From mn1@helix.nih.gov Tue Nov 25 10:37:16 1997 Received: from helix.nih.gov for mn1@helix.nih.gov by www.ccl.net (8.8.3/950822.1) id JAA12244; Tue, 25 Nov 1997 09:51:45 -0500 (EST) Received: (from mn1@localhost) by helix.nih.gov (8.8.5/8.8.5) id JAA15474; Tue, 25 Nov 1997 09:51:29 -0500 (EST) Date: Tue, 25 Nov 1997 09:51:29 -0500 (EST) From: "M. Nicklaus" Message-Id: <199711251451.JAA15474@helix.nih.gov> To: CHEMISTRY@www.ccl.net Subject: Re: G:Chemistry on beowulf systems? Cc: mn1@helix.nih.gov On Mon, 24 Nov 1997 15:26:14, Chris Harwell wrote: > Are there any chemists out there who are using a beowulf cluster > of pcs type of system for computing? > We are trying to gather some information about software success > stories and hardware choices for chemistry applications. > > Has anyone used gamess, gaussian, or Wien FLAPW on such a system? > What hardware choices did you make (processor. motherboard, mem, > hard disk, ethernet/fast/atm)? > What compiler did you use (gcc f2c, g77)? > What OS/Distribution did you use (linux/ redhat or debian?) > Any hindsight tempered suggestions to someone considering building such a > system? Yes. Check out http://www.lobos.nih.gov. Marc ------------------------------------------------------------------------ Marc C. Nicklaus National Institutes of Health E-mail: mn1@helix.nih.gov Bldg 37, Rm 5B29 Phone: (301) 402-3111 BETHESDA, MD 20892-4255 USA Fax: (301) 496-5839 http://rex.nci.nih.gov/RESEARCH/basic/medchem/mcnbio.htm Laboratory of Medicinal Chemistry, National Cancer Institute, & Lab. of Structural Biology, Div. of Computer Research and Technology ------------------------------------------------------------------------ From kotelyan@plmsc.psu.edu Tue Nov 25 10:49:19 1997 Received: from plmsc.psu.edu for kotelyan@plmsc.psu.edu by www.ccl.net (8.8.3/950822.1) id KAA12360; Tue, 25 Nov 1997 10:20:46 -0500 (EST) Received: (from kotelyan@localhost) by plmsc.psu.edu (8.8.2/8.8.2) id KAA23341; Tue, 25 Nov 1997 10:20:38 -0500 (EST) Date: Tue, 25 Nov 1997 10:20:38 -0500 (EST) From: Mike Kotelyanskii To: "Dr. Lutz Ackermann" cc: CHEMISTRY@www.ccl.net Subject: Re: CCL:CERIUS liquid phase heat capacity calcs In-Reply-To: <347A7F1E.17B5@wanadoo.fr> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Dear Lutz, I think that your message is a perfect example of how misleading the simulations package manual, written in a sort of "plug and play" fashion can be. Yes, the heat capacity is quite tricky for various reasons: 1) It is very sensitive to the force field. And force fields (at least those that I know) are not fitted for the heat capacity but rather for the structural parameters and densities and energies. 2) How is the heat capacity estimated in Cerius, if it is estimated from the fluctuatoins, then there are following sources of errors: 1) How is constant T constant P implemented: if it is the vanGunsteren Berendsen thermostat, which is very practical to use, but it does not reproduce the ensemble, therefore the averages are OK, but the fluctuations are not. 2) If it is not the case, your MD run can be not long enough, you need to collect much more data to estimate the fluctuation, than for the averages like energy, pressure, etc. 3) If the molecules you are simulating are polar, it is very important how is electrostatics calculated, is it Ewald summation, or the cut-off?? and where do the partial charges come from? (see above about the force-field) Unfortunately, I am not aware of any systematic study for the heat capacities, usually it is the heat of evaporation (or cohesive energy) which is reported and compared to experiments. So, if you need a heat capacity you may be more lucky by calculating energies (enthalpies) at different temperatures (pressures) and then getting the heat capacity from their dependencies upon T. And finally: Liquid densities within several percent is VERY BAD!!! Remember, liquids are virtually INCOMPRESSIBLE, and 1% deviation in density is equivalent to the kbar difference in pressure, and tens degrees difference in temperature.... And one more comment (risking to cause a lot of flames) IMHO the simulation is more useful (and more trustworthy and reliable) when comparing the differences between different systems, than the absolute values, so if you are interested in differences between different systems you may be more lucky in comparison with experiments, than in the absolute values. Sorry for the disappointment, but I hope it helps. Michael. On