From owner-chemistry@ccl.net Mon Jun 2 03:26:01 2008 From: "VITORGE Pierre 094605 Pierre.VITORGE.:.cea.fr" To: CCL Subject: CCL: IC50 deltaG Message-Id: <-37068-080602032452-22781-my73Je21t6eP4e5xM0fHQA%server.ccl.net> X-Original-From: "VITORGE Pierre 094605" Content-class: urn:content-classes:message Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="iso-8859-1" Date: Mon, 2 Jun 2008 09:24:37 +0200 MIME-Version: 1.0 Sent to CCL by: "VITORGE Pierre 094605" [Pierre.VITORGE[-]cea.fr] DG =3D -RTlnK at constant T and P is correct for any reaction, where K = is the corresponding equilibrium constant, usually Q, the concentration = and partial pressure quotient, including all the soluble and gaseous = species in the equilibrium, true chemical species -not just = stoichemetric formula-. DG refers to the standard state, which is = defined by the concentration units in K. Pierre Vitorge=20 http://www.vitorge.name=20 -----Message d'origine----- De=A0: owner-chemistry#%#ccl.net [mailto:owner-chemistry#%#ccl.net]=20 Envoy=E9=A0: dimanche 1 juin 2008 23:38 =C0=A0: VITORGE Pierre Objet=A0: CCL: IC50 deltaG Sent to CCL by: "R D" [rafi4dd+/-gmail.com] Hello, I need a clarification on fundamentals. I want to compare experimental = IC50 values to calculated binding free energy values.=20 Can I assume IC50 is approx =3D Kd and use DeltaG =3D RTlnIC50. for example if I want to convert an IC50 value of 1microM it will be = equivalent to delta G of -8.2Kcal/mol. Am I right? Are we using DeltaG =3D RTlnIC50 (instead of -RTlnIC50) because we are = focused on dissociation of receptor-ligand complex rather than = association of receptor ligand complex.=20 I apologise if the question is silly, but any reply will be greatly = helpful. Thank you. -=3D This is automatically added to each message by the mailing script = =3D-http://www.ccl.net/cgi-bin/ccl/send_ccl_messageSubscribe/Unsubscribe:=20Job: http://www.ccl.net/jobs=20Search Messages: http://www.ccl.net/htdig (login: ccl, Password: = search)http://www.ccl.net/spammers.txt From owner-chemistry@ccl.net Mon Jun 2 06:04:01 2008 From: "Pablo Echenique echenique.p:gmail.com" To: CCL Subject: CCL: computer for high-level QC calculations Message-Id: <-37069-080602055520-4932-pKOSwMJQ8w/jzq/xZm2oZA#%#server.ccl.net> X-Original-From: "Pablo Echenique" Content-Type: multipart/alternative; boundary="----=_Part_8133_6497630.1212397093558" Date: Mon, 2 Jun 2008 10:58:13 +0200 MIME-Version: 1.0 Sent to CCL by: "Pablo Echenique" [echenique.p**gmail.com] ------=_Part_8133_6497630.1212397093558 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Dear friends and colleagues, at my laboratory, we are about to buy a computer for performing high-level QC calculations (e.g., coupled cluster) in biological molecules (e.g., peptides). I write to kindly ask you for advice regarding the "best" (in whatever sense you consider appropriate) choice. One option would be a SGI Altix 4700 with 64 dual-core Itanium processors and 512GB of shared memory. The other option would be a number of smaller nodes, say 16 cores each, say 64GB of shared memory each. Gigabit or Infiniband connected. Some questions/remarks to orient the answers: 1. Are there QC problems that actually require 512GB of shared memory? Don'= t we face a CPU time bottleneck before that much memory is needed? 2. The second option would certainly present a lower cost per core. 3. The first option is said to be more easy to code parallel applications, so I suppose that more QC codes can be efficiently used on the Altix than o= n a multi-node architecture. 4. What specific machine would be the realization of the second option? Thank you in advance, --=20 Pablo Echenique Instituto de Biocomputaci=F3n y F=EDsica de los Sistemas Complejos (BIFI) Departamento de F=EDsica Te=F3rica Universidad de Zaragoza Pedro Cerbuna 12, 50009 Zaragoza Spain Tel.: +34 976761260 Fax: +34 976761264 echenique.p^-^gmail.com http://www.pabloechenique.com ------=_Part_8133_6497630.1212397093558 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Content-Disposition: inline Dear friends and colleagues,

at my laboratory, we are about to buy a= computer for performing high-level QC calculations (e.g., coupled cluster)= in biological molecules (e.g., peptides). I write to kindly ask you for ad= vice regarding the "best" (in whatever sense you consider appropr= iate) choice.

