From CHERM@FRCPN11.IN2P3.FR Mon Dec 12 04:19:18 1994 Received: from frcpn11.in2p3.fr for CHERM@FRCPN11.IN2P3.FR by www.ccl.net (8.6.9/930601.1506) id DAA05217; Mon, 12 Dec 1994 03:29:45 -0500 Message-Id: <199412120829.DAA05217@www.ccl.net> Received: from FRCPN11.IN2P3.FR by frcpn11.in2p3.fr (IBM VM SMTP V2R2) with BSMTP id 1961; Mon, 12 Dec 94 09:28:52 MET Received: from FRCPN11.in2p3.fr (NJE origin CHERM@FRCPN11) by FRCPN11.IN2P3.FR (LMail V1.2a/1.8a) with BSMTP id 2198; Mon, 12 Dec 1994 09:28:53 +0100 Date: Mon, 12 Dec 94 09:28:52 MET From: Henry Chermette Subject: DFT_95 To: chemistry@ccl.net X-Acknowledge-To: 6th INTERNATIONAL CONFERENCE ON THE APPLICATIONS OF THE DENSITY FUNCTIONAL THEORY IN CHEMISTRY AND PHYSICS AUGUST 29th - SEPTEMBER 1st 1995 PARIS (FRANCE) FIRST CIRCULAR The 6th edition of the International The SECOND CIRCULAR, containing Conference on the Applications of informations about registration the Density Functional Theory in fees, possible financial support Chemistry and Physics will be held in and accomodation, will be mailed Paris, August 29th - September 1st 1995. at the end of February 1995 to The meeting aims to bring together all who have returned the attached scientists interested in the basic pre-registration form before aspects of the theory, methodological January 15th 1995. developments and applications in The Scientific Program will include chemistry, biochemistry and physics. plenary lectures, oral contributions and a poster session. To receive the next mailing, return the questionnaire by mail or fax to: Dr. A. Goursot Ecole de Chimie 8, rue de l'Ecole Normale 34053 Montpellier, Cedex France fax: (33)-67 14 43 49 e-mail: dft95@frmop22.cnusc.fr The following speakers have already Scientific Committee agreed to present an invited paper: A. Bencini, Univ. Firenze (I) A. St-Amant (Canada) C. Daul, Univ. Fribourg (CH) E. J. Baerends (The Netherlands) W. Kohn, Univ. California (USA) R. Car (Switzerland) J. A. Pople, Northwestern Univ. (USA) D. Case (U.S.A.) N. Russo, Univ. Calabria (I) M. Cohen (U.S.A.) V. Smith, Queen's Univ. (CA) P. Fantucci (Italy) T. Ziegler, Univ. Calgary (CA) B. Johnson (U.S.A) M. Levy (U.S.A.) Organizing Commette R. Parr (U.S.A) M. Allavena, Univ. Paris VI J. Perdew (U.S.A.) C. Bureau, CEA, Paris D. Salahub (Canada) H. Chermette, Univ. Lyon I K. Schwartz (Austria) A. Goursot, ENSC, Montpellier C. Umrigar (U.S.A) J. Langlet, Univ. Paris VI J. Weber (Switzerland) C. Mijoule, Univ. Paris VI W. Yang (U.S.A.) A. Savin, Univ. Paris VI Complementary information can also be asked to: H. Chermette: cherm@frcpn11.in2p3.fr A. Goursot: dft95@frmop22.cnusc.fr C. Mijoule: cm@dim.jussieu.fr _____________________________________________________ DFT 95 pre-registration form Paris, August 29 - September 1, 1995 Name Affiliation Address Fax E-mail Probability of attending 90% 50% 10% Number of accompanying persons From kdb@oddjob.uchicago.edu Mon Dec 12 05:19:18 1994 Received: from oddjob.uchicago.edu for kdb@oddjob.uchicago.edu by www.ccl.net (8.6.9/930601.1506) id EAA05724; Mon, 12 Dec 1994 04:29:59 -0500 Received: by oddjob.uchicago.edu Mon, 12 Dec 94 03:29:58 CST Date: Mon, 12 Dec 94 03:29:58 CST From: "Keith Ball" Message-Id: <9412120929.AA17111@oddjob.uchicago.edu> To: chemistry@ccl.net Subject: Fast Diag. Routines: RESPONSES Here is a summary of the response to my mesage about possible fast diagonalization methods, posted to the Computational Chemistry List and sci.math, sci.numerical-analysis, sci.physics, and sci.chem. Concerning 'fast' algorithms (ie faster than O(n^3) ) there don't seem to be any for non-sparse matrices. The routines most recommended are tred2 (Householder reduction of a symmetric matrix to a tridiagonal matrix) followed by tqli (finding eigenvalues, and eigenvectors if desired, of the tridiagonal matrix). These routines are given in Press et al, _Numerical Recipes_, available in FORTRAN, C, and Pascal versions, to my knowledge. (2nd ed.) One can purchase the programs on disk, and they may