Mon, 24 Nov 1997, Dr. Lutz Ackermann wrote: > Dear Netters, > > I would like to draw on your experience concerning liquid simulation. I > have > spent the last three month experimenting (I am a novice in the field of > MM/MD simulations), trying to solve a problem, but am am not yet > satisfied. > > I am trying to reproduce isobaric heat capacity (Cp) data of single > component fluid phase systems (eg. liquid H2SO4), using the MSI > CERIUS2_3.0 > (OFF) software. MD runs (NPT ensemble) of typically > 50ps(equillibration)+120ps(observation) invariably produce Cp values, > that > are higher than the experimental ones (in general by 30 to 50 percent, in > some cases ut to a factor of 3!). This is true for: > - more or less complex molecules (3 to 10 atoms) > - different force fields > - different temperatures > - ensembles of different sizes (up to 500 molecules in the unit cell) > > At the same time, other properties like the density are calculated to a > satisfactory degree of accuracy (only a few percent difference to > experimental values). > > Maybe I am expecting to much, but the fact that CERIUS offers the > possibility of Cp calculations lead me to believe, that this is actually > possible (yielding reliable data, I mean). Are there any experts around, > who > have managed to calculate realistic heat capacities of liquids (using > CERIUS)? Or are there principal reasons, why this doesn't work? Or can > you > think of some silly mistake I may have made? Maybe you are aware of any > publications, that report the successful use of CERIUS in the calculation > of > the Cp of liquids (although I was told by MSI support that such > publications > do not exist, as "this is a tricky area"...)? Any other helpful > information? > > Thanks a lot, > > Lutz Ackermann (101,21869@germany.net / lutz@ri.ac.uk) > > -------This is added Automatically by the Software-------- > -- Original Sender Envelope Address: lutz.ackermann@wanadoo.fr > -- Original Sender From: Address: lutz.ackermann@wanadoo.fr > CHEMISTRY@www.ccl.net: Everybody | CHEMISTRY-REQUEST@www.ccl.net: Coordinator > MAILSERV@www.ccl.net: HELP CHEMISTRY or HELP SEARCH | Gopher: www.ccl.net 73 > Anon. ftp: www.ccl.net | CHEMISTRY-SEARCH@www.ccl.net -- archive search > Web: http://www.ccl.net/chemistry.html > > From richard@tc.cornell.edu Tue Nov 25 10:51:35 1997 Received: from theory.tc.cornell.edu for richard@tc.cornell.edu by www.ccl.net (8.8.3/950822.1) id JAA12203; Tue, 25 Nov 1997 09:48:03 -0500 (EST) Received: from ren.tc.cornell.edu (REN.TC.CORNELL.EDU [128.84.244.22]) by theory.tc.cornell.edu (8.8.4/8.8.3/CTC-1.0) with SMTP id JAA34219; Tue, 25 Nov 1997 09:48:03 -0500 Sender: richard@TC.Cornell.EDU Message-ID: <347AE523.794B@tc.cornell.edu> Date: Tue, 25 Nov 1997 09:48:03 -0500 From: Richard Gillilan X-Mailer: Mozilla 2.02 (X11; I; IRIX 5.3 IP22) MIME-Version: 1.0 To: chemistry@www.ccl.net CC: richard@TC.Cornell.EDU Subject: rigid body summary Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Many thanks to those who responded! I have had some very useful conversations. (I have shortened some of the responses in some cases as indicated by ... ) --------------------------------------------------------- My original post: A number of programs (like XPLOR) have implemented the virtual rigid body dynamics of T. Head-Gordon and C. Brooks (Biopolymers vol 31, 77-100 (1991). I've seen a few notes on CCL about "Moldy" that sound like it too can do this. Does anyone have info about the program? Also: 1. Does anyone know of recent studies using this technique, particularly applied to the protein-protein docking? 2. How does it compare with SHAKE type constraint methods: faster/slower or just different? -------------------------------------------------------- From: me (richard@tc.cornell.edu) Moldy, XPLOR, and recent versions of CHARMM and INSIGHT seem to support various kinds of rigid-body dynamics. MBO(N)D sounds particularly intriguing, see comments >from moldyn below. Here's somthing from NASA about it too: --- NASA Success Story (http://ntas.techtracs.org/4d.acgi$w3SuccItem(1000197)) DRUG RESEARCH NBOD2 is a computer program developed at NASA's Goddard Space Flight Center to solve equations of the motion of systems consisting of many objects. The program is used by DuPont to model the process of the binding between drugs and enzymes to help develop new drugs. Enzymes are complex proteins that help certain chemical reactions in the body. NBOD2 helps to determine