with a straight Pentium II or III at the same (or nearly the same) clock speed? I'm interested in performance comparisons using the usual chemistry applications, good cache-optimized code-- not implying there is no good cache-optimized chemistry code--or any other benchmark. ----------------------------------------------------------------------------- REPLY 1 (from dino@iris.inc.bme.hu): Hi, I was testing a normal pII450 against a pII450 xeon with 1024 kb cache. The application was Gaussian94 compiled with the f2c-gcc combination, the tests jobs were geom. optimizations and freq. calculations on 10-atom molecules with 70-250 basis functions and the results were disappointing: the xeon was only about 5 % faster, despite the bigger and faster cache. dino dino@iris.inc.bme.hu ------------------------------------------------------------------------------- REPLY 2 (from jochen@uni-duesseldorf.de): The German magazine c't has published SPEC values from a comprehensive comparison of P-III/600, Athlon-600 and some Xeon-450 (?) based machines. Basically the result is, AMD wins every category (x86-only, of course :-), I can look up the numbers or fax you a copy of the article if you want. (Its German, numbers are understandable, though, I guess) -- Jochen -------------------------------------------------------------------------------- MY FINDINGS/THOUGHTS: I found a review on the web with direct comparisons between regular PIIs and PIIIs and corresponding Xeon processors. This used benchmarks I'm not familiar with and may not be very relavent to scientific programs. The results are that Xeon helps by a few percent. The URL for the review is http://www.review-zone.com/hardware/processors/intel_pentium-iii_xeon_500/Page1. html One must have to work hard to squeeze out significant (or more likely marginal) improvement over the PIII. Is it worth the money to get a Xeon? Probably not. I spent some time hunting around Intel's web pages looking for a direct comparison and found nothing obvious. I think that tells you something. AMD's website says "a 550MHz AMD Athlon processor has more than 35 percent higher performance than either a 550MHz Pentium III or a Pentium III Xeon on the SPECfp_base95 benchmark" (http://www.amd.com/news/prodpr/9988.html). AMD now has a 650 MHz version with a 200 MHz bus. That sounds good. ____ Gary Kedziora // Department of Chemistry // Northwestern University email: kedziora@chem.nwu.edu//ph: (847)467-4857//fax: (847)491-7713 From chemistry-request@server.ccl.net Thu Sep 2 07:28:23 1999 Received: from matavnet.hu (matavnet.hu [145.236.224.246]) by server.ccl.net (8.8.7/8.8.7) with SMTP id HAA12253 for ; Thu, 2 Sep 1999 07:28:21 -0400 Received: (qmail 25564 invoked from network); 2 Sep 1999 13:24:17 +0200 Received: from line-212-114.dial.matav.net (HELO chemaxon.com) (145.236.212.114) by mail.matav.hu with SMTP; 2 Sep 1999 13:24:17 +0200 Message-ID: <37CE5F61.E3EB58C3@chemaxon.com> Date: Thu, 02 Sep 1999 13:28:34 +0200 From: Ferenc Csizmadia Organization: ChemAxon X-Mailer: Mozilla 4.61 [en] (WinNT; I) X-Accept-Language: en MIME-Version: 1.0 To: chemistry@ccl.net, chemweb@ic.ac.uk, CHMINF-L@LISTSERV.INDIANA.EDU, mol-diversity@listserv.arizona.edu Subject: Re: Welcome to ECSOC-3 References: <412567DF.006022B2.00@nts1.novartis.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Please excuse multiple postings. Esteban Pombo-Villar wrote: > ECSOC-3 : Third International Electronic Conference on Synthetic Organic > Chemistry http://www.mdpi.org/ecsoc-3.htm > > Dear Colleagues, > > September 1st marks the start of ECSOC-3 , the Third International Electronic > Conference on Synthetic Organic Chemistry. As you will see, the section > conveners have assembled an exciting program, bringing together many aspects of > contemporary interest beyond traditional organic synthesis. I trust you will > find ECSOC-3 stimulating and valuable. As in the previous conferences in the > ECSOC series, we hope a large international audience will be able to read and > react to the contributions. Feel free to comment, ask questions and generally > interact with the participants and section conveners, in what should be an > informal exchange of ideas. We have opted for using e-mail as the mode of > interaction, to