From chemistry-request@server.ccl.net Thu Jul 6 00:02:10 2000 Received: from mail.science.siu.edu (mail.science.siu.edu [131.230.107.5]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id AAA02861 for ; Thu, 6 Jul 2000 00:02:09 -0400 Received: from tapas (131.230.99.83) by mail.science.siu.edu with SMTP (Eudora Internet Mail Server 2.2.2); Wed, 5 Jul 2000 23:02:17 -0500 Message-ID: <001c01bfe6fe$9f1b2060$5363e683@thchem.siu.edu> From: "Tapas Kar" To: "art'" Cc: References: <200007050652.IAA03759@wacx73.chemie.uni-wuerzburg.de> <3963A64B.3F25AC8A@vaidila.vdu.lt> Subject: Re: CCL:G94 Cube file Date: Wed, 5 Jul 2000 22:59:45 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2314.1300 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 Chem3D can read and draw nice picture of MOs, densities etc. __________________________________ Tapas Kar Department of Chemistry 1245 Lincoln Dr #146F Southern Illinois University Carbondale, IL 62901-4409 Tel: 618-453-6485 (Office)/6433(Lab) Fax:618-453-6408 Email: tapaskar@siu.edu tapas@risky3.thchem.siu.edu ----- Original Message ----- From: art' Cc: Sent: Wednesday, July 05, 2000 4:19 PM Subject: CCL:G94 Cube file > Hi, CCLers, > > There is lack of formation about CUBE in the help of Gaussian. > These are files with data of MO or density information in 3D lattice. What kind of > soft can visualise Cube files? How they should be interpreted or analized? > Any information concerning that is wellcome. > > I will summurise > > A.Ziemys > *Vytautas Magnus Universitu Ins. of Biochemistry > Dept. of Biology Lab. of Enzymatic chemistry > Vileikos 8 Mokslininku 12 > LT-3035, Kaunas Vilnius > Lithuania Lithuania > > > -= This is automatically added to each message by mailing script =- > CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To Admins > MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH > CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net 70 > Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: jkl@ccl.net > > > > > From chemistry-request@server.ccl.net Wed Jul 5 23:51:51 2000 Received: from alanina.df.ibilce.unesp.br (alanina.df.ibilce.unesp.br [200.145.203.25]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id XAA02813 for ; Wed, 5 Jul 2000 23:51:50 -0400 Received: from localhost (irbis@localhost) by alanina.df.ibilce.unesp.br (8.10.1/8.9.3) with ESMTP id e660xC431957; Wed, 5 Jul 2000 21:59:12 -0300 X-Authentication-Warning: alanina.df.ibilce.unesp.br: irbis owned process doing -bs Date: Wed, 5 Jul 2000 21:59:12 -0300 (BRT) From: David Santo Orcero X-Sender: irbis@alanina To: Iraj Daizadeh cc: chemistry@ccl.net Subject: Re: CCL:comutational precision: removing round-off errors. In-Reply-To: <39636FA4.CF35D7C0@genxy.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hello! > If one takes 625 and divides this number by, lets say, 5 on a computer > one may get one of > two numbers that look like: 125.000000000000000001 and > 124.4999999999999999 . > > Is this true? and, if so, how does one remove possible rounding errors Yes, it is true. And there is no solution, sorry. Here is why: Numbers in floating point format are something like: Number= (-1)^sign * (1.mantisa) * base ^ (exponent - kexp) Mantisa can be in other format, but uses to be as (1.mantisa); after a calculus, the number is LRed to its first digit is 1 (on binary format). Thus, that 1 doesn't need to be almacenated on the memory of the computer. This is called "normalized floating point". Nearly all computers uses this. Base uses to be 2; kexp, as far as I renember, 127 in IEEE754 32 bits and 1023 in IEEE754 64bits. 0 is not representable. For this, it is used a special kind of representation, that allows representate +Inf, -Inf and NaN. It allows also +0 and -0, but this is another story... Then, you will have a number representated as: (sign, mantisa, exponent) As you see, not all numbers are exactly representated. Some numbers that are simple on decimal representation, on floating point would need infinite precision. The propagation of the error by the non-exact nature of floating point numbers is really complex, and for really short integration steps can be 1 or 2 order of magnitude higher that the error for the truncation of the Taylor's serie by the integration algorithm. Sorry for my bad english. Sorry also by the bad news. ;-) I hope that this message helps. Yours: David From chemistry-request@server.ccl.net Thu Jul 6 02:30:25 2000 Received: from mail-relay-blr.ernet.in (mail-relay-blr.ernet.in [202.141.1.18]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id CAA03441 for ; Thu, 6 Jul 2000 02:30:21 -0400 Received: from helix.orgchem.iisc.ernet.in (IDENT:root@orgchem.iisc.ernet.in [144.16.77.3]) by mail-relay-blr.ernet.in (8.9.0/8.9.0) with ESMTP id LAA26336; Thu, 6 Jul 2000 11:59:39 +0530 (GMT+0530) Received: from localhost (srb@localhost) by helix.orgchem.iisc.ernet.in (8.8.7/8.8.7) with ESMTP id LAA10918; Thu, 6 Jul 2000 11:23:07 +0530 Date: Thu, 6 Jul 2000 11:23:07 +0530 (IST) From: "R.B.Sunoj" To: info@gaussian.com cc: CHEMISTRY@ccl.net Subject: CIS calculation Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hello, I was trying to do a CIS calculation with a 6-31+G* basis set using G-94. Unfortunately I always end up in getting the following job termination!. Please have a look at it and give your valuable suggestions. I will be happy to summarise the responses. Tail of the out put --------------------- Standard basis: 6-31+G(d) (6D, 7F) There are 187 symmetry adapted basis functions of A symmetry. Crude estimate of integral set expansion from redundant integrals=1.000. Integral buffers will be 131072 words long. Raffenetti 1 integral format. Two-electron integral symmetry is turned on. 187 basis functions 320 primitive gaussians 32 alpha electrons 32 beta electrons nuclear repulsion energy 400.4330949861 Hartrees. One-electron integrals computed using PRISM. There are 3 eigenvalues of the overlap less than 1.0D-05 The smallest eigenvalue of the overlap matrix is 3.606D-06 Projected INDO Guess. Requested convergence on RMS density matrix=1.00D-08 within 64 cycles. Requested convergence on MAX density matrix=1.00D-06. Integral accuracy reduced to 1.0D-05 until final iterations. Initial convergence to 1.0D-05 achieved. Increase integral accuracy. SCF Done: E(RHF) = -382.484431956 A.U. after 20 cycles Convg = 0.5149D-08 -V/T = 2.0024 S**2 = 0.0000 Range of M.O.s used for correlation: 10 187 NBasis= 187 NAE= 32 NBE= 32 NFC= 9 NFV= 0 NROrb= 178 NOA= 23 NOB= 23 NVA= 155 NVB= 155 **** Warning!!: The largest alpha MO coeffient is 0.16393177D+03 Semi-Direct transformation. ModeAB= 4 MOrb= 23 LenV= 3211000 LASXX= 16662856 LTotXX= 16662856 LenRXX= 33769037 LTotAB= 17106181 MaxLAS= 69753572 LenRXY= 0 NonZer= 50431893 LenScr= 78858840 LnRSAI= 69753572 LnScr1= 107841358 MaxDsk= 671088640 Total= 290222807 SrtSym= T fsync failed in NtrExt1. ------------------------------------------------------------- with best wishes, Sunoj =========================== R. B. Sunoj Dept of Organic Chemistry Indian Institute of Science Bangalore 560 012 =========================== From chemistry-request@server.ccl.net Thu Jul 6 04:47:32 2000 Received: from hydra.chem.rug.nl (hydra.chem.rug.nl [129.125.35.229]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id EAA04121 for ; Thu, 6 Jul 2000 04:47:31 -0400 Received: from [129.125.7.13] (theochem1.chem.rug.nl [129.125.7.13]) by hydra.chem.rug.nl (AIX4.2/UCB 8.7/8.7) with ESMTP id KAA08328 for ; Thu, 6 Jul 2000 10:47:23 +0200 (MET DST) Mime-Version: 1.0 X-Sender: swart@hydra.chem.rug.nl Message-Id: In-Reply-To: <3961ECCC.113AACCE@citrin.chemie.uni-dortmund.de> References: <3961ECCC.113AACCE@citrin.chemie.uni-dortmund.de> Date: Thu, 6 Jul 2000 10:49:34 +0200 To: CHEMISTRY@ccl.net From: Marcel Swart Subject: Re: CCL:geometry with an applied electrical field? Content-Type: text/plain; charset="us-ascii" ; format="flowed" >Dear CCLers! > >Does anyone know what quantumchemical program is able to optimize a >geometry with an applied electrical field? All programs that I have >used until now allow only single point