what the properties of a new drug will be. It was supplied to DuPont by NASA's Computer Software Management and Information Center, or COSMIC(R). Now on to the replies: --------------------------------------------------------- From: "Gustavo A. Mercier Jr" Moldy was develope by Keith Refson in the UK and is available by ftp. It is written in C and is highly vectorized/parallelized. Also, it has been implemented in many machines. The code does not implement the virtual rigid body dynamics of H-G and Brooks. I remember browsing over their paper, but remember little of the details. So, I will stick to describing moldy. Moldy is intended for rigid body dynamics of molecules in liquids and crystal systems. It also incorporates the idea of a framework. This is like a slab of a charged crystal above which you have a group of atoms/molecules that move in an MD simulation. I guess that a framework could represent one face of a macromolecule above which a ligand is trying to find the binding site. Needless to say that was not the intended application. Moreover, I don't have experience with frameworks. Moldy only supports two body Van Der Waal potentials of several kinds, in addition to the two body electrostatic terms associated with partial charges in the molecules/atoms. The dynamics are done using quaternions, rather than Euler angles, and the coordinates of the center of mass. It supports the Gaussian Thermostat and the Noose-Hover thermostat, as well as variations of the latter for constant pressure computations. It has a few utilities that are useful in fetching out and visualizing the results. The integration scheme is a Beeman algorithm that is of higher order than the Verlet scheme commonly used in Charmm/X-Plor/AMBER. It uses E-wald summation together with a linked-cell method to compute the intermolecular interactions in a periodic system. ... -------------------------------------------------------- From: Lutz.Ehrlich@EMBL-Heidelberg.de I inspected Moldy's capabilities for rigid body protein docking (which I right now do with XPLOR); it also implements the Newton-Euler equations of motion which are used to propagate collections of rigid bodies. Moldy can be downloaded from ftp.earth.ox.ac.uk using anon ftp; relevant files are in the /pub directory (information taken from Moldy 2.12). The author is Keith Refson (keith@earth.ox.ac.uk). > 1. Does anyone know of recent studies using this technique, > particularly applied to the protein-protein docking? The recent protein-protein docking studies that I know of seem to choose different ways of simplifying the proteins involved: Either a simplified energy function (which takes into account only that part of the interaction deemed relevant by the authors) is minimized using a MonteCarlo-Simulated Annealing-like protocol, or the minimum is found by complete enumeration (as in the case of the Fourier-convolution based approaches). All these approaches trade a detailed description of the dynamics for docking speed. On the other side, some MD studies of docking have been performed (cf. Di Nola et al., 1994). I haven't found any studies on protein-protein docking using Head-Gordon's VRBD. As part of my current project, I'm studied protein-protein docking using VRBD within XPLOR. Let me know if you're interested in my experiences. > 2. How does it compare with SHAKE type constraint methods: > faster/slower or just different? Just different would be a better description, as you constrain the bond angles and dihedrals as well (dependent of course, on the way you choose your rigid bodies). Second, in the SHAKE approach, each atom still experiences the force due to it's interaction with the rest of the system. In VRDB (as proposed by Head-Gordon) , a force/torque acting on a rigid body is derived from the forces acting on it's constitutive atoms. Third, Head-Gordon in her work enforces planarity of the peptide bond by turning it into a rigid body. Thus, your dynamics will differ >from a plain bond-length SHAKE simulation. ----------------------------------------------------------------------------- From: Konrad Hinsen I have no practical experience with VRBD, but with other treatments of rigid-body and constrained systems, which may be of interest. First of all it is necessary to distinguish between the consequence of applying certain constraints and the limitations/efficiency of particular methods to do so. Obviously keeping a peptide group rigid is a stronger constraint