ensure that all scientists with access to Internet can > participate in this conference. The authors, the section conveners and the > organising committee welcome you to this Internet scientific event, and hope > that it will contribute to bring the global scientific community together. It > is your participation that will make this into a living event. > > I wish to thank the contributors, the members of the organising commitee, the > section conveners, the members of our international advisory board and the > sponsors whose continued support has made the third edition of this electronic > conference a reality. > > Please forward this mail to your colleagues, who may find ECSOC-3 of interest. > > Cordially yours, > > Esteban Pombo-Villar > > Dr. Esteban Pombo-Villar > > ------------------------------------------------------------------------ > > ECSOC-3 Chairman > http://www.mdpi.org/ecsoc-3.htm > WSJ-507.5.02 Tel. 41-61-3249865 > Novartis Pharma Ltd Fax 41-61-3249794 > CH4002 Basel, Switzerland e-mail: > esteban.pombo@pharma.novartis.com From chemistry-request@server.ccl.net Thu Sep 2 08:39:35 1999 Received: from apollo.syrres.com (apollo.syrres.com [204.168.121.242]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id IAA12525 for ; Thu, 2 Sep 1999 08:39:34 -0400 Received: from pc-citra ([204.168.121.167]) by apollo.syrres.com (Netscape Messaging Server 3.62) with SMTP id 303 for ; Thu, 2 Sep 1999 08:32:09 -0400 Message-Id: <4.1.19990902081220.009397b0@apollo.syrres.com> X-Sender: citra@apollo.syrres.com X-Mailer: QUALCOMM Windows Eudora Pro Version 4.1 Date: Thu, 02 Sep 1999 08:35:22 -0400 To: chemistry@ccl.net From: "Mario Citra" Subject: 3D conformational searches Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" I was wondering what everyone's thoughts are on 3D conformational searching algorithms such as 3DGEN or CONCORD. I am interested in searching for the various low energy conformers of relatively small molecules (20-50 heavy atoms) and using these results to perform single point semi-empirical QM calculations. A few questions come to mind: How close are the geometries of the low energy conformers found by the searching algorithms to geometries obtained by semi-empirical methods ? For example, if chemical X is found to have 10 low energy conformers by 3DGEN and I use these geometries as a starting point in an AM1 calculation can I reasonably expect to be close to a minimum. What software combines 3D conformational searching algorithms with QM calculations ? ********************************** * Regards, * * Mario J. Citra PhD * * Syracuse Research Corporation * * 6225 Running Ridge Road * * North Syracuse, New York 13212 * * * * phone 315-452-8406 * * email citra@syrres.com * ********************************** From chemistry-request@server.ccl.net Thu Sep 2 13:24:52 1999 Received: from mail.chat.ru (gnu.chat.ru [212.24.32.10]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id NAA13853 for ; Thu, 2 Sep 1999 13:24:51 -0400 Received: from [62.76.11.201] (helo=chat.ru) by mail.chat.ru with smtp (Exim 2.12 #3) id 11MaZC-000119-00 for CHEMISTRY@ccl.net; Thu, 2 Sep 1999 21:22:36 +0400 Date: Thu, 2 Sep 1999 23:18:53 +0500 From: "Qadir K. Timerghazin" X-Mailer: The Bat! (v1.036) S/N B2B97075 Reply-To: "Qadir K. Timerghazin" Priority: Normal Message-ID: <13971.990902@chat.ru> To: CHEMISTRY@ccl.net Subject: NBO Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear CCLers, I have some questions regarding NBO method: 1. Is it correct to perform NBO analysis of density obtained from DFT calculation? I mean not only Natural Populational Analysis but also NHO and NBOs. 2. After NBO analysis we obtain a set of NHOs and NBOs constructed from s, p, d, etc. orbitals. What kind of functions describe this basic orbitals? Are these Slater orbitals or simply orbitals of H-atom? If I want to visualize NBOs what fonctions should I use? Could someone clarify to me these points? With best wishes, Qadir mailto:qt@chat.ru ========================================================================= Qadir K. Timerghazin |Email: mailto:qt@anrb.ru Institute of Organic Chemistry,|Web: http://members.tripod.com/~qadir_t Ufa Scientific Center, | Russian Academy of Sciences, |Phone: +7 3472 356011 Prospekt Oktyabrya, 