From chemistry-request@ccl.net Wed Aug 19 06:16:25 1992 Date: 18 Aug 1992 16:41:02 -0500 (CDT) From: AHOLDER@VAX1.UMKC.EDU Subject: Semiempirical stuff To: CHEMISTRY@ccl.net Status: RO Hello netters. There have been some recent questions on the net about semiempirical methods. First Joe Leonard at Wavefuntion in Irvine, CA asked about the performance of semiempirical methods as regards the prediction of dipole moments. I can only second what others have said as to the plethora of data available in the parameterization papers by Dewar et.al. Joe further asked about "atom by atom" reparameterization of formerly completed parameters sets. This is an extremely dangerous concept. One of the real advantages of SE methods is that they have become standard over the years. "AM1" means "AM1". Everyone knows the pitfalls and problems with it (or will discover them). It is a known quantity and performs predictably. We should not disturb this stabil- ity. Re-doing old parameter sets is not really possible. (I know, I know, Dewar re-did Si and S for MNDO.) Elements parameterized AFTER a particular element is complete depend on that element's values. For example, some of the results for AM1 fluorine calculations are a bit irregular. It's not possible to go back and "re-do" that parameter set now, because all of the many AM1 elements parameterized since then depended on thise fluorine values! To change the old values would be to invalidate later parameter sets, as well as all calculations using the F parameters. As to the case of "fixing" published parameter sets (e.g. Rzepa and N), this is also a dangerous practice. The parameters in semiemp- irical methods are very sensitive and highly interdependent, and a method has been developed to optimized them simultaneously, They should NOT be altered by the end user in a manual fashion. This is especially true of the Gaussian functions in AM1 and its deriv- ative methods. These parameters can have a very strong effect and should only be applied to correct specific difficulties after the base values of the "chemical" parameters have been settled upon. Gaussians are best viewed as crude patches that are added only when needed. One must remember that semiempirical methods are a sub-class of quantum mechanics, a discipline that attempts to reproduce chemical phenomena by mathematically describing the activity of the mediators of chemistry, electrons. MM models use potential functions to approximate the effects of the activity of electrons on a macroscopic level. For this reason, a parameterization philosophy that works for MM is not viable in SE. Dr. Stewart at Fujitsu has made some comments about Michael Dewar's next- generation SAM1 method and has even presented some results. I must stress that these results are both preliminary and incomplete and were not published with the knowledge or approval of Michael Dewar. The parameters for these elements have only just reached final form and our results are somewhat different. A paper is in preparation on SAM1 at this time. Those interested and attending the ACS meeting in Washington will have the opportunity to hear more about SAM1, a new step in the devlopment of semiempirical methodology. Andy P.S. (Sorry for the sermonizing.) =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= DR. ANDREW HOLDER Assistant Professor of Computational/Organic Chemistry Department of Chemistry || BITNET Addr: AHOLDER@UMKCVAX1 University of Missouri - Kansas City || Internet Addr: aholder@vax1.umkc.edu Spencer Chemistry, Room 315 || Phone Number: (816) 235-2293 Kansas City, Missouri 64110 || FAX Number: (816) 235-1717 =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= --- Administrivia: This message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chemistry-request@ccl.net Wed Aug 19 06:32:25 1992 Date: 18-AUG-1992 22:35:08.01 From: WSONNEN@EAGLE.WESLEYAN.EDU Subject: CHARMM CRYSTAL BUILDER To: chemistry@ccl.net Status: RO I would like to build a crystal in CHARMM and write out the coordinates of all images explicitly. I am using CHARMm21.3. As a test i have tried the following script; in the test script, i attempt to write-out coordinates for three images (according to the CHARMM notes, the inverses will not be written). The result i obtain, however, is two coordinate sets, the first of which is the origional coordinates that i read in as my assymetric unit. I don't know what the second set of coordinates are (although this wouldn't be too difficult to figure out--just keep one image and its inverse and see which one matches). My questions are: (1) how can i get CHARMM to write out all the coordinates for the images that i generate? (2) Can i generate a given image without explicitly specifing its inverse? Any responses should probably be sent over the network for other CHARMM users. Thank you, Wayne Sonnen wsonnen@eagle.wesleyan.edu The following is the script (after psf generation and coords are read): CRYSTAL DEFINE ORTHORHOMBI 24.8700 40.3800 66.2000 90.0000 90.0000 90.0000 CRYSTAL ZERO 24.8700 40.3800 66.2000 90.0000 90.0000 90.0000 CRYSTAL STRESS 0.000 0.000 0.000 0.000 0.000 0.000 CRYSTAL BUILD NOPER 3 CUTOFF 99.0 (1/2-X,-Y,1/2+Z) (-X,1/2+Y,1/2-Z) (1/2+X,1/2-Y,-Z) READ CRYSTAL CARD * DEFINITION OF CRYSTAL FORxxx.msf * SYMMETRY (X,Y,Z) (1/2-X,-Y,1/2+Z) (-X,1/2+Y,1/2-Z) (1/2+X,1/2-Y,-Z) END IMAGES 2 0 0 0 3 0 0 0 4 0 0 0 2 -1 0 0 3 0 -1 0 4 0 0 -1 END UPDATE INBFRQ 50 IMGFRQ 50 IHBFRQ 50 CUTIM 99.0 PRINT COOR IMAGE OPEN WRITE UNIT 88 CARD NAME TEST_CRYSTAL.CRD WRITE COOR IMAGE CARD UNIT 88 * STOP --- Administrivia: This message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From jkl@ccl.net Wed Aug 19 10:02:29 1992 Date: Wed, 19 Aug 1992 10:02:25 -0400 From: jkl@ccl.net To: chemistry@ccl.net Subject: Reminder from the coordinator Status: R Dear Comp.Chem.List subscribers, Please adhere to these rules and we will all be happier... Please note that if you make a mistake it is multiplied over a thousand times... 