than fixing just the bond length, but it also comes with a higher reduction of fast degrees of freedom, which is after all the reason for applying mechanical constraints. For a discussion and an explicit comparison for helix dynamics, see K. Hinsen, G.R. Kneller Influence of constraints on the dynamics of polypeptide chains Phys. Rev. E 52, 6868 (1995) Once you have chosen the constraints you want to apply (that's a physical choice), you must then pick a method to implement them (that's an algorithmic choice). In this choice you must consider the limitations and efficiency criteria for each method. For SHAKE these are - limitation: no rigid groups of more than three atoms (planar) or four atoms (non-planar) (at least for the commonly implemented original version of SHAKE, a generalization has been published, but is not in general use) - efficiency problem: strongly interdependent constraint cause a long iteration For VRBD: - limitation: no constraints between rigid units - efficiency consideration: most efficient for large rigid units Note that many other techniques for implementing constraints have been proposed and/or tested. In principle, the results for a given set of constraints should be identical irrespective of the method used, but I am not aware of any practical comparison. ---------------------------------------------------------------------- From: hon@lisa.moldyn.com (Hon M Chun) There is a new modeling approach called MBO(N)D (for Multibody Order (n) Dynamics) that has more capabilities than the original Head-Gordon and Brooks work. It was developed at Moldyn as a result of application of flexible spacecraft simulation techniques to molecular dynamics modeling. The main difference is that, in addition to providing rigid body dynamics capabilities, MBO(N)D allows the bodies to be flexible, using body-based mode vectors to describe the allowed deformational motions. Another difference is that bond length constraints can be optionally applied across bonds connecting adjacent bodies. Bodies can be defined to be as small as rings and peptide planes, or as large as whole residues, secondary structures, or domains. Individual atoms can be modeled as particles. Of course, the user needs to decide what level of grouping is appropriate for the problem at hand. A multiple time scale integrator allows the efficient simulation of models where there co-exist bodies of various sizes, and even individual atoms. The code is interfaced to CHARMM, and uses a slightly extended set of keywords for the input script file. The combined code is also interfaced with INSIGHT 97.0, and is currently available from Molecular Simulations, Inc, which distributes CHARMM and MBO(N)D to commercial, academic, and government organizations. (The contact person at MSI is Mary Donlan; her e-mail address is mdonlan@msi.com.) A paper on MBO(N)D is being prepared for journal submission. I will be happy to send you a pre-print when it is ready. ----------------------------------------------------- From gbr@NPD.UFPE.BR Tue Nov 25 12:37:30 1997 Received: from NPD1.NPD.UFPE.BR for gbr@NPD.UFPE.BR by www.ccl.net (8.8.3/950822.1) id MAA12896; Tue, 25 Nov 1997 12:03:11 -0500 (EST) Received: from hartree.dqf.ufpe.br by NPD.UFPE.BR (PMDF V5.1-4 #20469) with ESMTP id <01IQFGS2MZRK002MS8@NPD.UFPE.BR> for CHEMISTRY@www.ccl.net; Tue, 25 Nov 1997 14:02:00 GMT-3 Date: Tue, 25 Nov 1997 14:00:38 -0200 From: Gerd Bruno da Rocha Subject: cyclodextrin inclusion complexes To: CHEMISTRY@www.ccl.net Message-id: <347AF626.6B365F32@npd.ufpe.br> Organization: Depto. de Quimica Fundamental - UFPE MIME-version: 1.0 X-Mailer: Mozilla 4.0 [en] (Win95; I) Content-type: text/plain; charset=x-UNICODE-2-0-UTF-7 Content-transfer-encoding: 7bit X-Priority: 3 (Normal) Dear All, Could someone give me any references or suggestions of the most appropriate force field for calculations of cyclodextrin inclusion complexes ? Thank you very much in advance! -- *********************************************************************** Universidade Federal de Pernambuco Centro de Ciencias Exatas e da Natureza Departamento de Quimica Fundamental http://www.dqf.ufpe.br Recife - PE - Brazil Gerd Bruno da Rocha E-Mail : gbr@npd.ufpe.br : gerd@dqf.ufpe.br MSc Student in Computational Chemistry *********************************************************************** From noda@zsu.zaporizhzhe.ua Tue Nov 25 12:41:19 1997 Received: from frya.zgik.zaporizhzhe.ua for noda@zsu.zaporizhzhe.ua by