71, | 450054, Ufa, Bashkortostan, | Russia | ========================================================================= From chemistry-request@server.ccl.net Thu Sep 2 13:04:36 1999 Received: from ccl.net (atlantis.ccl.net [192.148.249.4]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id NAA13737 for ; Thu, 2 Sep 1999 13:04:36 -0400 Received: from unomail.unomaha.edu (unomail.unomaha.edu [137.48.1.6]) by ccl.net (8.8.6/8.8.6/OSC 1.1) with SMTP id NAA18173 for ; Thu, 2 Sep 1999 13:00:31 -0400 (EDT) From: Douglas_Stack/CAS/UNO/UNEBR@unomail.unomaha.edu Received: by unomail.unomaha.edu(Lotus SMTP MTA v4.6.4 (830.2 3-23-1999)) id 862567E0.005CE132 ; Thu, 2 Sep 1999 11:54:29 -0500 X-Lotus-FromDomain: UNIVERSITY OF NEBRASKA To: CHEMISTRY@www.ccl.net Message-ID: <862567E0.005CDF29.00@unomail.unomaha.edu> Date: Thu, 2 Sep 1999 11:54:19 -0500 Subject: Summary: allyl vs formate pi system Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii Content-Disposition: inline Below are the comments (thanks to all!) that I received on the question of MO ordering and basis sets with regard to the allyl and formate pi systems. The interesting thing to note, is that the Huckel level (a basis set of three p-orbitals) the order of the MO matches HF at 3-21G and 6-31G*. It was the semi-empirical methods that had a sigma orbital as the HOMO for the formate system, whereas the other methods indicate that the MO with a node at the center carbon was the HOMO for formate. Evidently, the basis set choice is important. Below is first a copy of the original post ant then the responses I received. Thanks again to all those who replied. >> I have question about molecular orbitals (MO) generated by simple Huckel calculations involving just conjugated p-orbitals and MO generated via semi-empirical calculations using all of the atomic valence orbitals. Specifically, the differences in calculating the allyl pi system vs. the formate (HCO2-) pi system. Both systems have 3 p orbitals that combine to form the familiar Huckel orbitals, we'll call them HMO1, HMO2, and HMO3. Since the allyl cation system has just two electrons, HMO1 is the HOMO and HMO2 is the LUMO. If one does a semi-empirical calculation using all valence obitals on the allyl cation system (AM1 or PM3), a graphical representation of the HOMO matches the familiar all bonded HMO1 and the LUMO matches the HMO2. However, with the formate system, a semi-empirical calculation does not produce a HOMO the matches HMO2 (with four electrons in this pi system, HMO2 is the HOMO). The HOMO generated by the semi-empirical method not only has significant coefficients on both oxygens, but also has a large coefficients on the hydrogen (connected to the middle carbon). The HMO2 for formate matches HOMO-1 in the semi-empirical calculation not HOMO. The question is, why does the semi-empirical method for formate generate a HOMO that is different than that predicted by simple Huckel theory? I understand the latter has only 3 atomic orbitals as a basis set while the former has 13 by including all valence orbitals. Can this change the order of MO we are accustom to when we think of Huckel explanations of electron delocalization? Why would the allyl cation system match semi-empirical but not the formate, is symmetry involved? Douglas E. Stack >> As a follow-up to the previous post, if I run a HF/6-31G* calculation on the formate anion, the HOMO matches the Huckel HOMO and the LUMO macthes the Huckel LUMO. Douglas E. Stack >>Dr. Stack, I have no comment on the basis set dependence of the orbital ordering, other than to reiterate your observation that it exists. It is curious why this does not bother more computational chemists. I can point you to a paper that addresses the correspondence just between atomic charges that are "observables" and simple Huckel theory, thus avoiding all basis set artifacts. If you are interested, the reference is Slee and MacDougall, Can. J. Chem., vol. 66, p. 2961 (1988). Sincerely, Preston MacDougall ~~~ Preston J. MacDougall Associate Professor Department of Chemistry, Box X101 Middle Tennessee State University Murfreesboro, TN 37132 >> Doug, Several answers are possible depending the type of explanation you are looking for. First, you need to recognize that