1. READ THE HELP FILE before you post. You can get it from anon ftp at kekule@ccl.net in pub/chemistry or by sending a line: send help from chemistry to OSCPOST@ccl.net or OSCPOST@OHSTPY.BITNET 2. NEVER REPOST unless I ask you. If you sent it to chemistry@ccl.net it will be 99.9% distributed. If you are not sure, ask me (jkl@ccl.net) if I got it. In most cases you do not see your message coming back to you from the list since the mailer supresses messages with the same From: and To: fields. If you want to get your message, include Cc: to yourself. You are doing disservice to yourself when reposting, since people ask themselves why you did it :-)... Note that it may take a day or more to distribute your message depending on the volume of waiting messages. 3. ALWAYS USE SUBJECT: Not everybody is interested in everything. With the large volume on the list, it is an imperative. At some point I might prevent messages without Subject: field to appear on the list. 4. If you pass the information about the list to a friend, give him/her a copy of the help file or an information about the OSCPOST. Do not give out the chemistry@ccl.net address, so we do not have all this junk mail on the list. 5. If you messed up something, please DO NOT APOLOGIZE TO THE LIST. You are only increasing a volume of unnecessary mail Your coordinator (?) Jan Labanowski Ohio Supercomputer Center jkl@ccl.net From JADAMS@uoft02.utoledo.edu Wed Aug 19 10:12:26 1992 Date: 19 Aug 1992 10:12:06 -0400 (EDT) From: "Q. E. D" Subject: casscf calculations To: chemistry@ccl.net Status: R In Gaussian 90 I know that integrals will be recomputed as needed if one specifies SCF=DIRECT. Is there any similar keyword for CASSCF? This would be very helpful. I am running cas(10,10) [more correctly: trying to run] and my disks are not large enough to handle it. Jamie Adams Douglas A. Smith Research Group University of Toledo e-mail:jadams@uoft02.utoledo.edu From chemistry-request@ccl.net Wed Aug 19 10:51:57 1992 Date: Wed, 19 Aug 1992 11:29:59 +0200 From: oles@ulrik.uio.no To: C1790@slvaxa.umsl.edu Subject: MODELING POLYMERS Status: R How come I get so many copies of each message? Looks like the mail exploder has a fit of convulsions. Ole Swang --- Administrivia: This message is automatically appended by the mail exploder. CHEMISTRY@ccl.net --- everybody; CHEMISTRY-REQUEST@ccl.net --- coordinator only OSCPOST@ccl.net : send something from chemistry; FTP: www.ccl.net --- From chapron@drac.CENG.cea.fr Wed Aug 19 12:14:40 1992 Date: 19 Aug 92 18:10:56-2300 From: Yves chapron To: CHEMISTRY@ccl.net Subject: Charmm-Xplor files conversion code Status: R Yves CHAPRON Biophysique moleculaire & cellulaire Grenoble 19 aug 92 C.E.N.G. 85 X 38041 GRENOBLE cedex FRANCE - Ph (33)76884212 Fx (33)76885487 E.M. chapron@drac.ceng.cea.fr Dear Netters, Is there a code available that does Charmm-Xplor files conversions ? (topo files ...). Thanks Yves From jle@world.std.com Wed Aug 19 12:47:16 1992 Date: Wed, 19 Aug 92 12:47:05 -0400 From: jle@world.std.com (Joe M Leonard) To: chemistry@ccl.net Subject: Re: Semiempirical cal*n Status: R Continuing the discussion on semi-empirical methodology... 1) Am I correct in assuming that the consensus is to not change any parameters after they are published - replacing sets in toto as new methods/values are developed (MNDO -> AM1/PM3 -> ?)? If this is the case, how are situations indicative of inaccurate parameterization dealt with other than "they're incorrect, but will be fixed in the new program in 0-3 years"? 2) The Additional Gaussians added to AM1 seem to alter the semiemp wavefunction so as to make it less ab initio-like (Besler & Merz, for example) - is there an understanding of this behavior? Is the solution as simple as refitting ab initio basis sets to semiemp's valence-only wavefunctions? 3) Following up on (1), I guess all developers better get it right the first time... :-) 4) I'm not sure I'd call the parameterization approaches for MM and semiemp all that different - folks are interested in the same quantities: geometries, heats of formations/energies, spectra, etc. Granted, the physics is rather different, and hopefully the more general methods will turn out to be more flexible/expandable, but cynically, they're all a data-fitting problem... 5) I don't think Jimmie's with Fujitsu, although I think he has something to do with them (care to explain?).. 6) Is there extra Gaussians in mndo-d, or is it closer to a valence-only approach (w/o corrections)? Joe Leonard jle@world.std.com From m10!frisch@uunet.UU.NET Wed Aug 19 21:12:06 1992 Date: Wed, 19 Aug 92 16:38:24 EDT From: m10!frisch@uunet.UU.NET (Michael Frisch) Subject: Re: casscf calculations To: chemistry@ccl.net Status: R In Gaussian 90 I know that integrals will be recomputed as needed if one specifies SCF=DIRECT. Is there any similar keyword for CASSCF? This would be very helpful. I am running cas(10,10) [more correctly: trying to run] and my disks are not large enough to handle it. Jamie Adams Douglas A. Smith Research Group University of Toledo e-mail:jadams@uoft02.utoledo.edu Direct MCSCF became available for the first time in Gaussian 92. Mike Frisch -------