www.ccl.net (8.8.3/950822.1) id LAA12867; Tue, 25 Nov 1997 11:53:02 -0500 (EST) Received: from k1.zsu.zaporizhzhe.ua by frya.zgik.zaporizhzhe.ua with SMTP id RAA25181; (8.8.8/vak+evp/1.8e) Tue, 25 Nov 1997 17:50:31 +0200 (EET) Received: from pp200-nt-01 by k1.zsu.zaporizhzhe.ua with SMTP id RAA19393; (8.6.11/vak/1.8e) Tue, 25 Nov 1997 17:50:09 +0200 Date: Tue, 25 Nov 1997 17:50:09 +0200 Message-Id: <199711251550.RAA19393@k1.zsu.zaporizhzhe.ua> X-Sender: noda@win-serv.k1.zsu.zaporizhzhe.ua X-Mailer: Windows Eudora Light Version 1.5.2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: chemistry@www.ccl.net From: "Eugene G. Noda" Subject: HF and EC bond lengths At 04:48 PM 11/24/97 -0500, Jason Lye wrote: > > Please correct me if I am wrong; do you mean to say that metal-ligand > bonding in first row transition metals is predominantly due to 3s and 3p > orbitals, and that d-orbitals play only a minor role? > Dear Jason No, I can only insist on that 4s and 4p orbitals are redundant elements in theoretical treatment. As to the role of 3l electrons in the bonding, it should be proved by direct experiment which is not trivial to my knowledge. Fig.5. Besides d-orbitals, I would like to call your attention to overlapped ligand orbitals. They are calculated using ordinary Slater basis functions from the literature. You can see that amplitude of overlapping functions is not much different. Therefore, to neglect 3sp overlap integrals we can only in the case when 3sp orbitals of d-ion are extremely compact (or frosen to free atomic value). However, calcs show that situation is different. The average radius of 3s-3p orbs in matter is near the same as 3d one and these are stable numerical HF orbitals. As to 4s contribution, I am trying to catch it many years and, alas, unsuccessfully. This shell is so high in energy that even ions in matter have lost it :-) Or, seriously, it seems that 4s task is not simplier than negative ion problem in atomic calculations (at least in our approach). Bonding. Our modeling shows that at the _typical_ bonding distances the reordering of crystal subshells (splitted d orbital) leads to their crossing and thus to formation of new subshells with undetermined symmetry properties. To restore symmetry, the ligand orbitals should be involved and the manner of their combination determines the final coordination of the complex (naturaly, neglecting ligand-ligand interaction). So I believe that true theory of d-compounds should be a symmetry theory. However, contrary to usual point group analysis, this theory must be based on the fact that electronic symmetry of d-ion in matter is dynamical property and should be determined in self-consistent calculation. Hope these speculations will be stimulating, Eugene From billings@helix.nih.gov Tue Nov 25 12:42:58 1997 Received: from helix.nih.gov for billings@helix.nih.gov by www.ccl.net (8.8.3/950822.1) id MAA12964; Tue, 25 Nov 1997 12:33:01 -0500 (EST) Received: (from billings@localhost) by helix.nih.gov (8.8.5/8.8.5) id MAA01766; Tue, 25 Nov 1997 12:32:53 -0500 (EST) Date: Tue, 25 Nov 1997 12:32:53 -0500 (EST) From: Eric Billings Reply-To: Eric Billings To: Chris Harwell cc: ccl Subject: LoBoS, a Beowulf cluster at NIH In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On Mon, 24 Nov 1997, Chris Harwell wrote: > Are there any chemists out there who are using a beowulf cluster > of pcs type of system for computing? We have constructed an 80 processor Beowulf-class supercomputer called LoBoS (Lots of Boxes on Shelves). LoBoS will be scaled up to 136 processors next week. LoBoS' home page is: http://www.lobos.nih.gov/ and contains a more complete description of the equipment. We are in the process of adding benchmarks for different parallel topologies. BTW, the hardware pulse and queue management code seen on the web page are freely available. As a government lab we can't put much vendor-specific info on the web site (but we can talk about it). It is safe to say that, for our applications (QM and classical simulation) we are getting 10 times the price performance of the local supercomputer ! > Has anyone used gamess, gaussian, or Wien FLAPW on such a system? The test and current (Phase 2) systems have been in production use since April performing CHARMM, GAMESS and AMBER runs. >What hardware choices did you make (processor. motherboard etc.) ? We tested the DEC Alpha, PowerPC, Pentium Pro and PentiumII processors. The best price/performance was clearly (by a