already be on whatever machine you currently use (but I would check to see if they are 1st ed. and if there have been any problems with the 1st ed programs; i have no idea. Probably could just type in the necessary changes from the 2nd ed. book). I especially recommend that one read Chapter 11 on Eigensystems, if they have not already done so. The eispack library has versions of these routines, and lapack also has versions of these routines. For access to these libraries, the full bibliography for Press, and other quips and references, see below. There are also one or two routines offered to me, which I have included. I have not tried any of these, so I would check with the person who sent the code (name and e-mail are provided). Enjoy!! Keith Ball kdb@cloister.uchicago.edu ------------------------------------------------------------ From: "M. J. Saltzman" In sci.math you write: > I am wondering if anyone knows of any references to fast algorithms >for diagonalization (i.e. finding the eigenvalues and eigenvectors) >of a symmetric matrix. > The problem I am using this for consists of finding stationary points >(specifically saddle points) on the 3*N-dimensional Cartesian potential >surface for a system of N particles interacting via pairwise interactions. >The matrix in question is the Hessian second-derivative matrix. >Any references or suggestions would be greatly appreciated! >Please send your responses directly by e-mail. Check out LAPACK at Netlib. Send mail to netlib@ornl.gov with the message "send index". -- Matthew Saltzman Clemson University Math Sciences mjs@clemson.edu ---------------------------------------- From: David Heisterberg For the general case I don't know of anything that beats the old workhorse of Householder reduction followed by QL with implicit shifts, as implemented in Eispack (and now Lapack). This is assuming you want all eigenvalues and vectors. If you are repeatedly solving for eigenvalues and vectors of matrices that change only a small amount at each step, then maybe an inverse iteration technique would work. -- David J. Heisterberg (djh@ccl.net) Gee, it's so beautiful, I gotta The Ohio Supercomputer Center give somebody a sock in the jaw. Columbus, Ohio -- Little Skippy (Percy Crosby) ---------------------------------------- From: sling@euclid.chem.washington.edu (Song Ling) Hi Keith. I suggest that you look into existing program packages, one is called IMSL and another NAG, I think they are commercial. If you want to try it on your own, there are some fortran codes in a book called "Numerical Recipes" by Press et al, it is a very popular book, and I suggest that you find it at the bookstore of U of Chicago. Best, Song Ling ---------------------------------------- From: hinsenk@ERE.UMontreal.CA (Hinsen Konrad) All linear algebra libraries provide routines for that, and unless your matrix has some exotic properties and/or you are looking for only a small subset of eigenvalues, there aren't any faster routines than the standard ones. My personal preference are the routines from the LAPACK library, which are free and in my experience more stable and faster than many others. You can get them from many sources, e.g. from AT&T's netlib (ftp to netlib.att.com:netlib/lapack). ------------------------------------------------------------------------------- Konrad Hinsen | E-Mail: hinsenk@ere.umontreal.ca Departement de Chimie | Tel.: +1-514-343-6111 ext. 3953 Universite de Montreal | Fax: +1-514-343-7586 C.P. 6128, succ. A | Deutsch/Esperanto/English/Nederlands/ Montreal (QC) H3C 3J7 | Francais (phase experimentale) ------------------------------------------------------------------------------- ---------------------------------------- From: mike@mycenae.CChem.Berkeley.EDU (Mike L. Greenfield) For calculating eigenvalues and eigenvectors, I recommend the combination of tred2 and tql2 from the EISPACK library. You can obtain the Fortran source code by anonymous ftp to netlib.att.com. There are probably also LAPACK routines to