calculations with an E-field. > >Thanks in advance, HONDO although not as fancy as the major programs, does the trick. Since GAMESS and HONDO originally come from the same program, I guess you could also try to use GAMESS, since this is maybe somewhat easier to obtain. -- ------------------------------------------------ drs. Marcel Swart Theoretical Chemistry (MSC) Molecular Dynamics (GBB) Rijksuniversiteit Groningen Chemistry Department Nijenborgh 4 9747 AG Groningen The Netherlands tel : +31 - (0)50 - 3634377 fax : +31 - (0)50 - 3634441 E-mail : m.swart@chem.rug.nl WWW: http://theochem.chem.rug.nl/~marcel/ ------------------------------------------------ From chemistry-request@server.ccl.net Thu Jul 6 05:02:30 2000 Received: from vytautas.vdu.lt (mail@vytautas.vdu.lt [193.219.38.33]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id FAA04214 for ; Thu, 6 Jul 2000 05:01:40 -0400 Received: from vaidila.vdu.lt ([193.219.38.52] ident=root) by vytautas.vdu.lt with esmtp for chemistry@ccl.net id 13A7V2-0005Dm-00; Thu, 06 Jul 2000 10:59:12 +0200 Received: from vaidila.vdu.lt (vejas.vdu.lt [193.219.38.82]) by vaidila.vdu.lt (8.9.3/8.9.3) with ESMTP id KAA10858 for ; Thu, 6 Jul 2000 10:59:12 +0200 (GMT) Message-ID: <3964C8D9.8A6282F5@vaidila.vdu.lt> Date: Thu, 06 Jul 2000 10:58:49 -0700 From: "art'" X-Mailer: Mozilla 4.7 [en] (Win98; I) X-Accept-Language: en MIME-Version: 1.0 CC: chemistry@ccl.net Subject: SUMMURY: G94 Cube file References: <200007050652.IAA03759@wacx73.chemie.uni-wuerzburg.de> <3963A64B.3F25AC8A@vaidila.vdu.lt> <001c01bfe6fe$9f1b2060$5363e683@thchem.siu.edu> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi CCLers, Thank you all, who repyed so quickly. gOpenMol was the most mention free software for visualisation of Cube files. Here is all answers: ************************************************ Tapas Kar wrote: > Chem3D can read and draw nice picture of MOs, densities etc. > __________________________________ > Tapas Kar > Department of Chemistry > 1245 Lincoln Dr #146F > Southern Illinois University > Carbondale, IL 62901-4409 > > Tel: 618-453-6485 (Office)/6433(Lab) > Fax:618-453-6408 > > Email: tapaskar@siu.edu > tapas@risky3.thchem.siu.edu *********************************************** You can see GOPENMOL package at http://www.csc.fi/~laaksone/gopenmol/gopenmol.html Andrea Giovanni Alparone PhD Universita di Catania Dipartimento di Scienze C *********************************************** The next version of MOLEKEL http://igc.ethz.ch/molekel/ will be able to visualize gaussian cube files. The new release will be based on OpenGL and not IRIX GL anymore, those will be much more flexible according platforms. The release is planed for this fall. Stefan -- Dr. Stefan Portmann Laboratory of Physical Chemistry ETH-Zurich Universitaetsstr. 65 CH-8092 Zurich Switzerland Phone: +41 (0) 1 632 57 82 Fax: +41 (0) 1 632 16 15 e-mail: portmann@igc.phys.chem.ethz.ch WWW: http://igc.ethz.ch/portmann *********************************************** ou can read and visualise this data for example with molden, gaussview, gOpenMol. I prefer gOpenMol. Have a look at http://www.gaussian.com/00000430.htm and http://www.csc.fi/~laaksone/gopenmol/gopenmol.html and http://www.caos.kun.nl/~schaft/molden/molden.html Stefan Konietzny Dipl.-Chem. Heinrich-Heine-Universit?t D?sseldorf Institut f?r Anorganische Chemie und Strukturchemie II Conference Software Address Default *********************************************** hello you can use OpenDX (www.opendx.org), the cube files generated by g94 are easily adapted to DX format, this is a .dx data file , the `#' commented lines are those on the original g94 cube file. #SCF Total Density # 8 -6.658018 -6.512752 -6.658017 # 86 0.190134 0.000000 0.000000 # 70 0.000000 0.190134 0.000000 # 86 0.000000 0.000000 0.190134 # 6 6.000000 0.000000 0.000000 0.000000 # 6 6.000000 0.000000 0.000000 2.813980 # 6 6.000000 2.813980 0.000000 2.813980 # 6 6.000000 2.813980 0.000000 0.000001 # 8 8.000000 -1.687280 0.000000 -1.687281 # 8 8.000000 -1.687282 0.000000 4.501260 # 8 8.000000 4.501261 0.000000 4.501261 # 8 8.000000 4.501260 0.000000 -1.687282 object 0 class gridpositions counts 86 70 86 origin -6.658018 -6.512752 -6.658017 delta 0.190134 0.000000 0.000000 delta 0.000000 0.190134 0.000000 delta 0.000000 0.000000 0.190134 object 1 class gridconnections counts 86 70 86 #attribute "element type" string "cubes" attribute "ref" string "positions" object 2 class array type float rank 0 items 517720 data follows 1.65385E-08 2.05079E-08 2.53339E-08 3.11707E-08 3.81895E-08 4.65774E-08 [...517720 cube data points...] attribute "dep" string "positions" object "density" class field component "positions" value 0 component "connections" value 1 component "data" value 2 end *********************************************** The program Molden can visualize Gaussian Cube files. have a look at URL: http://www.cmbi.kun.nl/~schaft/molden/mapped.html It discusses how to create an isodensity surface colorcoded with the electrostatic potential. It also explains how to generate the gaussian cube files and how to use them with molden. Molden can render the outcome as postscript, vrml, povray and OpenGL (you need the molden helper program moldenogl for that: http://www.cmbi.kun.nl/~schaft/molden/opengl.html For more information have a look at the Molden Home Page: http://www.cmbi.kun.nl/~schaft/molden/molden.html You can download the software at our anonymous ftp site: ftp.cmbi.kun.nl, /pub/molgraph/molden/molden3.6.tar.Z Best Regards, Gijs Schaftenaar Author of Molden *********************************************** Dear Arturas, There is at least a solution at no cost: GOpenMol. This software is freely available at the following address: http://laaksonen.csc.fi/gopenmol/gopenmol.html It works under MS Windows 95/98/NT, and offers many possibilities, apart from the nice visualization of Gaussian cube files (through conversion to the corresponding .plt files, which is accomplished from one of the menu options) Of course, there are also commercial possibilities. Apart from GaussView, the graphical interface to Gaussian, from Gaussian Inc., the last versions of Chem3D, from Cambridge Software, are able to read directly Gaussian cube files, leading to nice representations of MOs, Electrostatic Potentials, Total Densities, Spin densities, etc. Regards, Jose I. Garcia ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Dr. Jose Ignacio Garcia-Laureiro Depto. Quimica Organica. Instituto de Ciencia de Materiales de Aragon Facultad de Ciencias (edificio D). Universidad de Zaragoza-CSIC c/ Pedro Cerbuna, 12. E-50009 Zaragoza (Spain) Phone: +34 976762271 e-mail: jig@posta.unizar.es Fax: +34 976762077 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ *********************************************** The gOpenMol-package is one free and good option for drawing the 3D isosurfaces (and a lot more) from gaussian cube files. The program is available for most platforms. See http://www.csc.fi/~laaksone/gopenmol/gopenmol.html Cheers, Atte atte.sillanpaa@oulu.fi +358 (0)8 553 1681 (work), KE 368 Dept. of Physical Chemistry +358 (0)40 592 7369 (gsm) Po-BOX 3000 FIN-90014 University of Oulu FINLAND | Entropy requires no maintenance. | *********************************************** There is extensive discussion of the format of the Cubes generated by G94 in the G94 User's Reference starting on page 54 along with skeleton Fortran code to read one. There are a number of programs which can read this GaussViewW, PCModel and Chem3D under Windows. GaussView under Unix and Cerius 2 may still have this capability. You can also read these in a number of generic isodensity codes by modifying the header information. Check our Web site www.gaussian.com under the applications showcase and the details for a number of options are discussed. -- Douglas J. Fox Director of Technical Support help@gaussian.com *********************************************** Have a nice day A.Ziemys From chemistry-request@server.ccl.net Thu Jul 6 06:04:17 2000 Received: from mail-c.bcc.ac.uk (mail-c.bcc.ac.uk [144.82.100.23]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id GAA04495 for ; Thu, 6 Jul 2000 06:04:17 -0400 Received: from socrates-a.ucl.ac.uk by mail-c.bcc.ac.uk with SMTP (Mailer); Thu, 6 Jul 2000 11:03:34 +0100 From: uccatvm Message-Id: <14897.200007061004@socrates-a.ucl.ac.uk> Subject: Re: CCL:d-aug-cc-pVDZ for sulphur To: chemistry@ccl.net (CCL) Date: Thu, 6 Jul 2000 11:04:03 +0100 (BST) In-Reply-To: <006f01bfe6df$d164c180$2fa1f8cf@cimav.edu.mx> from "Dr. Daniel Glossman" at Jul 06, 2000 09:49:00 AM X-Mailer: ELM [version 2.5 PL3] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear Daniel, > Where can I find the coefficients of the d-aug-cc-pVDZ basis > set for sulphur? I have checked the EMSL data base but that > basis set for sulphur is lacking. The d-aug-cc-pVnZ sets can be easily derived from the aug-cc-pVnZ sets by addition of an extra diffuse function for each angular momentum symmetry present in the set (in your case: s,p,d). The exponents of the additional functions are given by alpha*beta (for d-aug-cc-pVnZ), alpha*beta*beta (for t-aug-cc-pVnZ), etc.. Alpha is the exponent of the diffuse function in the aug-cc-pVnZ, and beta is the ratio of the exponents of the two most diffuse functions in this set (beta < 1). See Woon and Dunning, J.Chem.Phys. 103, 4572 (1995). And also: Van Mourik, Wilson, and Dunning, Mol.Phys. 96, 529 (1999) Appendix. HTH, Tanja -- ==================================================================== Tanja van Mourik phone University College London work: +44 (0)20-7679-4665 Christopher Ingold Laboratories home: +44 (0)1895-259-312 20 Gordon Street e-mail London WC1H 0AJ work: T.vanMourik@ucl.ac.uk United Kingdom home: tanja@netcomuk.co.uk ==================================================================== From chemistry-request@server.ccl.net Thu Jul 6 11:53:04 2000 Received: from mail.science.siu.edu (mail.science.siu.edu [131.230.107.5]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id LAA07095 for ; Thu, 6 Jul 2000 11:53:04 -0400 Received: from tapas (131.230.99.83) by mail.science.siu.edu with SMTP (Eudora Internet Mail Server 2.2.2); Thu, 6 Jul 2000 10:53:05 -0500 Message-ID: <001301bfe761$ea608020$5363e683@thchem.siu.edu> From: "Tapas Kar" To: "R.B.Sunoj" , Cc: References: Subject: Re: CCL:CIS calculation Date: Thu, 6 Jul 2000 10:50:31 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2314.1300 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 Try with Cis=direct and MAXDISK=xxMB. The error is due to disk space limitation. Depending on your disk put xx values. __________________________________ Tapas Kar Department of Chemistry 1245 Lincoln Dr #146F Southern Illinois University Carbondale, IL 62901-4409 Tel: 618-453-6485 (Office)/6433(Lab) Fax:618-453-6408 Email: tapaskar@siu.edu tapas@risky3.thchem.siu.edu ----- Original Message ----- From: R.B.Sunoj To: Cc: Sent: Thursday, July 06, 2000 12:53 AM Subject: CCL:CIS calculation > > Hello, > > I was trying to do a CIS calculation with a 6-31+G* basis set using G-94. > Unfortunately I always end up in getting the following job termination!. > Please have a look at it and give your valuable suggestions. I will be > happy to summarise the responses. > > Tail of the out put > --------------------- > > Standard basis: 6-31+G(d) (6D, 7F) > There are 187 symmetry adapted basis functions of A symmetry. > Crude estimate of integral set expansion from redundant integrals=1.000. > Integral buffers will be 131072 words long. > Raffenetti 1 integral format. > Two-electron integral symmetry is turned on. > 187 basis functions 320 primitive gaussians > 32 alpha electrons 32 beta electrons > nuclear repulsion energy 400.4330949861 Hartrees. > One-electron integrals computed using PRISM. > There are 3 eigenvalues of the overlap less than 1.0D-05 > The smallest eigenvalue of the overlap matrix is 3.606D-06 > Projected INDO Guess. > Requested convergence on RMS density matrix=1.00D-08 within 64 cycles. > Requested convergence on MAX density matrix=1.00D-06. > Integral accuracy reduced to 1.0D-05 until final iterations. > Initial convergence to 1.0D-05 achieved. Increase integral accuracy. > SCF Done: E(RHF) = -382.484431956 A.U. after 20 cycles > Convg = 0.5149D-08 -V/T = 2.0024 > S**2 = 0.0000 > Range of M.O.s used for correlation: 10 187 > NBasis= 187 NAE= 32 NBE= 32 NFC= 9 NFV= 0 > NROrb= 178 NOA= 23 NOB= 23 NVA= 155 NVB= 155 > > **** Warning!!: The largest alpha MO coeffient is 0.16393177D+03 > > Semi-Direct transformation. > ModeAB= 4 MOrb= 23 LenV= 3211000 > LASXX= 16662856 LTotXX= 16662856 LenRXX= 33769037 > LTotAB= 17106181 MaxLAS= 69753572 LenRXY= 0 > NonZer= 50431893 LenScr= 78858840 LnRSAI= 69753572 > LnScr1= 107841358 MaxDsk= 671088640 Total= 290222807 > SrtSym= T > fsync failed in NtrExt1. > > ------------------------------------------------------------- > > with best wishes, > Sunoj > > =========================== > R. B. Sunoj > Dept of Organic Chemistry > Indian Institute of Science > Bangalore 560 012 > =========================== > > > -= This is automatically added to each message by mailing script =- > CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To Admins > MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH > CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net 70 > Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: jkl@ccl.net > > > > > From chemistry-request@server.ccl.net Thu Jul 6 13:09:14 2000 Received: from ivory.trentu.ca (trentu.ca [192.75.12.103]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id NAA07390 for ; Thu, 6 Jul 2000 13:09:13 -0400 Received: from trentu.ca ([204.225.13.50]) by trentu.ca (PMDF V5.2-32 #29543) with ESMTP id <01JRG4XT8HQA0075OF@trentu.ca> for chemistry@ccl.net; Thu, 6 Jul 2000 13:08:58 EDT Date: Thu, 06 Jul 2000 13:13:21 -0400 From: elewars Subject: WORDS LOCAL AND NONLOCAL:SUMMARY To: chemistry@ccl.net Message-id: <3964BE31.E9DE0EA3@trentu.ca> MIME-version: 1.0 X-Mailer: Mozilla 4.7 [en] (WinNT; I) Content-type: text/plain; charset=us-ascii Content-transfer-encoding: 7bit X-Accept-Language: en Thurs, 2000 July 6 Thanks very much to all who responded to my question. The answers, #1--8, are summarized below. I hope I haven't overlooked anyone. EL ---- Original question: Wednesday, 2000 June 21 Hello, Does any one know why DFT functionals without gradient corrections are called _local_, while gradient-corrected functionals are called _nonlocal_? I am not asking about the theory behind these, but why these terms are used. I know that nonlocal has a certain meaning in quantum physics; does it have some special meaning in connection with mathematical functions or functionals? The term nonlocal applied to gradient-corrected functional has been said to be mathematically wrong (A. St-Amant, Reviews in Computational chemistry, vol. 7, chapter 5, 1996; p. 223). Thanks E. Lewars ===== Answers: Andrew T. Pudzianowski, andrew.pudzianowski@bms.com #1 Dr. Lewars - The term "local" comes from the local density approximation (LDA) in DFT, where it's assumed that the electron density at any given point in a molecule is the same as that in an electron gas, a system which has been solved exactly. The LDA provides correlation and is more accurate than the Hartree-Fock approximation in many cases, but it tends to "overbind" systems, i.e., it leads to overestimates of bond energies and complexation energies. The LDA is particularly bad in describing H-bonded systems. The term "nonlocal" has been used to refer to anything that goes beyond the LDA, including gradient-corrected functionals. Some of these have "local" components derived from the electron gas description, but I don't know if that's what's objectionable about the use of "nonlocal" terminology with these functionals. Regards, Andrew Pudzianowski --------- #2 Huub van Dam, h.j.j.vandam@dl.ac.uk The term nonlocal originates from the fact that in addition to the local terms you have gradient terms which say something about how the functional behaves in the immediate vicinity of the point where you're at. E.g. compare gradient corrected functionals with a first order Taylor