factor of 2-10) the Pentium Pro. The DEC Alpha remains an excellent processor so we are continuing to develop a 5 processor DEC Alpha cluster. Once we settled on the PentiumPro we tested 5 motherboards and found a small variance (+/- 5%) for well designed motherboards. Cheap motherboards which were/are available use the older Intel chipset and are 20-30% slower. > What compiler did you use (gcc,f2c, g77)? We tested 3 commercial compilers as well as f2c/gcc and g77. We are using the compiler from AbSoft. > What OS/Distribution did you use (linux/ redhat or debian?) All nodes use Linux. I think its safe to say that the Beowulf community is moving towards the use of RPM (RedHat) for installs and ensemble management software distribution, but we have been using Debian with good success. > Any suggestions to someone considering building such a system? 1. At a recent conference there was a general consensus that the hardware for up to 16 nodes was a "solved problem" when optimizing price performance. The equipment includes PentiumPros, a single Fast Ethernet NIC and a 16-port Fast Ethernet Switch (not hub). Add disk to suit your taste. 2. Since it sounds like you're doing QM work, you might consider the Dual PentiumPro configurations. You can amortize the cost of supporting hardware (disk, power supply, case etc) over two processors and improve your overall price/performance. 3. Don't be taken in by the Intel Bunny people and automatically buy a PentiumII. The PentiumII is a PentiumPro with MMX (useless for floating point ops) and faster memory access. Unfortunately, the advantages are offset by an off-chip cache that runs at half the speed of the PentiumPro. Our benchmarks showed that the PentiumII performance improvements barely scale up with the faster clock rates. - Eric Eric Billings Molecular Graphics and Simulation Laboratory National Institutes of Health From youkha@inpharmatics.com Tue Nov 25 13:44:23 1997 Received: from inet1.inetworld.net for youkha@inpharmatics.com by www.ccl.net (8.8.3/950822.1) id NAA13284; Tue, 25 Nov 1997 13:09:11 -0500 (EST) Received: from sydney.inpharma.com (n207211099143.inetworld.net [207.211.99.143] (may be forged)) by inet1.inetworld.net (8.8.7/8.8.7) with SMTP id KAA07527; Tue, 25 Nov 1997 10:08:49 -0800 (PST) Message-ID: <347A9046.7D74@inpharmatics.com> Date: Tue, 25 Nov 1997 00:45:58 -0800 From: Philippe Youkharibache Reply-To: youkha@inpharmatics.com Organization: InPharmatics Corp. X-Mailer: Mozilla 3.01Gold (Win95; I) MIME-Version: 1.0 To: Troy Wymore CC: chemistry@www.ccl.net Subject: Re: CCL:N-H Correlation function in CHARMM References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Troy Wymore wrote: > > I would like to construct a correlation function for N-H vectors in my > peptide from trajectory files using CHARMM version 24b2. People have done > this to make the connection between the motion of the peptide and NMR > relaxation. I know CHARMM has the NMR analysis facility which I have used > successfully; my problem is using CORREL to construct the correlation > function. > Troy, You could use the autocorrelation function (spectra) in DeCipher ... with the new version 970 that can read CHARMM trajectories. This is probably the best news from the Biosym-MSI merger. I have been told that there is documentation on www.msi.com/doc, but I have not checked, and that the software has been shipped to all licensed msi customers. Philippe. ======================================================================= Philippe Youkharibache, Ph.D. InPharmatics Corp. e-mail: youkha@inpharmatics.com 11440 W. Bernardo Court, Suite 300 tel: (619) 530 8550 San Diego, CA 92127 fax: (619) 530 8555 ======================================================================= From hou@Agouron.COM Tue Nov 25 15:37:18 1997 Received: from tbone.agouron.com for hou@Agouron.COM by www.ccl.net (8.8.3/950822.1) id PAA14031; Tue, 25 Nov 1997 15:21:35 -0500 (EST) Received: (from gproxy@localhost) by tbone.agouron.com (8.8.7/8.8.7) id MAA08269 for ; Tue, 25 Nov 1997 12:21:36 -0800 (PST) Received: from dns3.agouron.com(198.182.180.53) by tbone.agouron.com via smap (3.2) id xma008200; Tue, 25 Nov 97 12:21:14 -0800 Received: from horton.agouron.com (horton.agouron.com [198.49.92.17]) by dns3.agouron.com (8.8.8/8.8.8) with SMTP id MAA13502 for <@dns3.agouron.com:CHEMISTRY@www.ccl.net>; Tue, 25 Nov 1997 12:21:13 -0800 (PST) (envelope-from hou@horton.agouron.com) Received: (from hou@localhost) by horton.agouron.