One option would be a SGI Altix 4700 with 64 dual-core Itanium processo= rs and 512GB of shared memory.

The other option would be a number of= smaller nodes, say 16 cores each, say 64GB of shared memory each. Gigabit = or Infiniband connected.

Some questions/remarks to orient the answers:

1. Are there QC pr= oblems that actually require 512GB of shared memory? Don't we face a CP= U time bottleneck before that much memory is needed?
2. The second optio= n would certainly present a lower cost per core.
3. The first option is said to be more easy to code parallel applications, = so I suppose that more QC codes can be efficiently used on the Altix than o= n a multi-node architecture.
4. What specific machine would be the reali= zation of the second option?

Thank you in advance,


--
Pablo Echenique
Instituto de Biocomputaci=F3n y
F=EDsica de los Sistemas Complejos= (BIFI)

Departamento de F=EDsica Te=F3rica
Universidad de Zaragoz= a
Pedro Cerbuna 12, 50009 Zaragoza
Spain

Tel.: +34 976761260
Fax: +34 976761264

echenique.p^-^gmail.com
http://www.pabloechenique.com ------=_Part_8133_6497630.1212397093558-- From owner-chemistry@ccl.net Mon Jun 2 08:00:00 2008 From: "Karin Wichmann wichmann===cosmologic.de" To: CCL Subject: CCL: COSMOtherm / COSMO-RS Training Message-Id: <-37070-080602071646-7934-VQwyohoX4zQ0rJHxivgiLQ^^server.ccl.net> X-Original-From: Karin Wichmann Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-15; format=flowed Date: Mon, 02 Jun 2008 12:18:21 +0200 MIME-Version: 1.0 Sent to CCL by: Karin Wichmann [wichmann_-_cosmologic.de] COSMOlogic has a small number of seats available for a free COSMOtherm training on June 23. The training is a one-day course covering COSMO-RS theory for fluid thermodynamics and its applications in the areas of chemical engineering and life science. The course will take place at our location in Leverkusen (Germany). For further information please contact us at cosmotherm (AT) cosmologic.de -- ------------------------ Dr. Karin Wichmann COSMOlogic GmbH & Co. KG Burscheider Str. 515 D-51381 Leverkusen Germany Phone +49-2171-73168-3 Fax +49-2171-73168-9 ------------------------ From owner-chemistry@ccl.net Mon Jun 2 09:05:01 2008 From: "Armin M. Sobhani armin]![iums.ac.ir" To: CCL Subject: CCL: IC50 deltaG Message-Id: <-37071-080602085724-10573-NMxiVRLpZY/HlHUeri0fdw!A!server.ccl.net> X-Original-From: "Armin M. Sobhani" Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="iso-8859-1"; format=flowed Date: Mon, 02 Jun 2008 16:34:35 +0330 Mime-Version: 1.0 Sent to CCL by: "Armin M. Sobhani" [armin]^[iums.ac.ir] According to the following equations: [LR] =3D [L][R]total/Ki+[L] =3D [L][R]max/Ki+[L] Delta/Delta max =3D [LR]/[LR]max =3D [L]/(Ki+[L]) In half maximum inhibition we can write: Ki =3D [L] =3D IC50 This is for inhibitors (or antagonists). For=20 agonists, you may equally replace Ki with Kd in the above equations and= write: Kd =3D [L] =3D EC50 Because of this, in pharmacology we use IC50 to=20 estimate Ki (EC50 to estimate Kd). But=20 calorimetric methods (e.g. isothermal titration=20 calorimetry) are the most exact methods to=20 determine Ki. In worst cases (e.g. when drug=20 couldn't reach the site of action) IC50 and Ki=20 (EC50 and Kd) values maybe be completely different. Armin Madadkar Sobhani p-----__---_________----____----------------------------------------q | / \ / _ _ \ / ____\ Armin M. Sobhani, Pharm.D, Ph.D | | / \\ \\ \\ \/\ \___/_ Assistant Prof., Bioinformatics Dept| | / /\ \\ \\ \\ \ \_____ \ IBB, University of Tehran | | ( ___ \\ \\ \\ \/____/\ \ mailto:armin-$-ibb.ut.ac.ir | |( \__\_/\__\\__\\__\\__\/\_____/ Tel: +98 (21) 6696 9261 | | \/__/ \/__//__//__//__/\/____/ P.O.Box: 13145-1365, Tehran, IRAN | |~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| | Once you understand the universe at the atomic level everything | | else is easy. -Richard P. Feynman | b-------------------------------------------------------------------d At 01:07 =DE.