expansion. The reason I wouldn't use the term nonlocal is because of the word "vicinity". In practice the gradient corrected functionals are still so local that any long range correlation effect (e.g. dispersion) isn't accounted for at all. So in fact these functionals are much more local than the name nonlocal suggests. So if you really don't want to write "gradient corrected" all the time than the term "semi-local" wouldn't be too bad (although strictly mathematically that is probably wrong as well, because even gradient corrected functionals contain only quantities evaluated at 1 particular point even though it is a density and the norm of the gradient). ------------------------ #3 Thomas Bligaard Pedersen, bligaard@fysik.dtu.dk Hello, Locality in physics is often related to closeness in the time-space continuum - that a theory should yield its results without taking information into account that is not accessible due to information travelling only at the speed of light or slower. Accepting general relativity to be an axiom, means that all correct theories of nature should be local, which is not yet the case for quantum mechanics. This meaning of locality would then principally be of interest in TDFT, but nobody would really care about it, as it is related to extremely small corrections of general relativity, while it is plentily difficult to include all special relativistic terms. For many years the most used functionals in solid state physics were LDA and LSDA. That's probably why everything else going beyond the LSDA got to be called non-local, including the GGAs which are mathematically local. Interestingly Walter Kohn calls GGAs quasi-local when he speaks about them. Best regards Thomas B. Pedersen ------- #4 Lou Noodleman lou@scripps.edu Dear Dr. Lewars, Yes, words can cause big problems. DFT functionals without gradient corrections are called local, precisely because there are no gradient terms (and gradient terms in some sense sample the surroundings of a given point); gradient-corrected functionals are called nonlocal for exactly this reason. There is some confusion attending this, since in quantum chemistry, "nonlocal" is also used for situations where each orbital sees a different exchange potential (or exchange- correlation potential). The latter meaning is true for Hartree-Fock (and also probably in a number of explicitly correlated methods like MC-SCF and CI methods), but not in DFT methods either LDA or gradient- corrected (GGA) methods. Hybrid Hartree-Fock-DFT methods like B3LYP should be like Hartree-Fock, nonlocal in that each orbital sees a different exchange-correlation potential in the one-electron SCF equations. Nonlocal terms cause the worse scaling behavior of Hartree-Fock (n**4) versus DFT (n**3) as a function of the number of active electrons in the molecule. However, there are fitting tricks (and threshold tricks) that can help to control this worse scaling behavior in ab initio methods (as implemented in local MP2 methods in Jaguar), and also there are other tricks in Turbomole (I hear). A general rule in quantum chemistry is: "be sure you know how a technical term is being used in a particular context, what the author really means." Further, realize that the same term can be used to mean different things by different people in different contexts. For example, "correlation" can mean whatever is left out of Hartree-Fock (but is this spin restricted or unrestricted or broken symmetry?), but there are other equally valid frameworks. "Fermi correlation" has a statistical meaning (see McWeeney), and in my view, "correlation" as defined in a statistical context as in the theory of reduced density matrices has a much deeper meaning than the usual "correction to Hartree-Fock" meaning. Lou Noodleman Molecular Biology TPC15 The Scripps Research Institute La Jolla, CA 92037 . ------------------- #5 Ross Dickson, rdickson@molecularmining.com I believe the term "local" derives from the "local density approximation" (LDA), of Kohn & Sham (1965, Phys. Rev. A 1133). Although the idea is theirs, I can't say for certain that the term originated with them also, as I haven't the paper easily at hand. I further suspect that the term "non-local" arose as a casual or inexact means of distinguishing the non-(local density approximation) functionals from the at-the-time dominant LDA. (I hope TeX notation is clear enough for this.) A consequence of the Hohenberg-Kohn theorem is that e_{xc}(x) = e_{xc}[\rho]: The exchange-correlation energy density at a point x is a _functional_ of the charge density at all points. The local density approximation amounts to supposing that e_{xc}(x) = e_{xc}(\rho(x)), i.e. that the exchange-correlation energy density at a point x can be adequately approximated by a _function_ of the charge density at that point x. It's local in the very concrete sense of "doesn't depend on anything far away in coordinate space." The various degrees of gradient correction can be characterized as e_{xc}(x) = e_{xc}(\rho(x), \rho'(x), \rho''(x), ...) and so on: You're still using a function, not a functional, for e_{xc}, but employing more and more derivatives of the charge density as arguments (signified by ', '', etc.) This, I think, is why Alain St.-Amant suggested that "non-local" is a bad term for this family of functionals, and I am inclined to agree. One is still working with the charge density in the locality of the point at which one is evaluating the energy density (or the exchange-correlation potential), even though one is no longer using the local density approximation. Cheers, Ross Dickson, no longer a quantum chemist... ----------- #6 Per-Ola Norrby, peo@dfh.dk Hi, I think the term "local" in this context means that the value of the function in a mathematical point depends on that point only. The gradient is by definition the difference between the function values for two points separated by an infinitesimal distance. That is, the gradient and any functional depending on it can never be exactly "local". In principle, there could also be other types of "non-local" functionals, but I believe that the gradient correction is the first "non-local" fix that anybody ever introduces. I'm not aware of any non-local functional without gradient correction, but there could well be. Caveat: I'm a chemist, so my language is probably not mathematically correct... Per-Ola Norrby -- ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ * Per-Ola Norrby, Associate Professor * The Royal Danish School of Pharmacy, Dept. of Med. Chem. * Universitetsparken 2, DK 2100 Copenhagen, Denmark * Tel. +45-35306506, fax +45-35306040 * Email: peo@dfh.dk (preferred), peo@compchem.dfh.dk * WWW: http://compchem.dfh.dk/PeO/ ------------------- #7 Herbert Fruchtl, fruechtl@fecit.co.uk There is a mathematical theorem which states that if you know all derivatives of an 'analytical function' (i.e. one that can be expressed in a power series) in one point, you know the function in its complete definition range. Nonlocal functionals depend not only on the value of the density, but also its derivatives (mostly only the first one, the gradient). The gradient contains information about the surrounding area, not only the point you look at. So the expression nonlocal my be not very intuitive, but I don't think it's wrong. HTH, Herbert -------------- #8 Bert te Velde, tevelde@scm.com Hi, The 'local' functionals all use (only) the value rho(r), i.e. the density in a particular point in space. The formula for the density-functional potential (and/or) the energy-density in a given point in space is a more or less simple formula of rho(r), the charge density in the same point. The term 'non-local' has been used to stress that the (original) non-local gradient corrections used not only rho(r) but also its derivatives, so you could see it as implying that not only the local value of the density went into the formula but also (implicitly, through the derivatives) values at other (nearby) points in space: non only the LOCAL value of rho. Critics of the phrase 'non-local' refer to the fact that