From jkl@ccl.net Sun Oct 1 12:21:32 1995 Received: from bedrock.ccl.net for jkl@ccl.net by www.ccl.net (8.6.10/950822.1) id MAA09434; Sun, 1 Oct 1995 12:16:15 -0400 Received: from krakow.ccl.net for jkl@ccl.net by bedrock.ccl.net (8.6.10/950822.1) id MAA13616; Sun, 1 Oct 1995 12:16:12 -0400 From: Jan Labanowski Received: for jkl@ccl.net by krakow.ccl.net (8.6.10/920428.1525) id MAA11882; Sun, 1 Oct 1995 12:16:11 -0400 Date: Sun, 1 Oct 1995 12:16:11 -0400 Message-Id: <199510011616.MAA11882@krakow.ccl.net> To: chemistry@www.ccl.net Subject: Spin contamination in DFT - FORTRAN code Cc: jkl@ccl.net Once more about spin contamination in DFT. I hope that my FORTRAN is better than my broken English. I wrote a routine to compute it. Can you look at it and see if it is a nonsense or a pearl {:-)}? Maybe it is easier for you to look at it than at my previous message (I actually messed up in the previous message, since I did not square the SAB[i,j] at the very end). Sorry for my oldfashioned FORTRAN, but I learned it before optimizing compilers on a 64k machine and old habits die slow (but I still have a few years till retirement, I simply started late... {:-)}). Jan Labanowski jkl@ccl.net ---------------- Variable names as in my previous message ------- SUBROUTINE SPNCNT(S2EXCT, S2EXPT) IMPLICIT REAL*8(A-H,O-Z) INCLUDE "param.h" COMMON/BASIS/HMAT(NFOCK),TVECTR(NAUXFN),NDTT,NCNTRT,NNNS,NNNP,NNND COMMON/ORBEN/ELGALP(NCONTS),ELGBET(NCONTS),DOCA(NCONTS), & DOCB(NCONTS),NLA,NLB COMMON/MATRX/SINV(NCONTS,NCONTS),FOCK1(NFOCK),FOCK2(NFOCK), & FOCK3(NFOCK) COMMON/MODAT/AMAT(NFOCK),BMAT(NFOCK), & CALPHA(NCONTS,NCONTS),CBETA(NCONTS,NCONTS) COMMON/INFOA/CENT(3,NATOMS),CHARGE(NATOMS),NATOMC(NATOMS),NCENTR, & NALPHA,NBETA,NSTATE,NHOLEA(NCONTS),NHOLEB(NCONTS) DIMENSION WORK(NCONTS, NCONTS), WVEC(NCONTS) C -- Read in overlap matrix REWIND 43 DO 100 I=1,NCNTRT 100 CALL MYREAD(43,SINV(1,I),NCNTRT) C -- "exact spin square" S2EXCT = 0.25D0*(NALPHA - NBETA)*(NALPHA - NBETA + 2.0D0) C -- Calculate MO overlap matrix SAB = CALPHA' * SINV * CBETA C -- CALPHA' is a transpose of CALPHA C -- Make a matrix WORK(1:NCNTRT,1:NLB) = SINV * CBETA DO 400 K1 = 1, NCNTRT DO 150 K2 = 1, NCNTRT C -- since SINV(K1,K2) = SINV(K2,K1) and FORTRAN orders matrices along C -- columns, it is better for vectorization to do SINV(K2,K1) WVEC(K2) = SINV(K1,K2) 150 CONTINUE DO 300 J = 1, NLB SUM = 0.0 DO 200 K2 = 1, NCNTRT SUM = SUM + CBETA(K2,J)*WVEC(K2) 200 CONTINUE WORK(K1,J) = SUM 300 CONTINUE 400 CONTINUE C -- Calculate SAB = CALPHA' * WORK (save it in SINV) DO 700 I = 1, NLA DO 600 J = 1, NLB SUM = 0.0 DO 500 K1 = 1, NCNTRT SUM = SUM + CALPHA(K1,I)*WORK(K1,J) 500 CONTINUE SINV(I,J) = SUM 600 CONTINUE 700 CONTINUE C -- Calculate expectation value of square of total spin S2EXPT = S2EXCT + NBETA DO 900 I = 1, NLA DO 800 J = 1, NLB X = SINV(I,J) S2EXPT = S2EXPT - DOCA(I)* DOCB(J) * X*X 800 CONTINUE 900 CONTINUE C -- Restore overlap matrix REWIND 43 DO 1000 I=1,NCNTRT 1000 CALL MYREAD(43,SINV(1,I),NCNTRT) RETURN END From !rgab@trpntech.com Sun Oct 1 14:21:34 1995 Received: from netcomsv.netcom.com for !rgab@trpntech.com by www.ccl.net (8.6.10/950822.1) id OAA10268; Sun, 1 Oct 1995 14:19:18 -0400 Received: from .UUCP by netcomsv.netcom.com with UUCP (8.6.12/SMI-4.1) id LAA01898; Sun, 1 Oct 1995 11:17:41 -0700 Message-Id: <199510011817.LAA01898@netcomsv.netcom.com> Date: 1 Oct 1995 10:49:53 -0800 From: "Richard Bone" Subject: The bond order question To: "CCL" Subject: The bond order question From how I understood the original posting on bond-orders, (and from my experience of working in a pharmaceutical company), I assumed that the questioner did not have *wavefunctions* for his molecules. Consequently, I couldn't see that methods of Bader, Mayer, etc., had any relevance. From the original posting, I understood that the questioner had his molecules stored with atom-types and (presumably) some connectivity information and from those pieces of information wished to ascribe (*integer* - by use of the term 'kekule') bond orders (single, double, triple) to the bonds. I do not know of a program which explicitly does this, but I suspect that one has to write subroutines for the task. One would need a table of 'common' valencies for the atoms somewhere in there. This gives rise to certain problems for hypervalent elements, e.g., S, P, and for those which can commonly exist in more than one oxidation state, e.g., N. Richard Bone __________________________________________________________ Richard G. A. Bone, PhD. Computational Chemist Terrapin Technologies, Inc. 750-H Gateway Blvd. South San Francisco CA 94080-7020 USA Tel. +1 (415) 244 9303 FAX +1 (415) 244 9388 E-mail rgab@trpntech.com