=D9 2008/06/02, you wrote: >Sent to CCL by: "R D" [rafi4dd+/-gmail.com] >Hello, > >I need a clarification on fundamentals. I want=20 >to compare experimental IC50 values to calculated binding free energy= values. > >Can I assume IC50 is approx =3D Kd and use DeltaG =3D RTlnIC50. > >for example if I want to convert an IC50 value=20 >of 1microM it will be equivalent to delta G of -8.2Kcal/mol. Am I right? > >Are we using DeltaG =3D RTlnIC50 (instead of=20 >-RTlnIC50) because we are focused on=20 >dissociation of receptor-ligand complex rather=20 >than association of receptor ligand complex. > >I apologise if the question is silly, but any reply will be greatly= helpful. > >Thank you. From owner-chemistry@ccl.net Mon Jun 2 10:26:01 2008 From: "Armin Madadkar Sobhani armin-#-ibb.ut.ac.ir" To: CCL Subject: CCL: IC50 deltaG Message-Id: <-37072-080602074519-10974-u9tGJiZjnb+6cnPNmzyCHA^server.ccl.net> X-Original-From: "Armin Madadkar Sobhani" Date: Mon, 2 Jun 2008 07:45:15 -0400 Sent to CCL by: "Armin Madadkar Sobhani" [armin * ibb.ut.ac.ir] According to the following equations: [LR] = [L][R]total/Ki+[L] = [L][R]max/Ki+[L] Delta/Delta max = [LR]/[LR]max = [L]/(Ki+[L]) In half maximum inhibition we can write: Ki = [L] = IC50 Because of this, in pharmacology we use IC50 to estimate Ki. But calorimetric methods (e.g. isothermal titration calorimetry) are the most exact methods to determine Ki. In worst cases (e.g. when drug couldn't reach the site of action) IC50 and Ki values may be completely different. Armin Madadkar Sobhani p-----__---_________----____----------------------------------------q | / \ / _ _ \ / ____\ Armin M. Sobhani, Pharm.D, Ph.D | | / \\ \\ \\ \/\ \___/_ Assistant Prof., Bioinformatics Dept| | / /\ \\ \\ \\ \ \_____ \ IBB, University of Tehran | | ( ___ \\ \\ \\ \/____/\ \ mailto:armin||ibb.ut.ac.ir | |( \__\_/\__\\__\\__\\__\/\_____/ Tel: +98 (21) 6696 9261 | | \/__/ \/__//__//__//__/\/____/ P.O.Box: 13145-1365, Tehran, IRAN | |~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| | Once you understand the universe at the atomic level everything | | else is easy. -Richard P. Feynman | b-------------------------------------------------------------------d From owner-chemistry@ccl.net Mon Jun 2 11:01:01 2008 From: "Yann Tambouret yannpaul###chem.bu.edu" To: CCL Subject: CCL:G: ab initio calc. and metal complex Message-Id: <-37073-080602102406-14554-9O/yc7DhgOtwZ71vwyNiNg_-_server.ccl.net> X-Original-From: Yann Tambouret Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 02 Jun 2008 09:52:23 -0400 MIME-Version: 1.0 Sent to CCL by: Yann Tambouret [yannpaul * chem.bu.edu] Dear CCL users, I've been asked by a colleague to help with a proposal due in a week. He's proposing to study a metal-complex-catalized reaction involving small organic molecules. He has a suspected mechanism and how the metal complex is involved, but has asked me to do some ab initio calculations to show the complex's effect on the organic compound's electronic structure. I could use some help finding good examples of these types of calculations and any suggestions on how to judiciously cut corners. For example, is there any reasonable approximation that can be made for the structure of less-relevant ligands? What general advice can you offer in the use of effective core potentials? Thanks in advance for your replies. I have access to Gaussian 03 and NWChem. Sincerely, Yann Tambouret Boston University yannpaul(-at-)chem.bu.edu From owner-chemistry@ccl.net Mon Jun 2 12:44:01 2008 From: "Elaine Meng meng^cgl.ucsf.edu" To: CCL Subject: CCL: IC50 deltaG Message-Id: <-37074-080602121631-13061-G8zJQe1fF5fM8it41CrSKg..server.ccl.net> X-Original-From: "Elaine Meng" Date: Mon, 2 Jun 2008 12:16:27 -0400 Sent to CCL by: "Elaine Meng" [meng+*+cgl.ucsf.edu] Hello, The Cheng-Prusoff equation is used to calculate K(i) from IC50, when IC50 is from a competition assay. It factors in the concentration and affinity of the competing ligand. See for example: http://www.unmc.edu/Pharmacology/receptortutorial/competition/analysis_sample4.htm http://www.tocris.com/PharmGlossUS.php Elaine ----- Elaine C. Meng, Ph.D. meng%x%cgl.ucsf.edu UCSF Computer Graphics Lab and Babbitt Lab Department of Pharmaceutical Chemistry University of California, San Francisco http://www.cgl.ucsf.edu/home/meng/index.html From owner-chemistry@ccl.net Mon Jun 2 13:19:01 2008 From: "Raj S r.subramanian|,|ipc.uni-stuttgart.de" To: CCL Subject: CCL:G: problems in Message-Id: <-37075-080602125500-21751-D03+LAjvPQaYJCXCeiGvGA#server.ccl.net> X-Original-From: "Raj S" Date: Mon, 2 Jun 2008 12:54:56 -0400 Sent to CCL by: "Raj S" [r.subramanian^ipc.uni-stuttgart.de] Dear ccl users, I am using Gaussian03 package for optimization and freq. calculations. After opt. and freq. calculations i notice in the .log file the total spin is marked ******. For e.g., look below (A) (A) (A) (A) of initial guess=******* Requested convergence on RMS density matrix=1.00D-08 within1000 cycles. Requested convergence on MAX density matrix=1.00D-06. Requested convergence on energy=1.00D-06. No special actions if energy rises. Restarting incremental Fock formation. SCF Done: E(UB+HF-LYP) = -2934.91465147 A.U. after 20 cycles Convg = 0.9816D-08 -V/T = 2.0810 S**2 = ******* Annihilation of the first spin contaminant: S**2 before annihilation 110.0343, after 110.0001 1 Symmetry operations used in ECPInt. I dont know what is it *******?Is this some error... many thanks for your reply raja From owner-chemistry@ccl.net Mon Jun 2 13:54:00 2008 From: "Michael K. Gilson gilson(_)umbi.umd.edu" To: CCL Subject: CCL: IC50 deltaG Message-Id: <-37076-080602130551-27036-zd5Ei9k+Niz3MufhMLxLzg]![server.ccl.net> X-Original-From: "Michael K. Gilson" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 02 Jun 2008 13:05:31 -0400 MIME-Version: 1.0 Sent to CCL by: "Michael K. Gilson" [gilson,,umbi.umd.edu] No, you can't assume IC50 is approximately Kd and use DG = RT ln IC50. IC50 is the concentration of inhibitor that leads to half-maximal inhibition of the targeted enzyme or receptor. If it is a competitive inhibitor, then it is competing with substrate (in the case of an enzyme) or with a labelled ligand (in the case of a receptor). For a given value of Ki, the value of IC50 will still vary depending upon how tightly the substrate or labeled ligand binds the protein, and also upon its concentration. The higher the affinity of the substrate or labeled ligand and the higher its concentration, the more inhibitor will be needed to have an effect, and hence the higher IC50 will be -- even though Ki is unchanged. The relationship is given by the Cheng-Prusoff equation. See, e.g., http://www.ncgc.nih.gov/guidance/section11.html However, if two inhibitors are assayed for same substrate or labeled ligand and same concentration thereof, then Delta Delta G = -RT ln r, where r is the ratio of the IC50 values. (The analysis is different if the inhibitor is not competitive.) Regards, Mike R D rafi4dd:gmail.com wrote: > Sent to CCL by: "R D" [rafi4dd+/-gmail.com] > Hello, > > I need a clarification on fundamentals. I want to compare experimental IC50 values to calculated binding free energy values. > > Can I assume IC50 is approx = Kd and use DeltaG = RTlnIC50. > > for example if I want to convert an IC50 value of 1microM it will be equivalent to delta G of -8.2Kcal/mol. Am I right? > > Are we using DeltaG = RTlnIC50 (instead of -RTlnIC50) because we are focused on dissociation of receptor-ligand complex rather than association of receptor ligand complex. > > I apologise if the question is silly, but any reply will be greatly helpful. > > Thank you.> > > > -- Michael K. Gilson, M.D., Ph.D. CARB Fellow and Professor Center for Advanced Research in Biotechnology University of Maryland Biotechnology Institute 9600 Gudelsky Drive Rockville, MD 20850 Voice: 240-314-6217 Fax: 240-314-6255 gilsonumbi.umd.edu Lab Page: gilsonlab.umbi.umd.edu BindingDB: www.bindingdb.org From owner-chemistry@ccl.net Mon Jun 2 15:45:00 2008 From: "Nuno A. G. Bandeira nuno.bandeira~!