the potential (or energy-function) is not in itself a non-local function: it still is a 'simple' function v(r), respectively e(r). Consequently, the term 'nonlocal' has gradually disappeared and has been replaced by the more appropriate phrase 'gradient-corrected', or 'generalized gradient approximation (GGA)'. Best regards Bert ============ ========= From chemistry-request@server.ccl.net Thu Jul 6 13:31:32 2000 Received: from ivory.trentu.ca (trentu.ca [192.75.12.103]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id NAA07529 for ; Thu, 6 Jul 2000 13:31:31 -0400 Received: from trentu.ca ([204.225.13.50]) by trentu.ca (PMDF V5.2-32 #29543) with ESMTP id <01JRG5PEHT6E00758S@trentu.ca> for chemistry@ccl.net; Thu, 6 Jul 2000 13:31:12 EDT Date: Thu, 06 Jul 2000 13:35:36 -0400 From: elewars Subject: X-ALPHA, DFT TODAY:SUMMARY To: chemistry@ccl.net Message-id: <3964C368.F3E62E02@trentu.ca> MIME-version: 1.0 X-Mailer: Mozilla 4.7 [en] (WinNT; I) Content-type: text/plain; charset=us-ascii Content-transfer-encoding: 7bit X-Accept-Language: en Thurs 2000 July 6 Here is a summary of the answers to my X-alpha question. I thank all who responded (#1--#4). Question: > > Monday 2000 June 12 > > > > Hello, > > > > the Xalpha method, developed by John slater (1951) mainly for solids and > > atoms, and used till at least the early 1970's , was perhaps the first > > really useful DFT method. For molecules it has now been replaced by > > Kohn-Sham-based approaches to DFT. > > > > (1) Is the Xalpha method still used for solids/crystals (e.g. by > > solid-state physicists?). If not, what replaced it? > > (2) In modern molecular DFT, probably the most popular method is > > B3LYP/6-31G*. What seems to be the second most widely-used? > > > > Thanks > > E. Lewars ==== Answers #1 Dr Robert J. Doerksen, rjd@bastille.cchem.berkely.edu > Dear Dr. Lewars, > > I just have a small comment on (1): > You may be amused by the article Dennis Salahub wrote giving a > retrospective on the X-Alpha method, in Theoretical Chemistry Accounts, > in their 20th century highlights issue: > > Dennis R. Salahub: From X-scattered wave to end-of-the-century > applications of density functional theory in chemistry. Perspective on > "Chemical bonding of a molecular transition-metal ion in a crystalline > environment" (Johnson KH, Smith FC Jr (1972) Phys Rev B 5: 831-843) > > That's in: Theor Chem Acc 103 (2000) 3/4, 311-312 > > Cheers, > > Robert > > Dr. Robert J. Doerksen ************************************************* > Head-Gordon Group * > Department of Chemistry work phone (510) 642-9304 > University of California fax (510) 643-6232 > Berkeley, CA, USA 94720 e-mail: rjd@bastille.cchem.berkeley.edu > **** webpage: http://www.cchem.berkeley.edu/~mhggrp/doerksen/rjdhome.htm --- #2 Per-Ola Norrby, peo@dfh.dk Hi, > > >(2) In modern molecular DFT, probably the most popular method is > >B3LYP/6-31G*. What seems to be the second most widely-used? > > I'd vote for BP86. It's been around longer, is almost as > accurate, and substantially faster (at least in some codes...). > > /Per-Ola > -- > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > * Per-Ola Norrby, Associate Professor > * The Royal Danish School of Pharmacy, Dept. of Med. Chem. > * Universitetsparken 2, DK 2100 Copenhagen, Denmark > * Tel. +45-35306506, fax +45-35306040 > * Email: peo@dfh.dk (preferred), peo@compchem.dfh.dk > * WWW: http://compchem.dfh.dk/PeO/ ------------ #3 Thomas Heine, Thomas.Heine@chiphy.unige.ch > (1) mainly LDA, later on GGA was used as well > (2) B3LYP is the mostly used by chemists, it is almost never used by > physicists, since (i) they don't like HF exchange since it makes the > programs very slow, (ii) it just works well for lighter elements. > Perdew-Wang 96, PW91 and the newer PBE is used quite often, I think in a > competitive amount with old-fashioned LDA (e.g. VWN). > > Thomas > University of Geneva > > PS: Let me know what others tell you --------- #4 Thomas Bligaard Petersen, bligaard@fysik.dtu.dk Hello, > > I propose a response to question 1: > > The Xalpha was also replaced by the Kohn-Sham approch. In solid state physics the > LDA (of Kohn-Sham) was the most used exchange-correlation functional. The role was > taken over by the LSDA (Gunnarson and Lundquist) that describes magnetic systems > better. These are still used, and especially in more theoretical part of solid > state theory, as they are some of the only true ab-initio approches in DFT. For > applications the PW91 (Perdew-Wang) has become very important, as it corrects some > important deficiencies of the LDA. The PBE (Perdew-Burke-Ernzerhofer) removes > some problems of oscillating xc-potential in the PW91, whereby the numerical > behaviour of the calculations are somewhat improved. Other then that it yields > results very similar to PW91, and has more or less replaced the use of PW91 today. > The B3LYP yields quite bad results in solids, this is probably related to the > Hartree-Fock approch having some problems here. In the future orbital dependent > potentials are probably going to have growing importance in solid state physics, > but the selfconsistent implementation of these functionals are not yet in common > use, even if the functionals appear to give vastly improved results for some > systems. One such functional is the PKZB meta-GGA (Perdew-Kurth-Zupan-Blaha). > > In short the LSDA and the PBE are the most used today. ================= ========= From chemistry-request@server.ccl.net Thu Jul 6 04:15:00 2000 Received: from c008.sfo.cp.net (c008-h017.c008.sfo.cp.net [209.228.14.206]) by server.ccl.net (8.8.7/8.8.7) with SMTP id EAA03976 for ; Thu, 6 Jul 2000 04:15:00 -0400 Received: (cpmta 19908 invoked from network); 6 Jul 2000 01:14:51 -0700 Received: from pcb4122.univ-lemans.fr (193.52.30.155) by smtp.cristal.org (209.228.14.206) with SMTP; 6 Jul 2000 01:14:51 -0700 X-Sent: 6 Jul 2000 08:14:51 GMT Message-Id: <4.2.0.58.20000706101248.00976bf0@mail.cristal.org> X-Sender: pop@mail.cristal.org X-Mailer: QUALCOMM Windows Eudora Pro Version 4.2.0.58 Date: Thu, 06 Jul 2000 10:14:54 +0200 To: CHEMISTRY@ccl.net From: Armel Le Bail Subject: Torsion angle Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Hello, Is there some open source code (or part of code) allowing to locate automatically where can occur possible variation of torsion angles in a given molecule ? Fortran preferably. Thanks. -- Armel Le Bail http://www.cristal.org/ From chemistry-request@server.ccl.net Thu Jul 6 04:40:46 2000 Received: from hydra.chem.rug.nl (hydra.chem.rug.nl [129.125.35.229]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id EAA04078 for ; Thu, 6 Jul 2000 04:40:46 -0400 Received: from [129.125.7.13] (theochem1.chem.rug.nl [129.125.7.13]) by hydra.chem.rug.nl (AIX4.2/UCB 8.7/8.7) with ESMTP id KAA22616 for ; Thu, 6 Jul 2000 10:40:37 +0200 (MET DST) Mime-Version: 1.0 X-Sender: swart@hydra.chem.rug.nl Message-Id: In-Reply-To: <39636FA4.CF35D7C0@genxy.com> References: <39636FA4.CF35D7C0@genxy.com> Date: Thu, 6 Jul 2000 10:42:49 +0200 To: CHEMISTRY@ccl.net From: Marcel Swart Subject: Re: CCL:comutational precision: removing round-off errors. Content-Type: text/plain; charset="us-ascii" ; format="flowed" >Hello. > >If one takes 625 and divides this number by, lets say, 5 on a computer >one may get one of >two numbers that look like: 125.000000000000000001 and >124.4999999999999999 . > >Is this true? and, if so, how does one remove possible rounding errors >produced when >this occurs during lets say integration.... This is true, but depends on the machine on which you run the program. However, if you use FORTRAN90, you have somewhat control of the error. If you use selected_real_kind(15,70) to define the kind of all reals, you can check if that kind is available on that machine. The 15,70 means 15 decimals precision, and exponent up to power 70. From chemistry-request@server.ccl.net Thu Jul 6 06:33:48 2000 Received: from stark.udg.es (stark.udg.es [130.206.128.61]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id GAA04643 for ; Thu, 6 Jul 2000 06:33:47 -0400 Received: from iqc.udg.es (woodstock.udg.es [130.206.128.118]) by stark.udg.es (980427.SGI.8.8.8/980728.SGI.AUTOCF) via ESMTP id DAA93506 for ; Thu, 6 Jul 2000 03:34:46 -0700 (PDT) Message-ID: <39646F75.52C52C81@iqc.udg.es> Date: Thu, 06 Jul 2000 12:37:25 +0100 From: Pedro Salvador X-Mailer: Mozilla 4.7 [en] (Win95; I) X-Accept-Language: en MIME-Version: 1.0 To: Computational Chemistry List Subject: Problem compiling G98 on Linux Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear all I've got the following message error when linking the g98 on a Linux RedHat machine: g77 -O2 -march=pentiumpro -malign-double -m486 -fexpensive-optimizations -O3 -ffast-math -funroll-loops -c ml0.f rm -f ml0.f ml0.c make[1]: Leaving directory `/home/perico/g98' g77 -O2 -march=pentiumpro -malign-double -m486 -fexpensive-optimizations -O3 -ffast-math -funroll-loops -o g98 ml0.o util.so util.so: undefined reference to `wait_' util.so: undefined reference to `fork_' collect2: ld returned 1 exit status make: *** [g98] Error 1 endif It seems that it can't resolve a couple of functions which however are defined in the mdutil.c file. Are they essential? Any suggestions? thank you in advance Pedro -- /-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\ I Pedro Salvador Sedano I I \ Ph.D. Student I Butch: You okay? / I Universitat de Girona I Marcellus: No. I'm pretty I / Institut de Quimica Computacional I fuckin' far from okay. \ I E-mail: perico@stark.udg.es I I \ WWW :http://stark.udg.es/~perico I Pulp Fiction / I Telf: +34 972 418358 I I \-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/-\-/ From chemistry-request@server.ccl.net Thu Jul 6 06:49:32 2000 Received: from cmcind.pharm.uu.nl (cmcind.pharm.uu.nl [131.211.56.43]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id GAA05284 for ; Thu, 6 Jul 2000 06:49:31 -0400 Received: from cmcind (localhost [127.0.0.1]) by cmcind.pharm.uu.nl (980427.SGI.8.8.8/970903.SGI.AUTOCF) via ESMTP id MAA23637 for ; Thu, 6 Jul 2000 12:49:21 +0200 (MDT) Message-Id: <200007061049.MAA23637@cmcind.pharm.uu.nl> To: chemistry@ccl.net Reply-To: Malcolm Gillies Subject: Summary: OpenGL on Linux and Insight/Sybyl In-reply-to: <200007030955.LAA12754@cmcind.pharm.uu.nl> Date: Thu, 06 Jul 2000 12:49:21 +0200 From: Malcolm Gillies I've compiled a (slightly edited) list of replies to my question, as well as some ensuing discussions. Thank you to everyone who responded. The general impression I gained is that hardware accelerated OpenGL on Linux has similar performance to SGI for many molecular graphics applications, and that people who are using it are quite satisfied. In particular, various in-house and non-commercial programs are reported to work well. Unfortunately, noone seems to have succeeded in getting MSI Insight (aka Biosym InsightII) working with a remote Linux OpenGL display. At first glance, this appears to be a minor problem with Insight recognizing the remote display type correctly, but there may also be other more subtle difficulties to overcome. I'd be interested to know MSI's response to this problem. I received one clear report that Sybyl can be used in the way I propose, though apparently there is a problem with object selection. Given that Tripos are supporting configurations using a PC X server with OpenGL (e.g. Exceed 3D), I would expect that fixes or workarounds shouldn't be too difficult. I didn't hear from anyone who is using the new SGI Visual Workstations. -- Malcolm Gillies PhD student, computational medicinal chemistry Dept Medicinal Chemistry, Faculty of Pharmacy, Utrecht University, The Netherlands ------- Messages Date: Mon, 03 Jul 2000 11:55:21 +0200 From: Malcolm Gillies Subject: CCL:OpenGL on Linux and Insight/Sybyl - any experiences? Message-Id: <200007030955.LAA12754@cmcind.pharm.uu.nl> I'm interested in hearing about any experiences using Linux with an X server supporting OpenGL/GLX as a remote display for modelling programs such as MSI Insight or Tripos Sybyl. In particular, I'd like to hear if the following kind of setup is workable in practise: * Insight running on SGI * Linux 2.2.* kernel on x86 * Nvidia video card (GeForce 256/GeForce 2 GTS/Quadro) * Nvidia OpenGL Linux reference drivers and Xfree86 But I'm also interested in information on other video cards, drivers, or the new SGI Visual Workstations (VPro graphics) under Linux. If this sort of configuration is practical, then it could be a cheap way of expanding a modelling lab with extra workstations for light use by students, etc. ------- Message 2 Date: Mon, 03 Jul 2000 14:29:59 +0200 From: Sigismondo Boschi Message-Id: <39608747.8A7F4472@cineca.it> Dear Malcolm, I have installed the glx module for the Matrox G400. It is a module in development, for Xfree 3.x You can find it and information about it at http://www.matroxusers.com/ It works quite fine interfacing MSI cerius2 (InsightII does not recognize it giving the error "Couldn't get initial visual in app_init_display" indeed! Xvision for windows does the same) running both on IBM rs6000 (SP) and Origin2000. The main lack is the "depth cueing". For the rest it works fine, very fast rendering atoms and isosurfaces, comparable with an SGI O2. The PC is a PentiumIII 450MHz, Linux Redhat 6.1, XFree86 3.3.5, matrox G400 single-head 16MB. Also other software runs fine: gaussview and gOpenMol for example. Checking now the site "matroxusers" I have found they have included the DRI driver (direct rendering interface) also for this board in the very new Xfree86 4.0.1. More information at http://www.xfree86.org and in particular http://www.xfree86.org/4.0.1/RELNOTES.html Actually I have not tried it: I have found it just now. I will let you know if in the next days I will be able to try it. Please, post the summary to CCL! Thanks Kind regards Sigismondo -- ------------------------------------------------------------------------- |Sigismondo Boschi, Ph.D. |tel: +39 051 6171559 | |CINECA (High Performance Systems) |fax: +39 051 6137273 - 6132198 | |via Magnanelli, 6/3 | | |40033 Casalecchio di Reno (BO) ITALY |WWW: http://www.cineca.it | ------------------------------------------------------------------------- ------- Message 3 Date: Mon, 03 Jul 2000 09:27:47 -0400 From: "Bienstock.Rachelle" Message-Id: <4D693F933DD8D311A18C00E018B0057652C825@trollope.niehs.nih.gov> Malcolm: Although I am by no means an expert in this area, it was my impression that the MSI Insight software package currently available will not run on Linux. First, because it is commercial software package, you are only given the executibles which are currently only sold and available as compiled code under Irix. MSI is working on a commerical version which will be available for Linix. You may want to consult with MSI to see when this will be available. Rachelle ------- Message 4 Date: Mon, 03 Jul 2000 09:54:06 -0400 From: "Bienstock.Rachelle" Message-Id: <4D693F933DD8D311A18C00E018B0057652C827@trollope.niehs.nih.gov> > I'm aware that neither Insight nor Sybyl will run on Linux. What I'm > actually proposing is that the software continue to run on an SGI > machine, but with a remote Linux X server as display. Yes, I guess it should work. Actually, I know at several universities they have been running the MSI Insight software using the Hummingbird Exceed software on the PC as an X-terminal emulator. ------- Message 5 Date: Mon, 03 Jul 2000 10:07:00 -0400 From: "Bienstock.Rachelle" Message-Id: <4D693F933DD8D311A18C00E018B0057652C828@trollope.niehs.nih.gov> > Right, the crux of the question is if they are using -opengl graphics or > -axxess X graphics; the latter has been possible for some time, but comes > at the cost of much more primitive graphics. > > Do you know if they are using a version of Exceed which supports OpenGL > graphics (I think it's called Exceed 3D)? You know,I believe it was using the Axxess X graphics. At one time I had thought of setting this up (to run the software at home on my PC) and received lots of replies from people who were using the Insight II software this way with Exceed. If you search the Comp. Chem. archives you may find some info. Everyone