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id MAA11911 for CHEMISTRY@www.ccl.net; Tue, 25 Nov 1997 12:18:51 -0800 Date: Tue, 25 Nov 1997 12:18:51 -0800 From: hou@Agouron.COM (Xinjun Hou) Message-Id: <199711252018.MAA11911@horton.agouron.com> To: CHEMISTRY@www.ccl.net Subject: Re: CCL:G:LoBoS, a Beowulf cluster at NIH Reply-To: hou@Agouron.COM On Mon, 24 Nov 1997, Chris Harwell wrote: >... >3. Don't be taken in by the Intel Bunny people and automatically buy a >PentiumII. The PentiumII is a PentiumPro with MMX (useless for >floating point ops) and faster memory access. Unfortunately, the >advantages are offset by an off-chip cache that runs at half the speed >of the PentiumPro. Our benchmarks showed that the PentiumII performance >improvements barely scale up with the faster clock rates. Another issue is the cost of processor in the future: PPro's architecture (on-chip cache) makes it very costly in production. This is another reason Intel came up with PentiumII with off-chip cache. The off-chip cache structure is also cheaper to expand, both in terms of size and clock speed. So bottom line is that PPro with on-chip cache has a better architecture than Pentium II for computional extensive work, but Intel might not lower the price further or devepop it more (high clock speed or more cache). The cost/performance ration of two CPU would change in the future. Xinjun C Xinjun J. Hou (hou@agouron.com) Agouron Pharmaceuticals, Inc. C10110000110100101101110011010100111010101101110010010000110111101110101 From martin@saluki-mail.siu.edu Tue Nov 25 16:37:19 1997 Received: from saluki-mail.siu.edu for martin@saluki-mail.siu.edu by www.ccl.net (8.8.3/950822.1) id PAA14158; Tue, 25 Nov 1997 15:58:49 -0500 (EST) Received: from 131.230.89.21 (harpo.thchem.siu.edu [131.230.89.21]) by saluki-mail.siu.edu (AIX4.2/UCB 8.7/8.7) with SMTP id OAA178320 for ; Tue, 25 Nov 1997 14:58:14 -0600 (CST) Message-Id: <199711252058.OAA178320@saluki-mail.siu.edu> X-Mailer: Microsoft Internet Mail & News for Macintosh - 3.0c (405) Date: Tue, 25 Nov 1997 15:01:04 -0600 Subject: Benzene excited states again From: Martin Cuma To: chemistry@www.ccl.net Mime-version: 1.0 Content-type: text/plain; charset="us-ascii" Content-transfer-encoding: 7bit Hello, thanks all for the reply to my previous question. The latest reference is J. Chem. Phys. 106, 6051 (1997). However, it did not solve my quiestion, since this work and all the references thereof calculate the excited states in the ground state geometry (i.e. vertical excitation energies). What I need is the optimized geometries of the first pi excited states (i.e. from there the adiabatic exc. energies). Is there any reference for that? Thanks in advance, MC Martin Cuma Dept. of Chemistry Southern Illinois University at Carbondale From berriz@chasma.harvard.edu Tue Nov 25 18:37:20 1997 Received: from chasma.harvard.edu for berriz@chasma.harvard.edu by www.ccl.net (8.8.3/950822.1) id RAA14425; Tue, 25 Nov 1997 17:41:19 -0500 (EST) Received: by chasma.harvard.edu (AIX 3.2/UCB 5.64/4.03) id AA21994; Tue, 25 Nov 1997 17:38:26 -0500 Date: Tue, 25 Nov 1997 17:38:26 -0500 From: berriz@chasma.harvard.edu (Gabriel Berriz) Message-Id: <9711252238.AA21994@chasma.harvard.edu> To: chemistry@www.ccl.net Subject: ISO coordinates for canonical helix Dear netters: Where from in the Internet can download the coordinates for a canonical alpha-helix (say, poly-ala)? (If there's a choice, PDB format would be slightly preferable.) Thanks for your help, Gabriel Berriz berriz@chasma.harvard.edu From info@microsimulations.com Tue Nov 25 22:37:28 1997 Received: from maple.nis.net for info@microsimulations.com by www.ccl.net (8.8.3/950822.1) id WAA15111; Tue, 25 Nov 1997 22:29:14 -0500 (EST) Received: from merlin.nis.net by maple.nis.net with smtp (Smail3.2.0.95 #47) id m0xaYkz-000SozC; Tue, 25 Nov 1997 23:07:21 -0500 (EST) Received: from PACKARDBELL1 [207.99.101.85] by merlin.nis.net with smtp (Exim 1.62 #1) id 0xaYgy-0000Kj-00; Tue, 25 Nov 1997 23:03:13 -0500 Message-ID: <347BBD9B.6131@microsimulations.com> Date: Tue, 25 Nov 1997 22:11:39 -0800 From: Willie Cui Reply-To: info@microsimulations.com Organization: microsimulations X-Mailer: Mozilla 3.0 (Win95; I; 16bit) MIME-Version: 1.0 To: Andy Holder CC: chemistry@www.ccl.net Subject: Re: CCL:M:MJS Dewar died October 10 References: <9711201406.AA44424@134.193.11.2> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Andy, I share you respect to Prof. Dewar