~ist.utl.pt" To: CCL Subject: CCL: problems in Message-Id: <-37077-080602152007-8801-F5DxofajEjGzAi8JTnEpcw]^[server.ccl.net> X-Original-From: "Nuno A. G. Bandeira" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 02 Jun 2008 19:42:38 +0100 MIME-Version: 1.0 Sent to CCL by: "Nuno A. G. Bandeira" [nuno.bandeira+*+ist.utl.pt] Raj S r.subramanian|,|ipc.uni-stuttgart.de wrote: S**2 = ******* > Annihilation of the first spin contaminant: > S**2 before annihilation 110.0343, after 110.0001 There's something very wrong with your input. The average of S^2 can never be this big. Most likely the spin multiplicity card is read from one of the cartesian coordinates. -- Nuno A. G. Bandeira, AMRSC Graduate researcher and molecular sculptor Inorganic and Theoretical Chemistry Group, Faculty of Science University of Lisbon - C8 building, Campo Grande, 1749-016 Lisbon,Portugal http://cqb.fc.ul.pt/intheochem/nuno.html Doctoral student ]![ IST,Lisbon -- From owner-chemistry@ccl.net Mon Jun 2 16:31:00 2008 From: "Steven Gwaltney gwaltney-#-ccs.msstate.edu" To: CCL Subject: CCL: computer for high-level QC calculations Message-Id: <-37078-080602082509-26615-bCaPOzjGcSYG1m6pFbl7xg]![server.ccl.net> X-Original-From: Steven Gwaltney Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Mon, 02 Jun 2008 06:51:56 -0500 MIME-Version: 1.0 Sent to CCL by: Steven Gwaltney [gwaltney^_^ccs.msstate.edu] The first question you need to answer is what code you intend to use. Three I know of that can run massively parallel coupled-cluster calculations are NWChem, GAMESS, and ACES 3. I'm sure others exist, as well. Once you had chosen a code, then contact the authors. They should be able to give you information about what kinds of architectures will run their code most efficiently. In general, though, most of the massively parallel codes are designed to run on distributed memory computers. I can imagine some calculations that could efficiently use 512 GB of shared memory, but I can't think of any that would require it. Steve Pablo Echenique echenique.p:gmail.com wrote: > Dear friends and colleagues, > > at my laboratory, we are about to buy a computer for performing > high-level QC calculations (e.g., coupled cluster) in biological > molecules (e.g., peptides). I write to kindly ask you for advice > regarding the "best" (in whatever sense you consider appropriate) choice. > > One option would be a SGI Altix 4700 with 64 dual-core Itanium > processors and 512GB of shared memory. > > The other option would be a number of smaller nodes, say 16 cores each, > say 64GB of shared memory each. Gigabit or Infiniband connected. > > Some questions/remarks to orient the answers: > > 1. Are there QC problems that actually require 512GB of shared memory? > Don't we face a CPU time bottleneck before that much memory is needed? > 2. The second option would certainly present a lower cost per core. > 3. The first option is said to be more easy to code parallel > applications, so I suppose that more QC codes can be efficiently used on > the Altix than on a multi-node architecture. > 4. What specific machine would be the realization of the second option? > > Thank you in advance, > > > -- > Pablo Echenique > > Instituto de Biocomputación y > Física de los Sistemas Complejos (BIFI) > > Departamento de Física Teórica > Universidad de Zaragoza > Pedro Cerbuna 12, 50009 Zaragoza > Spain > > Tel.: +34 976761260 > Fax: +34 976761264 > > echenique.p .. gmail.com > http://www.pabloechenique.com -- Dr. Steven Gwaltney Phone: 662-325-7602 Assistant Professor Fax: 662-325-1618 Department of Chemistry, Mail: Box 9573 Center for Environmental Health Sciences, Mississippi State University and HPCC Center for Computational Sciences Mississippi State, MS 39762 From owner-chemistry@ccl.net Mon Jun 2 18:57:01 2008 From: "Zsolt Zsoldos zsolt,simbiosys.ca" To: CCL Subject: CCL: computer for high-level QC calculations Message-Id: <-37079-080602185218-15495-0TPh7aCun2NxYKmI5//6Yg^server.ccl.net> X-Original-From: "Zsolt Zsoldos" Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=ISO-8859-1 Date: Mon, 2 Jun 2008 18:51:58 -0400 MIME-Version: 1.0 Sent to CCL by: "Zsolt Zsoldos" [zsolt+/-simbiosys.ca] I agree with Steve and would like to add a furhter emphasis to the code compatibility issue regarding the Itanium processor. The processor is very different from the x86 architecture: http://www.nekochan.net/wiki/index.php/Itanium Even though older versions have an x86 compatibility emulation layer, but the performance of that is really bad and in newer versions the compatibility is dropped and software emulation is used instead: http://news.cnet.com/Intel-scraps-once-crucial-Itanium-feature/2100-1006_3-= 6028817.html Since you are looking for good performance, the code