told me though that the software ran so slowly that it was almost to the point of being unusuable- i.e. rotating large molecules etc. ------- Message 6 Date: Mon, 03 Jul 2000 09:40:34 -0400 From: Zsolt Zsoldos Message-Id: <396097D2.4F791EB0@simbiosys.ca> Dear Malcolm, My experience is not with Nvidia card, so I cannot comment on its performance. However, we have an SGI Indigo2 Impact and a number of Linux computers. One of the Linux machines have an ELSA Gloria L/MX graphics card and we use it with the XiG 3D Accelerated-X OpenGL driver and X server combination (see http://www.xig.com/). This combination on a 2.2.5-15smp kernel 2x350MHz Pentium II computer can provide a very comparable performance to the Impact. We use it with our own modelling package, not Insight or Sybyl as we are a computational chemistry software company ourselves. We use standard OpenGL/GLX graphics in our software which runs on both the SGI and the Linux-PC. We use it in every combination (running the software on either platform with both local and remote display) and each combination works fine for the OpenGL graphics. On the other hand, we experience some color mapping problems with non-OpenGL graphics (menubars, button colors etc) wen running on Linux and remote displaying on SGI. But this might be to do with the Java Swing library. The mentioned XiG X server provides drivers for a large number of 3D accelerated OpenGL cards, including the cheap Permedia2 based cards, like Diamond FireGL 10 00Pro. Those definitely provide a very cheap alternative to SGI with usable performance. I hope you find this information useful. I'd be very interested to see a summary of the replies you get, especially if you hear anything about the SGI Visual Workstations graphics under Linux. Best regards, Zsolt - -- Zsolt Zsoldos, PhD. Simbiosys Inc. phone: 1 - 416 - 741-4263 135 Queen's Plate Dr., Unit 355 fax: 1 - 416 - 741-5084 Toronto, Ont, M9W 6V1, Canada http://www.simbiosys.ca/ ------- Message 7 Date: Mon, 03 Jul 2000 10:56:56 -0300 From: antonio luiz oliveira de noronha Message-Id: In-Reply-To: <200007030955.LAA12754@cmcind.pharm.uu.nl> Mr. Gillies In my lab we use Linux MANDRAKE 7.1 and we had in our workstations TRIPOS/SYBYL and MSI/INSIGHTII. Both work fine with the linux box. Our video card is pretty simple, it's a Trident with 2 M. Also, vwe only use the PCs with linux as terminals, but work very fine! Best Regards!! ___________________________________________________________________ | | |Antonio Luiz Oliveira de Noronha | |Medicinal Chemistry of Organotin Compounds | |NEQUIM - Medicinal Chemistry Group | |Chemistry Departament - DQ | |UFMG - Federal University of Minas Gerais - Brazil | |Av. Antonio Carlos, 6627 - Campus Pampulha | |CEP 31270-901 - Belo Horizonte - MG - Brazil | |VOICE +55 31 499 5765 - FAX +55 31 499 5700 | |___________________________________________________________________| ------- Message 8 Date: Mon, 03 Jul 2000 16:03:16 -0300 From: antonio luiz oliveira de noronha Message-Id: Mr. Gillies I check it. Yup, the INSIGHT does not work, but the SYBYL works fine, using the x graphics. Regards!! ------- Message 9 Date: Mon, 03 Jul 2000 15:58:52 +0200 From: Ezequiel Quintana Morales Message-Id: <39609C1C.E1DE1154@iiq.cartuja.csic.es> Hi, We use on a daily basis, linux boxes as terminals of SGI's running Sybyl. We use SuSE Linux starting from 6.2 to 6.4, and Nvidia TNT2 or GeForce 256 DDR video cards. For us that is the best option (the cheapest). They all run fine (GeForce better, :-)). The configuration is very simple and y ou only need to load the iris fonts into the linux machines. However, we have found that running Sybyl in OGLX mode, the selection on screen of objects can hung the SGI. In XDEV mode there is no problem. Hope this help. If you want more details, please let me know. Regards, Ezequiel - -------------------------------------------------------------------- Dr. Ezequiel Quintana Morales Instituto de Investigaciones Quimicas Centro de Investigaciones Cientificas Isla de la Cartuja c/ Americo Vespucio, s/n 41092 Sevilla Tel: + 34 95 4489564 Fax: + 34 95 4460565 - -------------------------------------------------------------------- ------- Message 10 Date: Mon, 03 Jul 2000 17:59:12 +0200 From: Ezequiel Quintana Morales Message-Id: <3960B850.EF602CF6@iiq.cartuja.csic.es> Malcolm, > I guess you are using the Nvidia reference Linux driver. Is it > reasonably stable? I don't use the nvidia driver but the one that come with the SuSE distribution. I tried several: Utah-glx and nvidia drivers but SuSE in collaboration with UTAH-glx has one which works fine and it's easier to install (just from the CD). > > They all run fine (GeForce better, :-)). The configuration is very > > simple and y ou only need to load the iris fonts into the linux > > machines. However, we have foun d that running Sybyl in OGLX mode, > > the selection on screen of objects can hung th e SGI. In XDEV mode > > there is no problem. > > That's interesting. So the SGI hangs, rather than the X server on > the Linux machine? > > I was hoping that things would work stably in OGLX mode. I don't know if you has used Sybyl too much, but it's a very tricky software. If you select objects using the dialog boxes everything is ok, but if you need to point with the mouse on the screen Sybyl crashes and most of the times hangs the computer (OGLX mode). For TRIAD (NMR module of Sybyl), the best configuration I had tested is XDM/Fvwm2(or mwm), but never KDM/KDE, and if you want a working rubber band you need to have activated Num Lock. > If possible, could you tell me which software versions you are using? > > - Sybyl 6.6 > - Xfree86 3.3.6 (but 3.3.5 and 4.0 are also ok) > - Nvidia drivers As I told you before I use the drives which come with SuSE 6.2 and 6.4 (6.1 doesn't have them and I haven't tried 6.3). For SuSE 6.2 they are: glxriva-1.0.8 mesa-3.0.48 For SuSE 6.4: glx-000229-20 mesa-3.1.35 ------- Message 11 Date: Tue, 04 Jul 2000 11:15:20 +0200 From: Ezequiel Quintana Morales Message-Id: <3961AB28.727ADB3C@iiq.cartuja.csic.es> Malcolm, > I had the impression that the SuSE drivers only support the > Riva 128/128ZX/TNT/TNT2 graphic boards of NVIDIA, and not the GeForce? SuSE 6.4 supports also GeForce 256, GeForce 256 DDR and Quadro. ------- Message 13 Date: Mon, 03 Jul 2000 16:24:44 +0100 From: "BAYARD Francois" Organisation: COMS Message-Id: <6920061747@mulliken.cpe.fr> I am also very interested by this topic. (I use sybyl) Thanks to forward the answers you will collect. Francois BAYARD Laboratoire COMS - CNRS/CPE UMR 9986 Bat F308 CPE LYON 43 Bd du 11 Novembre 1918 - BP2077 F-69616 VILLEURBANNE CEDEX - FRANCE - ------------------------------- http://www.cpe.fr/lcoms tel (+33) 4 72 43 18 08 fax (+33) 4 72 43 17 95 ------- Message 15 Date: Mon, 03 Jul 2000 17:25:25 -0500 From: John Stone Message-Id: <20000703172524.C26538@geneseo.ks.uiuc.edu> Hi Malcolm, I saw your post on CCL asking about hardware accelerated OpenGL on Linux. We've recently been woking with using the NVidia TNT2, GeForce 256, and Geforce 2 GTS boards with Linux and XFree 4.0, and have had some great results. We are running VMD, our own molecular visualization tool, which you may have already seen or heard of: http://www.ks.uiuc.edu/Research/vmd/ Running VMD, a 700MHz Athlon with a GeForce 2 GTS easily doubles an Onyx2 InfiniteReality2 on rendering speed with a 4,000 atom molecule displaying in VDW representation. The GeForce board isn't quite as fast on lines, and it doesn't have the same hardware antialiasing that our Onyx2 does, but for $300 US, there's no reason to question the utility of having a GeForce... :-) We've also run VMD with some of the other PC video boards that support hardware 3-D on XFree 4.0, but I'd have to say that the only ones worth looking at are the NVidia GeForce 2 GTS boards... For $300 US, you can get a linux box setup that'll roast most of the SGI boxes that are out there, certainly so for the kinds of rendering done in molecular graphics. There are still a number things that the SGI machines are better at, but rendering