From rrk@iris3.chem.fsu.edu Fri Jan 29 20:05:32 1993 Date: Sat, 30 Jan 93 01:05:32 -0500 From: rrk@iris3.chem.fsu.edu (Randal R. Ketchem) Message-Id: <9301300605.AA28721@iris3.chem.fsu.edu> To: chemistry@ccl.net Subject: Re: computer, general >Hello Netters, > I am using an IBM RS/6000 computer under both UNIX and X-windows. > I am trying to view PostScript file on screen. Is there any >simple way to do it? > Thanks in advance. > > Frank Liu > Argonne National Laboratory > Ghostview -- An X11 user interface for ghostscript. Ghostview is full function user interface for ghostscript 2.4. Brief list of features: - Ghostview parses any known version of Adobe's Document Structuring Conventions. - Page size is automatically determined from the Document Structuring Comments. The user is able to override the values from the comments. - Window size is set to the bounding box for Encapsulated PostScript figures. - Default page size is Letter and can be changed via Xresources or application defaults file to A4 (or any other valid size) for our European friends. - Scrollbars appear when necessary. - Page orientation is automatically determined from the Document Structuring Comments. The user is able to override the values from the comments. - Ability to view at 4 orientations: Portrait, Landscape, Upside-down, and Seascape (for those who rotate landscape the other direction). - Ability to preview in any supported visual. (Can preview in gray-scale or color on a Color monitor.) - Ability to mark pages for printing, or saving. (Good for people that printed a 100 page document and lost page 59 due to a printer jam.) - Can popup zoom windows at printer resolution (1 display dot = 1 printer dot). The Ghostview distribution includes a Ghostview Widget that people are encouraged to use in other programs. Ghostview-1.3 is available via anonymous ftp from: prep.ai.mit.edu:/pub/gnu/ghostview-1.3.tar.Z appenzell.cs.wisc.edu:/pub/ghostview-1.3.tar.Z From friedman@tammy.harvard.edu Sat Jan 30 03:44:03 1993 Date: Sat, 30 Jan 93 08:44:03 -0500 From: friedman@tammy.harvard.edu (Dawn Friedman) Message-Id: <9301301344.AA10445@tammy.harvard.edu> To: chemistry@ccl.net Subject: Gaussian excited states: replies Long post warning; also, most if not all of these were posted to the list, so the redundancy factor is high. I originally posted the following questions. There were several requests for a summary of responses. Thanks, everyone, for your helpfulness and patience (the latter virtue, clearly, being one I was having a touch of trouble with when I posted.) 1) Why doesn't Gaussian allow you to specify wavefunction state and symmetry explicitly, rather than as a series of hints? 2) What package is being used by folks on the list who do excited states and calculate the relative energies in correlation diagrams? Is there a package out there that doesn't have its own agenda about state and symmetry? 3) If there's no answer to 1) and 2), then has anyone else run into this problem? --Dawn An overall summary answer might be: 1) Because there's no single approach to all excited states that can be used with confidence by people using Gnn as a black (or at least shadowy) box, and most of these users want simple input for ground states. 2) For singly excited states, G92 provides CIS. For lowest excited state of a given symmetry, GAMESS, HONDO, and some others can be used (see below for some contacts); but for other excited states, COLUMBUS is recommended, which really does have "no agenda". For studying a correlation diagram or a space of potential energy surfaces for a group of excited states, some CI method is recommended, perhaps a multireference or other non-trivial approach, which can be found as an option in several packages (see below). 3) Yes, there is an answer, and yes, people have. ************************************************************** From fredvc@esvax.dnet.dupont.com Fri Jan 29 13:52:21 1993 Date: Fri, 29 Jan 93 13:44:13 -0500 Message-Id: <9301291844.AA03866@esds01.es.dupont.com> From: fredvc@esvax.dnet.dupont.com To: "friedman@tammy.harvard.edu"@ESDS01.dnet.dupont.com Subject: RE: Gaussian and excited states Most people I know use HONDO/GAMESS or similar program systems when they want to exert tight control over orbital occupancy. GAUSSIANnn was never, to my knowledge, intended for extensive use in this fashion. You might want to contact Mark Gordon (Iowa State) about the availability of GAMESS for your platform. ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ FREDERIC A. VAN-CATLEDGE Scientific Computing Division || Office: (302) 695-1187 Central Research & Development Dept. || FAX: (302) 695-9658 The Du Pont company || P. O. Box 80320 || Internet: fredvc@esvax.dnet.dupont.com Wiilmington DE 19880-0320 || ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ From dave@bogosity.chem.psu.edu Fri Jan 29 15:25:33 1993 Date: Fri, 29 Jan 93 15:19:56 -0500 From: dave@bogosity.chem.psu.edu (David Hartsough) Message-Id: <9301292019.AA08102@bogosity.chem.psu.edu> To: friedman@tammy.harvard.edu Subject: Excited States... You may have seen this one already, but a reference to check is "Toward a Systematic Molecular Orbital Theory for Excited States" Foresman, Head-Gordon, Pople, and Frisch. J. Phys. Chem. 1992, 96, 135-149. Please send me a summary of replies if possible... Dave -------------------------------------------------------------- David Hartsough, dave@bogosity.chem.psu.edu Department of Chemistry Pennsylvania State University University Park PA 16802 (814)863-7591 -------------------------------------------------------------- From caldwell@utdallas.edu Fri Jan 29 16:02:49 1993 From: "Richard A. Caldwell" To: friedman@tammy.harvard.edu Subject: Re: Gaussian and excited states Message-Id: <93Jan29.145811cst.15706@utdallas.edu> Date: Fri, 29 Jan 1993 14:57:26 -0600 Please post the results, or at least let me know what you fond out. GAMESS, which we use, does it; but we only use it for lowest triplet and singlet so I can't speak authoritatively. A note to mike schmidt at Iowa State will give you a definitive answer. Good luck, Dick Caldwell From dudisds@picard.ml.wpafb.af.mil Fri Jan 29 16:13:40 1993 Date: Fri, 29 Jan 93 16:08:50 EST From: dudisds@picard.ml.wpafb.af.mil (Doug S. Dudis) Message-Id: <9301292108.AA01974@Picard.ml.wpafb.af.mil> To: friedman@tammy.harvard.edu Subject: Re: Gaussian and excited states We've run into very similar problems in our own work in excited states. We find GAMESS much friendlier in this respect. Doug Dudis Polymer Branch AF Materials Lab From FOX@CMCHEM.CHEM.CMU.EDU Fri Jan 29 16:19:38 1993 Date: Fri, 29 Jan 1993 16:12:47 EST From: FOX@CMCHEM.CHEM.CMU.EDU Message-Id: <930129161247.21201ec6@CMCHEM.CHEM.CMU.EDU> Subject: RE: Gaussian and excited states To: friedman@tammy.harvard.edu Tammy, Hartree-Fock on excited states is a tricky business, even when they are separated by symmetry. Especially if you are interested in frequencies and the like to add thermochemical corrections to barrier heights. Gaussian's aim has been the investigation of ground states with a minimum of input required of the user, hence we diagonalize full Fock matrices built by applying symmetry operators rather than by building block diagonal Fock matrices and doing number of symmetry operations small diagonalizations. The latter allows great control over what state you have but if you ever go beyond the energy, i.e. the perturbations required to find frequencies, this structure is gone and you can collapse to the lower state without warning. G92 implements the CIS method as a cleaner route to excited states, admittedly only singly excited ones, where we can implement derivatives, frequency analysis and some analysis to make sure you are following the same root. The latter is normally not a problem as all roots, regardless of symmetry, are available so except for regions where states switch order it is easy to follow one particular state. CIS is not perfect. You don't get nearly as much orbital relaxation as you would with a symmetry separated solution to the SCF equations. It is not clear how you add correlation in a reliable fashion and second derivatives are not yet available. The latter two are research topics which I know are actively being pursued. I would recommend the Columbus series of programs if you are serious about playing games with explicitly specifying molecular states. They were distributing this out of Argonne National Labs but I have not kept a real close eye on that program. Doug Fox help@gaussian.com From Louis.Grace@um.cc.umich.edu Fri Jan 29 16:40:45 1993 Date: Fri, 29 Jan 93 16:35:12 EST From: Louis.Grace@um.cc.umich.edu To: friedman@tammy.harvard.edu Message-Id: <19429611@um.cc.umich.edu> Subject: Gaussian and excited states Dear Dawn, I have never calculated