must be running natively on the IA64 architecture. So indeed, decide on the package first, then ask the developers/distributors whether thay support IA64 and what is the performance compared to other alternatives, e.g. an AMD Opteron. Zsolt Zsoldos http://www.simbiosys.ca/ On Mon, Jun 2, 2008 at 7:51 AM, Steven Gwaltney gwaltney-#-ccs.msstate.edu wrote: > > Sent to CCL by: Steven Gwaltney [gwaltney^_^ccs.msstate.edu] > The first question you need to answer is what code you intend to use. Thr= ee > I know of that can run massively parallel coupled-cluster calculations ar= e > NWChem, GAMESS, and ACES 3. I'm sure others exist, as well. Once you ha= d > chosen a code, then contact the authors. They should be able to give you > information about what kinds of architectures will run their code most > efficiently. > > In general, though, most of the massively parallel codes are designed to = run > on distributed memory computers. I can imagine some calculations that co= uld > efficiently use 512 GB of shared memory, but I can't think of any that wo= uld > require it. > > Steve > > Pablo Echenique echenique.p:gmail.com wrote: >> >> Dear friends and colleagues, >> >> at my laboratory, we are about to buy a computer for performing high-lev= el >> QC calculations (e.g., coupled cluster) in biological molecules (e.g., >> peptides). I write to kindly ask you for advice regarding the "best" (in >> whatever sense you consider appropriate) choice. >> >> One option would be a SGI Altix 4700 with 64 dual-core Itanium processor= s >> and 512GB of shared memory. >> >> The other option would be a number of smaller nodes, say 16 cores each, >> say 64GB of shared memory each. Gigabit or Infiniband connected. >> >> Some questions/remarks to orient the answers: >> >> 1. Are there QC problems that actually require 512GB of shared memory? >> Don't we face a CPU time bottleneck before that much memory is needed? >> 2. The second option would certainly present a lower cost per core. >> 3. The first option is said to be more easy to code parallel application= s, >> so I suppose that more QC codes can be efficiently used on the Altix tha= n on >> a multi-node architecture. >> 4. What specific machine would be the realization of the second option? >> >> Thank you in advance, >> >> >> -- >> Pablo Echenique >> >> Instituto de Biocomputaci=F3n y >> F=EDsica de los Sistemas Complejos (BIFI) >> >> Departamento de F=EDsica Te=F3rica >> Universidad de Zaragoza >> Pedro Cerbuna 12, 50009 Zaragoza >> Spain >> >> Tel.: +34 976761260 >> Fax: +34 976761264 >> >> echenique.p .. gmail.com >> http://www.pabloechenique.com > > -- > Dr. Steven Gwaltney Phone: 662-325-7602 > Assistant Professor Fax: 662-325-1618 > Department of Chemistry, Mail: Box 9573 > Center for Environmental Health Sciences, Mississippi State University > and HPCC Center for Computational Sciences Mississippi State, MS 39762 > > > > -=3D This is automatically added to each message by the mailing script = =3D-http://www.ccl.net/chemistry/sub_unsub.shtmlConferences: > http://server.ccl.net/chemistry/announcements/conferences/> > > From owner-chemistry@ccl.net Mon Jun 2 20:44:01 2008 From: "Brian Salter-Duke brian.james.duke[#]gmail.com" To: CCL Subject: CCL:G: problems in Message-Id: <-37080-080602192533-8955-vLQuCHF8xaZaTJwRdw4zCQ{:}server.ccl.net> X-Original-From: Brian Salter-Duke Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Tue, 03 Jun 2008 09:25:05 +1000 MIME-Version: 1.0 Sent to CCL by: Brian Salter-Duke [brian.james.duke]=[gmail.com] Raj S r.subramanian|,|ipc.uni-stuttgart.de wrote: > Sent to CCL by: "Raj S" [r.subramanian^ipc.uni-stuttgart.de] > Dear ccl users, > I am using Gaussian03 package for optimization and freq. calculations. After opt. and freq. calculations i notice in the .log file the total spin is marked ******. For e.g., look below > (A) (A) (A) (A) > of initial guess=******* > Requested convergence on RMS density matrix=1.00D-08 within1000 cycles. > Requested convergence on MAX density matrix=1.00D-06. > Requested convergence on energy=1.00D-06. > No special actions if energy rises. > Restarting incremental Fock formation. > SCF Done: E(UB+HF-LYP) = -2934.91465147 A.U. after 20 