molecules isn't one of them. The only thing I can think of that the current generation NVidia boards lack, that we want to have in our lab, are stereo sync outputs for use with the CrystalEyes stereo glasses. Fortunately, there are some new low-end LCD stereo glasses coming out for PCs that work with almost any video board, it'll only be a matter of time before we see support for them in XFree/Linux.... Anyway, if you have questions about Linux and XFree 4.x, or on getting/using VMD, let us know, and we can help you out. John Stone - -- Theoretical Biophysics Group Beckman Institute http://www.ks.uiuc.edu/~johns/ University of Illinois Phone: (217) 244-3349 405 N. Mathews Ave FAX: (217) 244-6078 Urbana, IL 61801, USA Unix Is Good For You!!! ------- Message 16 Date: Tue, 04 Jul 2000 10:30:08 +0200 From: Stefan Konietzny Message-Id: <3961A090.B5F8B6C8@uni-duesseldorf.de> Hi Malcolm, if it need not to be Linux, you should also try the Hummingbird Exceed 3D X-Server for Windows. They send you an evaluation copy when you ask for. Of course it's not free, but it worked really fine with my ErazorIII (TNT2) Card. Most of those programs on our origin2000 that where denoted "for SGI Workstation only" worked fine on my PC-Display. The Linux solution is cheaper, but the Windows solution is still much cheaper than to buy several SGI's. Don't blame me for giving the advise to use Windows :-), i'm a Linux user, but the Exceed-X-Server worked better on my PC than the 3.3.6 plus the appropriate extensions from Nvidia. Have fun, Stefan. ------- Message 17 Date: Tue, 04 Jul 2000 13:43:10 +0200 From: Stefan Konietzny Message-Id: <3961CDCE.F3FA2C68@uni-duesseldorf.de> Hi Malcolm, it was Gaussview in OpenGl mode, Cerius2 and InsightII. It's a few weeks ago, so i'm not really sure any more. Further on all those little programs that our sysop put in the toolchest with the comment "for SGI only", that are 3D plots of CPU usage etc. If you like to use Insight without OpenGL in the meantime try (is it InsightII?) to start it with the --axxess option; that forces the program not to use OpenGL but usual X-Graphics. Stefan. ------- End of Forwarded Messages From chemistry-request@server.ccl.net Thu Jul 6 08:14:57 2000 Received: from bacchus.pc1.uni-duesseldorf.de (root@bacchus.pc1.uni-duesseldorf.de [134.99.152.11]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id IAA05946 for ; Thu, 6 Jul 2000 08:14:57 -0400 Received: (from jochen@localhost) by bacchus.pc1.uni-duesseldorf.de (8.9.3/8.8.7) id OAA27941; Thu, 6 Jul 2000 14:13:59 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Message-ID: <14692.30727.176109.78229@bacchus.pc1.uni-duesseldorf.de> Date: Thu, 6 Jul 2000 14:13:59 +0200 (CEST) From: Jochen Kuepper To: Iraj Daizadeh Cc: chemistry@ccl.net Subject: CCL:comutational precision: removing round-off errors. In-Reply-To: <39636FA4.CF35D7C0@genxy.com> References: <39636FA4.CF35D7C0@genxy.com> X-Mailer: VM 6.72 under 21.1 (patch 8) "Bryce Canyon" XEmacs Lucid Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by server.ccl.net id IAA05947 Iraj Daizadeh writes: > If one takes 625 and divides this number by, lets say, 5 on a computer > one may get one of > two numbers that look like: 125.000000000000000001 and > 124.4999999999999999 . > > Is this true? Yes. > and, if so, how does one remove possible rounding errors > produced when > this occurs during lets say integration.... Look up a numerics book on floating point numbers. Some good and free docs are lying around the net ;-) Jochen -- Heinrich-Heine-Universität, Institut für Physikalische Chemie I Universitätsstr. 1, Geb. 26.43 Raum 02.29 40225 Düsseldorf, Germany phone ++49-211-8113681 http://www.Jochen-Kuepper.de fax ++49-211-8115195 From chemistry-request@server.ccl.net Thu Jul 6 09:41:31 2000 Received: from ccl.net (atlantis.ccl.net [192.148.249.4]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id JAA06271 for ; Thu, 6 Jul 2000 09:41:31 -0400 Received: from hscmg02.hou.ucarb.com (hscmg02.hou.ucarb.com [144.68.4.25]) by ccl.net (8.9.3/8.9.3/OSC 2.0) with ESMTP id JAA01763 for ; Thu, 6 Jul 2000 09:41:18 -0400 (EDT) Received: by hscmg02.hou.ucarb.com with Internet Mail Service (5.5.2650.10) id ; Thu, 6 Jul 2000 08:41:55 -0500 Message-ID: <47F7EAA0389AD011840500805FEAB9C6043C920A@sctms01.sct.ucarb.com> From: "Fulton CR (Charles)" To: "'CCL'" Cc: "'Krzysztof Radacki'" Subject: RE: Problem with stability of linux Date: Thu, 6 Jul 2000 08:41:48 -0500 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.10) Content-Type: text/plain What motherboard and processors are you using? Have you tried using the latest stable kernel (2.2.16)? Link to Linux-SMP FAQ - http://www.irisa.fr/prive/dmentre/smp-howto/ , specifically look at what kernel compile options you need to have your .config. (mtrr, etc). There is some great information in the howto/faq. Hardware could be the problem.. (are you overclocking, etc?) -Charlie ------------------------------------------------------------------------ "Open your source and they will come" - various. Charles Fulton Catalyst Skill Center Union Carbide Corporation S. Charleston, WV. phone: (304) 747-3175 email: fultoncr@ucarb.com > ---------- > From: Krzysztof Radacki[SMTP:krys.radacki@ac.rwth-aachen.de] > Sent: Monday, July 03, 2000 1:08 PM > To: chemistry@ccl.net > Subject: CCL:Problem with stability of linux > > > Hi, > I got some problems with the linux-box on which I work. > Real often the maschine hang during gaussian calculations > in SMP-modus with following message (just example): > ==== > Unable to handle kernel NULL pointer dereference at virtual adress > 00000018 > current -> tss.cr3 = 0e0a6000, %cr3 = 0e0a6000 > *pde = 00000000 > Oops = 0000 > CPU = 1 > EIP = 0010: [] > EFLAGS: 00010207 > ax: a7553000 ebx: daad4000 ecx: 00000000 edx: c023c000 > esi: 00000000 edi: 00000000 edp: daad5fbe esp: daad5f9c > de: 0018 ec: 0018 sw: 0019 > Process l502.exe (pid:727, process nr:7, stackpage: daad5f9c > Stack: 00000000....... some lines > CallTrace : [] > Code: 8b 41 18 c1 f8 d1 03 41 1c 89 41 18 > 8b 49 34 81 f9 00 c0 23 > ====== > The questions: why and what should I do again :) > Is it problem of software (I've change the kernel and I can't see any = > difference), > is it SMP-hardware or maybe memory? How can I test it? > > regards > Krzys Radacki > > > -= This is automatically added to each message by mailing script =- > CHEMISTRY@ccl.net -- To Everybody | CHEMISTRY-REQUEST@ccl.net -- To > Admins > MAILSERV@ccl.net -- HELP CHEMISTRY or HELP SEARCH > CHEMISTRY-SEARCH@ccl.net -- archive search | Gopher: gopher.ccl.net > 70 > Ftp: ftp.ccl.net | WWW: http://www.ccl.net/chemistry/ | Jan: > jkl@ccl.net > > > > > From chemistry-request@server.ccl.net Thu Jul 6 22:20:35 2000 Received: from celeborn.otago.ac.nz (celeborn.otago.ac.nz [139.80.64.4]) by server.ccl.net (8.8.7/8.8.7) with ESMTP id WAA09736 for ; Thu, 6 Jul 2000 22:20:34 -0400 From: timr@alkali.otago.ac.nz Received: from alkali.otago.ac.nz (alkali.otago.ac.nz [139.80.16.4]) by celeborn.otago.ac.nz (8.9.3/8.9.3) with ESMTP id OAA08759 for ; Fri, 7 Jul 2000 14:20:28 +1200 (NZST) Message-Id: <200007070220.OAA08759@celeborn.otago.ac.nz> Received: from ALKALI/SpoolDir by alkali.otago.ac.nz (Mercury 1.40); 7 Jul 100 14:20:02 +1200 Received: from SpoolDir by ALKALI (Mercury 1.41); 7 Jul 100 14:19:47 +1200 Organization: University of Otago To: chemistry@ccl.net Date: Fri, 7 Jul 2000 14:19:45 +1200 MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Subject: used g94 manuals X-Confirm-Reading-To: X-pmrqc: 1 Priority: normal X-mailer: Pegasus Mail for Win32 (v3.12b) Hi, I was wondering if anyone had surplus gaussian 94 user manuals that they no longer required, and wanted to sell (or give away). As more groups at our University are becoming interested in computational chemistry, our copy is going walkabout a lot!! The $US 70 price to buy another copy seems rediculous. If anyone can help, please email me. Thanks very much, Tim ------------------------------- Tim Robinson Vikings Group Department of Chemistry University of Otago 03 479 7929 timr@alkali.otago.ac.nz