and feel the lost of him passing away. On your message, you raised a very import point that his contribution to chemistry worthy of the Nobel prize. I am wondering if computational chemistry is overlooked by the Nobel prize committee. Best! Andy Holder wrote: > > --part_B099F4740018449300000003 > Content-Type: Text/Plain; charset=US-ASCII > Content-Disposition: Inline > > Dear Netters and Friends, > > On October 10 of this year MJS Dewar died of congestive heart failure in a > hospital in Gainesville, FL. After speaking with his son Steuart, it seems > that his death was not difficult. > > Since most (if not all) of us use his methods and ideas in our work, I > thought it beneficial to take a moment and review just a few of Michael > Dewar's contributions to computational chemistry. There is a short (and > woefully incomplete) list below. Even more than his direct contributions to > the literature (over 600 papers), Professor Dewar sparked people to think > differently and in unconventional ways. As many of us know, he was involved > in many hot controversies in the literature, all of which had the result of > advancing science. Michael Dewar would not let you just plod along in the > same old rut. His sparkling prose and inimitable speaking style challenged > audiences and stimulated discussions for decades. I think it fitting that > Michael died in early October, when the Nobel Prize recipients are announced. > His contributions are certainly worthy of the prize, and he was a perrenial > candidate for it. > > 1. Perturbation molecular orbital (PMO) theory > 2. MINDO3 semiempirical method > 3. MNDO semiempirical method > 4. AM1 semiempirical method > 5. SAM1 semiempirical method > 6. AMPAC and MOPAC programs > 7. Elucidation of basic reaction mechanisms throughout organic chemistry > through the application of theoretical methods > 8. Basic contributions in the theory and implementation of semiempirical > methodology > > A little recent history is probably in order. Professor Dewar left the > University of Texas about 1990 and accepted a research professorship at the > University of Florida. He worked there on development of the SAM1 method > until 1992, when he retired. Michael completely left chemistry when he > retired, and enjoyed reading (virtually anything), music, old movies, and > Chinese cooking to mention just a few of his many interests. Mary Williamson > Dewar, his and wife and best friend of over 50 years, passed away in 1994. > There was no memorial service for Michael, and his ashes will be spread with > those of Mary in the Lake District of northern England, where they honeymooned > many years ago. He is survived by his two sons C.E. Steuart Dewar of > Morganton, GA and Robert B.K. Dewar of New York City. > > He challenged those of us who had the great fortune to work with him in ways > we never imagined. He was a loyal and caring mentor and was always gracious > and kind, even in those painful moments when we insisted on learning the "hard > way". > > Not just those of us who knew him, but SCIENCE will miss him. A great man has > left us. > > Regards, Andy Holder > > PS. There are preliminary efforts underway to develop an endowment in > Professor Dewar's name at the University of Texas at Austin. If you > are interested in being added to the mailing list for this, please let > me know. > > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > > UUUU UUU MMM MMKK KKKK CCCC | ANDREW J. HOLDER > UU U MM MMK K CC CC | Assoc. Prof. of Comp./Org. Chemistry > UU U MMM M MK KK CCC | Dept. of Chemistry > UU U M MM MK KK CC CC | University of Missouri-Kansas City > UUUUU MMM M MMKK KK CCCC | Kansas City, MO 64110 > KK | aholder@cctr.umkc.edu > K | (816) 235-2293, FAX (816) 235-5502 > > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > > > --part_B099F4740018449300000003-- > > -------This is added Automatically by the Software-------- > -- Original Sender Envelope Address: aholder@cctr.umkc.edu > -- Original Sender From: Address: aholder@cctr.umkc.edu > CHEMISTRY@www.ccl.net: Everybody | CHEMISTRY-REQUEST@www.ccl.net: Coordinator > MAILSERV@www.ccl.net: HELP CHEMISTRY or HELP SEARCH | Gopher: www.ccl.net 73 > Anon. ftp: www.ccl.net | CHEMISTRY-SEARCH@www.ccl.net -- archive search > Web: http://www.ccl.net/chemistry.html -- Willie Cui voice: 201-512-0486 MicroSimulations fax: 201-512-0489 478 Green Mountain Road email: info@microsimulations.com Mahwah, NJ 07430 URL: http://www.microsimulations.com