cycles > Convg = 0.9816D-08 -V/T = 2.0810 > S**2 = ******* > Annihilation of the first spin contaminant: > S**2 before annihilation 110.0343, after 110.0001 > 1 Symmetry operations used in ECPInt. > > I dont know what is it *******?Is this some error... It means the number is out of bounds of the format used to print. The problem is 110.0343 for S**2. This is way too high. What spin state do you have and what S**2 would you expect for a pure spin state (i.e. from ROHF not UHF)? Brian. > many thanks for your reply > raja> > > From owner-chemistry@ccl.net Mon Jun 2 23:49:01 2008 From: "Christopher Cramer cramer#%#umn.edu" To: CCL Subject: CCL:G: problems in Message-Id: <-37081-080602225706-1140-Po7VVrXiOm/fIOnJYe5r7w+*+server.ccl.net> X-Original-From: Christopher Cramer Content-Type: multipart/alternative; boundary=Apple-Mail-2-77062252 Date: Mon, 2 Jun 2008 21:26:19 -0500 Mime-Version: 1.0 (Apple Message framework v924) Sent to CCL by: Christopher Cramer [cramer(0)umn.edu] --Apple-Mail-2-77062252 Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit There is a curious bias for low spin amongst CCL readers. Irrespective of whether Gaussian's FORMAT statement ever envisioned 3 digits to the left of the decimal place in S^2, a value of 110 is perfectly reasonable. It implies 20 unpaired electrons all high-spin coupled ( S^2 = S * (S + 1) = 10 * 11 = 110). This is a very nice single determinantal wave function and might be expected in the case of, say, a tetranuclear manganese complex having 4 ferromagnetically coupled d5 Mn ions. Of course, the particular input deck of Raj S might indeed have an error, but there is nothing intrinsically wrong with the S^2 value. Instead, the formatting issue simply reflects what is likely to be some of the oldest (original?) code in Gaussian, written back in the day when small organic molecules where mostly the rule... Interestingly, the FORMAT used in the spin annihilation line is more flexible. CJC On Jun 2, 2008, at 6:25 PM, Brian Salter-Duke brian.james.duke[#]gmail.com wrote: > > Sent to CCL by: Brian Salter-Duke [brian.james.duke]=[gmail.com] > Raj S r.subramanian|,|ipc.uni-stuttgart.de wrote: >> Sent to CCL by: "Raj S" [r.subramanian^ipc.uni-stuttgart.de] >> Dear ccl users, >> I am using Gaussian03 package for optimization and freq. >> calculations. After opt. and freq. calculations i notice in >> the .log file the total spin is marked ******. For e.g., >> look below >> (A) (A) (A) (A) >> of initial guess=******* >> Requested convergence on RMS density matrix=1.00D-08 within1000 >> cycles. >> Requested convergence on MAX density matrix=1.00D-06. >> Requested convergence on energy=1.00D-06. >> No special actions if energy rises. >> Restarting incremental Fock formation. >> SCF Done: E(UB+HF-LYP) = -2934.91465147 A.U. after 20 cycles >> Convg = 0.9816D-08 -V/T = 2.0810 >> S**2 = ******* >> Annihilation of the first spin contaminant: >> S**2 before annihilation 110.0343, after 110.0001 >> 1 Symmetry operations used in ECPInt. >> I dont know what is it *******?Is this some error... > > It means the number is out of bounds of the format used to print. > The problem is 110.0343 for S**2. This is way too high. What spin > state do you have and what S**2 would you expect for a pure spin > state (i.e. from ROHF not UHF)? > > Brian. > >> many thanks for your reply >> raja> > > > > -= This is automatically added to each message by the mailing script > =- > To recover the email address of the author of the message, please > changehttp://www.ccl.net/chemistry/sub_unsub.shtml> > Search Messages: http://www.ccl.net/htdig (login: ccl, Password: > search)> > -- Christopher J. Cramer University of Minnesota Department of Chemistry 207 Pleasant St. SE Minneapolis, MN 55455-0431 -------------------------- Phone: (612) 624-0859 || FAX: (612) 626-2006 Mobile: (952) 297-2575 cramer/a\umn.edu http://pollux.chem.umn.edu/~cramer (website includes information about the textbook "Essentials of Computational Chemistry: Theories and Models, 2nd Edition") --Apple-Mail-2-77062252 Content-Type: text/html; charset=US-ASCII Content-Transfer-Encoding: quoted-printable There is a curious bias for low = spin amongst CCL readers.

