From chemistry-request@server.ccl.net Wed Jun 19 04:04:38 2002 Received: from iris (mail@[152.66.63.5]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g5J84b521464 for ; Wed, 19 Jun 2002 04:04:38 -0400 Received: from janka (helo=localhost) by iris with local-esmtp (Exim 3.35 #1 (Debian)) id 17KaSD-0007S5-00 for ; Wed, 19 Jun 2002 10:04:37 +0200 Date: Wed, 19 Jun 2002 10:04:36 +0200 (CEST) From: Matrai Janka To: chemistry@ccl.net Subject: charges Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hello Everybody, does somebody know a source of charges for methoxy group on aromatic ring? I would like to perform MD simulations with GROMOS and I have to make the topology file first. For that purpose I need charges. Thank you for the help in advance Janka From chemistry-request@server.ccl.net Wed Jun 19 10:17:19 2002 Received: from sg7.chemie.uni-konstanz.de ([134.34.7.80]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g5JEHI527189 for ; Wed, 19 Jun 2002 10:17:18 -0400 Received: from sg8.chemie.uni-konstanz.de (sg8.chemie.uni-konstanz.de [134.34.7.81]) by sg7.chemie.uni-konstanz.de (8.9.3/8.9.3) with ESMTP id QAA1232753 for ; Wed, 19 Jun 2002 16:13:48 +0200 (CEST) Received: from localhost (hein@localhost) by sg8.chemie.uni-konstanz.de (8.9.3/8.9.3) with ESMTP id QAA592609 for ; Wed, 19 Jun 2002 16:13:48 +0200 (MDT) Date: Wed, 19 Jun 2002 16:13:47 +0200 From: Frank Schaper Reply-To: To: Subject: Q: DFT on open shell systems Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear CCLers, I have tried to calculate some organometallic complexes with unpaired electrons (Turbomole). I'm a little bit surprised by the results, however. In particular the calculated relative energies were much more sensitive to the functional used (VWN vs. BP), changing form +100 kJ/mol to -100 kJ/mol, compared to my experiences with closed shell systems. I would thus appreciate, if someone could help me with the following questions: - Is a strong dependence on the DFT-functional a "normal" behavior for open-shell systems? - Is there any way to take a shortcut for geometry optimizations? E.g. calculating the geometry closed-shell and using single-point energies for different open-shell spin states. - Is a geometry optimization mandatory for every spin state (possible in my case S=1, 3 and 5) or are single-point calculations sufficient? I'm very much afraid they are not. - Can the calculated energies of the same complex with different spin states be compared directly or is the spin-pairing-energy missing to some extend? Thanks in advance Hein -- Frank Hein Schaper - AG Brintzinger - Universitaet Konstanz email : Frank.Schaper@uni-konstanz.de - Frank@Schaper.com mail@FrankSchaper.de web : www.FrankSchaper.de s-mail : Frank Schaper - AG Brintzinger - Universitaetsstr. 10 78434 Konstanz From chemistry-request@server.ccl.net Wed Jun 19 10:27:37 2002 Received: from ivory.trentu.ca ([192.75.12.103]) by server.ccl.net (8.11.6/8.11.0) with ESMTP id g5JERb527426 for ; Wed, 19 Jun 2002 10:27:37 -0400 Received: from trentu.ca ([206.248.117.30]) by trentu.ca (PMDF V5.2-32 #37862) with ESMTP id <01KJ412Q04TW000CJY@trentu.ca> for chemistry@ccl.net; Wed, 19 Jun 2002 10:27:24 EST Date: Wed, 19 Jun 2002 10:29:37 -0400 From: elewars Subject: SUMMARY of Direct SCF, RAM To: chemistry@ccl.net Message-id: <3D109551.6797F53B@trentu.ca> MIME-version: 1.0 X-Mailer: Mozilla 4.79 [en] (WinNT; U) Content-type: text/plain; charset=us-ascii Content-transfer-encoding: 7bit X-Accept-Language: en 2002 June 19 Here is the summary of responses to my direct SCF/RAM question. Thanks very much to all who wrote. EL ----------- QUESTION 2002 June 14 Does it seem likely to everyone out there that the introduction of very big (and fast-access) electronic memories (RAM) will make direct SCF obsolete? ================= RESPONSES #1 jle@theworld.com I know Warren felt that way 10+ years ago when he (and his group) did Spartan with in-core stuff. Direct/recalc methods vectorized well, but the newer machines don't program that way - we might see a change if/when people overcome the intertia and rewrite stuff. Joe ============== #2 Joop van Lenthe Theoretical Chemistry Group Padualaan 14 3584CH Utrecht The Netherlands Email joop@chem.uu.nl Joop van Lenthe Yes, in the sense that the calculations we used to run with direct scf now can bw run incore, However since the cpu's get fater all the time as well, there are always calculations that won't fit in memory. I tried this with a 500 processor (1gbyte/processor) SGI. Joop ====================== #3 Alan.Shusterman@directory.reed.edu (Alan Shusterman) I haven't kept up with hardware memory developments so I would like to know how big a memory you are talking about? I remember when I bought my first 128M workstation back in 1991. I couldn't go above ~100 basis functions in an RHF model w/o using direct SCF (and MP2 models were MUCH smaller). Assuming quadratic scaling and a 1000 basis function RHF model (not that large by today's standards), I would need about 13G of RAM to avoid direct SCF. Is this an option? How expensive is it? -Alan [2 GB of RAM is now readily available for Pentiums. I expect 10 GB or so in a few years.--EL] ====================== #4 "Van Dam, HJJ (Huub)" Organization: CCLRC Daresbury Laboratory You might be right that traditional direct SCF (as in recalculating all integrals every SCF cycle) could be on the defense. However the number of integrals required still tends to grow fast with the system size, making it unlikely that all of them can be stored. The most successful approach I know of is to store all integrals with large values or which are particularly expensive to compute in memory and to recalculate the remaining ones if needed. This approach makes the best use of the resources available without the restricting requirement that all integrals MUST fit in memory. Huub ========================================================================