From daizadeh@indigo.ucdavis.edu Fri Oct 31 17:34:24 1997 Received: from indigo.ucdavis.edu for daizadeh@indigo.ucdavis.edu by www.ccl.net (8.8.3/950822.1) id QAA02007; Fri, 31 Oct 1997 16:40:30 -0500 (EST) Received: by indigo.ucdavis.edu (940816.SGI.8.6.9/940406.SGI.AUTO) id NAA10221; Fri, 31 Oct 1997 13:42:24 -0800 From: "Iraj Daizadeh" Message-Id: <9710311342.ZM10219@indigo.ucdavis.edu> Date: Fri, 31 Oct 1997 13:42:03 -0800 X-Mailer: Z-Mail (3.2.0 26oct94 MediaMail) To: Doug Fox , chemistry@www.ccl.net Subject: Solution. Print out of MOs from two different sets: Cc: fink@indigo.ucdavis.edu, Ahmed Bouferguene , John McKelvey , Stephan Irle Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hello. The overall project was to simply construct an overlap matrix of MO's from different sets. I have summarized replies to the previous message I submitted (utter nonsense!). The problem was to save the MO's from an initial run of Gaussian. Then during a second run read these in and project onto them atomic overlaps and the MOs of the current calculation. Lets take a closer look at the problem: Example: trivial model system H-H+ run 1 H -- H+ -- > coefficients of expansion of MO's print out from G94 run2 H+--H -- > read in MO's from checkpoint file and construct: (Smo)ij = sum_{mu,nu} Cmu,i * (Sao)mu,nu * Cnu,j Now to do this I did the following: First I printed out the coefficients of Mo's from the first run by introducing a link into G94 called l612.F. The line is shown below: Call FileIO (2,-IRwCa,NBSq,V,0) do 200 i=1,NBsq 200 write (IUnitA,2010) V(i) 2010 format (E40.30) Also I print out Overlap matrix over atomic orbitals: NTT = Nbasis*(NBasis+1)/2 Call FileIO(2,-IRwSao,NTT,V,0) do 211 i=1,NTT 211 write (IOvlp,2010) V(i) Now I rerun G94 and print out the IRwCa again for the new molecule. (BE WARNED: Fortran is the transpose of C/C++; that is: F(i,j) = C[j][i];) In my C code I construct the overlap matrix myself. This is a very cumbersome method for doing this but it works!!! Now for the C_{pho,k}'s from the same calculation, of course, the Smo Matrix should be unitary. But for different combinations of MO set this is not the case *yet some symmetry should be there ** it depends on your system of course ** *. This is my no means the best possible methodology I would appreciate any other solutions. Any comments, questions, or suggestions would be appreciated. Iraj. Hi Iraj, Use the option IOP(3/33=1). *********************************************************************** * Lorsque la chance s'envole, * * Lorsque la raison decole, * * Il nous reste la picole. * *********************************************************************** Ahmed Bouferguene --- Hi Iraj, Yes the matrix you get is the matrix S_{\mu , \nu} which is the overlap between atomic orbitals. Now, the overlap between molecular orbitals, if my memory is still good should be zero since this is one of the constraint in the derivation of the Hartree-Fock equation. Ahmed Bouferguene -- Add IOP(3/33=1) to get all of the one electron operator matrices in the log file. Doug Fox -- Hi Iraj, as far as I know there is no such option. However, the overlap between different sets of orbitals during an SCF calculation is evaluated if you set iop(5/14=1) and scf=dm which will activate the conventional (and very slow) direct minimization technique proposed by Seeger and Pople (J. Chem. Phys., 65 (1976), 265-271) in L503. You may want to cut out the responsible part of the code (mainly subroutine BESSRT) and wrap your own code (reading of MO input, overlap matrix etc.) around it. I would very much appreciate if you could post a summary if you get more replies on this topic. Stephan Irle -- G9x does exactly this automatically (default,) I believe. John If the intention is to use this as an initial guess it is done by default. Doug Fox -- Iraj, If the point is to get this into the checkpoint file you can use, %chk=mychk # ... Guess=(Read,Save,Only) which will pick up the previous MO's, project onto the new basis set, save them to the checkpoint file and then only do properties before quiting. Then you could pick them up from the checkpoint file for visualizing etc. -- -- Iraj Daizadeh Department of Chemistry University of California One Shields Ave. Davis, CA 95616-5295 Phone: 916.754.8695 Fax: 916.752.8995 email: daizadeh@indigo.ucdavis.edu