HMO2 of formate anion is a sigma orbital and not a pi orbital. This can be seen by graphing the orbital isosurface. It is also "obvious" from the fact that HMO2 contains a contribution from hydrogen; semi-empirical methods do not put p-type orbitals on hydrogen, so hydrogen cannot contribute to a pi-type orbital in this molecule. Short answer: Huckel theory ignores the sigma system. Since chemists are trained using Huckel theory they often mistakenly assume that sigma orbitals are always low-energy and pi orbitals are always the frontier orbitals. Not true! Semi-empirical methods include sigma orbitals and can (and often do) arrive at different conclusions. Look at the HOMO of formaldehyde; it is not a pi orbital. I think a sigma orbital also sneaks into the pi manifold of benzene. A longer answer: The semi-empirical HMO2 of formate anion looks like it contains an antibonding interaction between the oxygens and hydrogen. This can't be too strong since these atoms are fairly well separated. On the other hand, the bonding interactions (CO and CH) in HMO2 appear to be weak. This balance of interactions suggests that HMO2 should be a high energy orbital. Huckel HOMO of allyl anion, by comparison, is a pi nonbonding orbital. We shouldn't be surprised that an orbital with sigma antibonding components has a higher energy than a nonbonding orbital (pi or sigma). An even longer answer: Huckel theory completely ignores the effects of electronegativity and electron-electron interactions. For example, Huckel theory says that the orbital energies of allyl cation and anion are the same. This is wrong. The Huckel HOMO for allyl anion is delocalized over the two end carbons. These carbons are replaced by electronegative oxygens in formate anion, and perturbation theory predicts a significant drop in MO energy. In fact, this orbital appears as HOMO-1 in the semi-empirical calculation. In general, MO energies reflect the energies of the AOs that they are made out of (this is a gross simplification since we also need to think about electron-electron interactions, and whether the MO contains bonding or antibonding orbital interactions). An MO that is composed entirely of oxygen AO might be lower in energy than an orbital that mixes the same type of oxygen AO with a higher energy hydrogen AO. Hope this hasn't confused you too much, -Alan ------------ Alan Shusterman Department of Chemistry Reed College Portland, OR www.reed.edu/~alan >> Dear Prof. Stack, I have checked your results for formate anion and it seems to me that the semiempirical pi-orbitals are in agreement with the Hueckel calculations. The pi-orbitals are HOMO-3, HOMO-1 and HOMO+1. If I understand your question correctly, you had an impression that semiempirical pi-orbitals are ordered differently than Hueckel orbitals. That is, you're not concerned with the relative positions of pi- and sigma- semiempirical orbitals but only of pi- -orbitals. For such a small system it would be really strange if pi-orbitals were ordered differently than Hueckel predicts, but this does not seem to be the case. Maybe you mistook some of the sigma-orbitals as a pi-orbital as it looks like from significant contributions of H-atom orbitals which you mentioned (pi-orbitals [usually!] do not include contributions from non-p atomic orbitals). Pi-orbitals are not necessarilly confined to HOMO-LUMO region, that is, they are more or less interspersed among the sigma-orbitals and only the coefficients enable identification of the pi-orbitals. If you have more questions, let me know. Sincerely, Darko Babic Institute "Rudjer Boskovic" HR-10001 Zagreb, P.O.B. 1016 Croatia >> I suppose a bunch of people told you this, but maybe not. The MOS in formate are degenerate because of symmetry, so any arbitrary linear combination of these is also a solution. Huckel thry gives you the symmetry adapted linear combination if you do it right; computer MO programs cannot be counted on to do this for you. HOwever, if you add & subtract the degenerate ones you DID get, they should give you the 2 you'd like to see... Irene Newhouse >> Douglas E. Stack Assistant Professor Department of Chemistry University of Nebraska at Omaha Omaha, NE 68182-0109 (402) 554-3647 (402) 544-3888 (fax) destack@unomaha.edu