excited states, but I have done a few Gaussian calculations of ground states. If you are actually able to map the MO's given the Gaussian output, then, of course, you know the true symmetries of your MO's. If you aren't doing this, but are going by the symmetries which Gaussian assigns, I thought the following might interest you: In the ground state, at least, molcules which have known symmetry ele- ments must be put in so that the z-matrix reflects those symmetry elements. That is, any bond lengths or angles which must be equal to create a mirror plane, axis of rotation, etc., must be specified by a single variable. If one uses separate variables to describe them, the optimized geometry which Gaussian finds for the molecule may not have the proper symmetry. The actual structure and, I would think, the MO coefficients will be very much the same as they would be if the molecule were optimized with the variables set up to give the right symmetry. However, Gaussian will not notice the symmetry, and will label it as being different. This will happen even if bonds or angles which are supposed to be symmetric are different by only a small amount. A good example is benzene. If the z-matrix is built by going around the ring, Gaussian will not give the resulting structure d6h symmetry. If one uses a dummy variable, X, in the center and assigns all C-X distances as being equal, and all C-X-C angles as equal (60deg) then Gaussian will give it d6h symmetry. The structures and MO's arrived at in the two cases should not differ significantly, but in the non-d6h case, one would have to actually plot the MO's on a graph to see what their symmetries really are. I hope this is helpful (and that it has something to do with your pro- blem and is not blatantly obvious). Good luck! Louis Grace Department of Chemistry The University of Michigan Ann Arbor, Michigan 48104 From chemistry-request@ccl.net Fri Jan 29 18:27:22 1993 Date: Fri, 29 Jan 93 15:07:15 CST From: shepard@dirac.tcg.anl.gov (Ron Shepard) Message-Id: <9301292107.AA13616@dirac.tcg.anl.gov> To: chemistry@ccl.net Subject: excited states OK, I'm willing to get into this, but lets keep it friendly... If by excited state you mean the lowest state of a particular symmetry, then you have lots of options. HONDO, GAMESS (US or UK), MELD, ACES, etc. I think GAUSSIAN is the exception rather than the rule in treating spatial symmetry explicitly, and that is a primary requirement for these cases. However, if you mean a general excited state, not necessarily the lowest of its symmetry, then you probably have to use a CI method. If you are interested in reaction paths or global potential energy surfaces, you might want to explore multireference CI or general selected-CI methods. These are the most general, and for surfaces the most accurate, but also probably the most expensive. Of course I will mention COLUMBUS, but also GAMESS, MELD, MOLPRO, the MOLECULE/SWEDEN codes, and MRDCI should be included. There are also other selected-CI methods that should be considered. All of these have advantages and disadvantages. It depends on what kind of questions you are asking as to which is the best choice. -Ron Shepard shepard@tcg.anl.gov From chm6@midway.uchicago.edu Fri Jan 29 22:15:52 1993 Date: Fri, 29 Jan 93 21:11:19 CST From: "charles h martin" Message-Id: <9301300311.AA11702@midway.uchicago.edu> To: friedman@tammy.harvard.edu Subject: excited states Hi. I saw your message on the network and was wondering what system you are studying. I am working with Karl Freed at the University of Chicago developing a package for the effective valence shell many-body perturbation method. This and other many-body methods can give you many states simultaneously, as well as various properties. Of course, this stuff is still in development, but maybe we could trade some info. Thanks Chuck Martin *********************************************************************** From figuei@lutece.rutgers.edu Sat Jan 30 07:15:49 1993 Date: Sat, 30 Jan 93 12:15:49 -0500 From: figuei@lutece.rutgers.edu (Francisco Figueirido) Message-Id: <9301301715.AA17601@lutece.rutgers.edu> To: sliu@anchr1.chm.anl.gov Subject: computer, general You should try ghostscript, which can be ftp'ed from prep.ai.mit.edu. I never compiled it on an RS/6000 (I have none), but it should be doable. It is a great PostScript previewer that work under X Windows and other systems.