Irrespective of whether = Gaussian's FORMAT statement ever envisioned 3 digits to the left of the = decimal place in S^2, a value of 110 is perfectly reasonable. It implies = 20 unpaired electrons all high-spin coupled ( S^2 =3D S * (S + 1) =3D 10 = * 11 =3D 110). This is a very nice single determinantal wave function = and might be expected in the case of, say, a tetranuclear manganese = complex having 4 ferromagnetically coupled d5 Mn = ions.

Of course, the particular input deck of = Raj S might indeed have an error, but there is nothing intrinsically = wrong with the S^2 value. Instead, the formatting issue simply reflects = what is likely to be some of the oldest (original?) code in Gaussian, = written back in the day when small organic molecules where mostly the = rule... Interestingly, the FORMAT used in the spin annihilation line is = more flexible.

CJC

On = Jun 2, 2008, at 6:25 PM, Brian Salter-Duke brian.james.duke[#]gmail.com = wrote:


Sent to CCL by: Brian Salter-Duke = [brian.james.duke]=3D[gmail.com]
Raj S = r.subramanian|,|ipc.uni-stuttgart.de wrote:
Sent to CCL by: "Raj  S" = [r.subramanian^ipc.uni-stuttgart.de]
Dear ccl users,
I = am using Gaussian03 package for optimization and freq. calculations. = After opt. and freq. calculations i notice in the .log file the total = spin <S**2> is marked ******. For e.g., look = below
(A) (A) (A) = (A)
<S**2> of initial = guess=3D*******
Requested = convergence on RMS density matrix=3D1.00D-08 within1000 = cycles.
Requested convergence = on MAX density matrix=3D1.00D-06.
Requested convergence on =             en= ergy=3D1.00D-06.
No special = actions if energy rises.
= Restarting incremental Fock formation.
SCF Done:  E(UB+HF-LYP) =3D  -2934.91465147 =     A.U. after   20 = cycles
=             Co= nvg  =3D    0.9816D-08 =             -V= /T =3D  2.0810
=             S*= *2   =3D  *******
Annihilation of the first spin = contaminant:
S**2 before = annihilation   110.0343,   after =   110.0001
=   1 Symmetry operations used in = ECPInt.
I dont know what is it = *******?Is this some error...

It means the number is = out of bounds of the format used to print. The problem is 110.0343 for = S**2. This is way too high. What spin state do you have and what S**2 = would you expect for a pure spin state (i.e. from ROHF not = UHF)?

Brian.

many thanks for = your reply
raja> =



-=3D This is automatically added to each = message by the mailing script =3D-
To recover the email address of = the author of the message, please change
the strange characters on = the top line to the /a\ sign. You can also
look up the X-Original-From: = line in the mail header.

E-mail to subscribers: CHEMISTRY/a\ccl.net or use:
=     http://www.ccl.ne= t/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST/a\ccl.net = or use
    http://www.ccl.ne= t/cgi-bin/ccl/send_ccl_message
http://www.ccl.net/c= hemistry/sub_unsub.shtml

Before posting, check wait time at: = http://www.ccl.net

Job: http://www.ccl.net/jobs = Conferences: http:/= /server.ccl.net/chemistry/announcements/conferences/

Search = Messages: http://www.ccl.net/htdig =  (login: ccl, Password: search)

If your mail bounces from = CCL with 5.7.1 error, check:
    http://www.ccl.net/spammers.txt

RTFI: http://www.cc= l.net/chemistry/aboutccl/instructions/


<= /div>


Christopher J. Cramer

University of = Minnesota

207 Pleasant St. SE

Minneapolis, MN = 55455-0431

Phone:  (612) 624-0859 || = FAX:  (612) = 626-2006

cramer/a\umn.edu

http://pollux.chem.umn.edu/~cr= amer

    of Computational = Chemistry:  Theories and Models, = 2nd Edition")




= --Apple-Mail-2-77062252--