do this; I don't know their names. I think these two routines are also discussed in the eigenvalues section of Numerical Recipes. I assume you've looked at Jon Baker's paper (and references therein) on finding transition states? He discusses a Cerjan-Miller type algorithm in detail: @Article{Baker860, author = "Jon Baker", title = "An Algorithm for the Location of Transition States", journal = "J. Comput. Chem.", year = 1986, volume = 7, pages = "385--395" } Good luck with your calculations. Mike Greenfield Dept. of Chemical Engineering UC Berkeley mike@mycenae.cchem.berkeley.edu ---------------------------------------- From: cooksj@ttown.apci.com (Stephen J. Cook) Keith, I'm not sure how fast they are, but have you looked at the netlib archives? These can be found at either research.att.com (or) netlib.att.com One of these names is obsolete, but I know that one of them works. These machines hold linpack,lapack,eispack and other numerical algorithm suites that may be helpful to you. They are all available in Fortran. Good luck. Steve Cook -- ********************************************************************* * Steve Cook cooksj@ttown.apci.com * * Air Products and Chemicals, Inc. Tel. (610) 481-2135 * * 7201 Hamilton Blvd. FAX (610) 481-2446 * * Allentown, PA 18195 * * USA * ********************************************************************* * Emacs - the choice of a GNU generation * ********************************************************************* * Disclaimer: The opinions expressed here are those of the author. * * Any resemblance between my opinions and those of Air * * Products is purely coincidental... * ********************************************************************* ---------------------------------------- From: bittner@hpcf.cc.utexas.edu (bittner) I use the LAPACK equlvalents to the EISPACK tred2 and tql2 routines. All of which I know are on rainbow.uchicago. BTW, will their be a 2nd Keith Ball symposium for mathematical physics? I forgot the subject for the first symposium...but it was great idea. ---------------------------------------------------------------------- Eric R. Bittner phone: (512)-471-1092 Dept. of Chemistry fax: (512)-471-8696 Univ. of Texas (formerly U of C) ----------------------------------------------------------------------- Wenn is das Nunstuck git und Slottermeyer? Ja! ...Beiherhund das Oder die Flipperwaldt gersput! ---------------------------------------- From: Ryszard Czerminski X 217 Hi Keith, I am not quite sure if this one will be fast enough for you, but it served me well over past ~10 years. Ryszard Czerminski SUBROUTINE HOUSE (Z,NZ,N,D,E,ERROR) IMPLICIT REAL*8 (A-H,O-Z) C C DIAGONALIZES REAL, SYMMETRIC MATRIX USING HOUSEHOLDER METHOD. C C INPUT: C Z - Matrix to be diagonalized C NZ - first dimension in calling program C N - order of Z C C OUTPUT: C D - eigenvalues in decreasing order C Z - genvectors C ERROR=0 - normal solution C ERROR=J - failure in J eigenvalue after 30 iterations C C MACHINE DEPENDENT PARAMETERS: C MACHEP - relative machine precision C SMALL - smallest floating point number C INTEGER ERROR REAL*8 MACHEP C DIMENSION D(N),E(N),Z(NZ,N) real*8 D(N),E(N),Z(NZ,N) real*8 b,c,f,g,h,p,r,s,small,tol c DATA SMALL/5.4D-79/ IBM 370 c MACHEP=2.D0**(-56) IBM 370 C DATA SMALL/2.225D-308/ titan, IBM PC real*8 ms-fortran 4.0 C = 2**(-1022) C MACHEP=2.D0**(-52) titan, IBM pc real*8 ms-fortran 4.0 C DATA SMALL/0.495D-323/ sun : 2**(-1074) C DATA SMALL/0.495D-323/ sun : 2**(-1074) DATA SMALL/0.495D-323/ C DSQRT(X)=DDSQRT(X) C DSIGN(X,Y)=DDSIGN(X,Y) C DABS(X)=DDABS(X) MACHEP=2.D0**(-52) C TOL=SMALL/MACHEP IF (N .EQ. 1) GO TO 320 DO 300 II = 2, N I = N + 2 - II L = I - 2 F = Z(I-1,I) G = 0.0E0 IF (L .LT. 1) GO TO 140 DO 120 K = 1, L 120 G = G + Z(K,I) * Z(K,I) 140 H = G + F * F IF (G .GT. TOL) GO TO 160 E(I) = F H = 0.0E0 GO TO 280 160 L = L + 1 G = -DSIGN(DSQRT(H),F) E(I) = G H = H - F * G Z(I-1,I) = F - G F = 0.0E0 DO 240 J = 1, L Z(I,J) = Z(J,I) Z(J,I) = Z(J,I) / H G = 0.0E0 DO 180 K = 1, J 180 G = G + Z(K,J) * Z(I,K) JP1 = J + 1 IF (L .LT. JP1) GO TO 220 DO 200 K = JP1, L 200 G = G + Z(J,K) * Z(K,I) 220 E(J) = G / H F = F + G * Z(J,I) 240 CONTINUE HH = F / (H + H) DO 260 J = 1, L F = Z(I,J) G = E(J) - HH * F E(J) = G DO 260 K = 1, J Z(K,J) = Z(K,J) - F * E(K) - G * Z(I,K) 260 CONTINUE 280 D(I) = H 300 CONTINUE 320 D(1) = 0.0E0 E(1) = 0.0E0 DO 500 I = 1, N L = I - 1 IF (D(I) .EQ. 0.0E0) GO TO 380 DO 360 J = 1, L G = 0.0E0 DO 340 K = 1, L 340 G = G + Z(I,K) * Z(K,J) DO 360 K = 1, L Z(K,J) = Z(K,J) - G * Z(K,I) 360 CONTINUE 380 D(I) = Z(I,I) Z(I,I) = 1.0E0 IF (L .LT. 1) GO TO 500 DO 400 J = 1, L Z(I,J) = 0.0E0 Z(J,I) = 0.0E0 400 CONTINUE 500 CONTINUE ERROR = 0 IF (N .EQ. 1) GO TO 1500 DO 1100 I = 2, N 1100 E(I-1) = E(I) F = 0.0E0 B = 0.0E0 E(N) = 0.0E0 DO 1240 L = 1, N J = 0 H = MACHEP * (DABS(D(L)) + DABS(E(L))) IF (B .LT. H) B = H DO 1110 M = L, N IF (DABS(E(M)) .LE. B) GO TO 1120 1110 CONTINUE 1120 IF (M .EQ. L) GO TO 1220 1130 IF (J .EQ. 30) GO TO 1400 J = J + 1 P = (D(L+1) - D(L)) / (2.0E0 * E(L)) R = DSQRT(P * P + 1.0E0) H = D(L) - E(L) / (P + DSIGN(R,P)) DO 1140 I = L, N 1140 D(I) = D(I) - H F = F + H P = D(M) C = 1.0E0 S = 0.0E0 MML = M - L DO 1200 II = 1, MML I = M - II G = C * E(I) H = C * P IF (DABS(P) .LT. DABS(E(I))) GO TO 1150 C = E(I) / P R = DSQRT(C * C + 1.0E0) E(I+1) = S * P * R S = C / R C = 1.0E0 / R GO TO 1160 1150 C = P / E(I) R = DSQRT(C * C + 1.0E0) E(I+1) = S * E(I) * R S = 1.0E0 / R C = C / R 1160 P = C * D(I) - S * G D(I+1) = H + S * (C * G + S * D(I)) DO 1180 K = 1, N H = Z(K,I+1) Z(K,I+1) = S * Z(K,I) + C * H Z(K,I) = C * Z(K,I) - S * H 1180 CONTINUE 1200 CONTINUE E(L) = S * P D(L) = C * P IF (DABS(E(L)) .GT. B) GO TO 1130 1220 D(L) = D(L) + F 1240 CONTINUE NM1 = N - 1 DO 1300 I = 1, NM1 K = I P = D(I) IP1 = I + 1 DO 1260 J = IP1, N IF (D(J) .LE. P) GO TO 1260 K = J P = D(J) 1260 CONTINUE IF (K .EQ. I) GO TO 1300 D(K) = D(I) D(I) = P DO 1280 J = 1, N P = Z(J,I) Z(J,I) = Z(J,K) Z(J,K) = P 1280 CONTINUE 1300 CONTINUE GO TO 1500 1400 ERROR = L 1500 RETURN END c------------------------------------------------ subroutine horder(a,e,n,ch) character*1 ch real*8 a(n,n),e(n),x c c ch = 'i' - increasing order of eigenvalues c ch = 'd' - decreasing order of eigenvalues c do 30 i=1,n-1 do 20 j=i+1,n if(e(i).gt.e(j)) then if(ch.eq.'i') then do 10 k=1,n x = a(k,i) a(k,i) = a(k,j) a(k,j) = x x = e(i) e(i) = e(j) e(j) = x 10 continue endif endif 20 continue 30 continue return end ---------------------------------------- From: dickson@zinc.chem.ucalgary.ca (Ross M. Dickson) Have you looked into EISPACK or its successor, LAPACK? They're public-domain FORTRAN libraries for eigenproblems. Although I'm not intimately familiar with the algorithm available therein for real symmetric matrices, I know there is code for that special case. The developers tried to choose the most efficient yet reliable algorithms possible for each matrix type covered. They can be obtained as a whole or in pieces through NETLIB: For info, send a message consisting solely of the word 'help' to netlib@ornl.gov. Ross Dickson, dickson@zinc.chem.ucalgary.ca Department of Chemistry, The University of Calgary, Alberta, Canada ---------------------------------------- From: ucacsam The Householder method is among the best you can find. See texts by J H Wilkinson, L Fox and many others for descriptions. Basically the method calculates orthogonal matrices that are used to transform your original matrix to a (symmetric) tridiagonal form, which for an n x n matrix can be done in about 2/3 of the cost of one complete matrix multiplication (of two n x n matrices - that's very clever!!). From then on the process uses Laguerre's method for finding the eigenvalues and you can the vectors as well. All rapidly convergent and numericaly stable. A simpler method, which takes rather longer is known as Jacobi's method. Have fun. Paul Samet ---------------------------------------- From: saj@chinet.chinet.com (Stephen Jacobs) How big? Up to about order 1000 brute force (and maybe some sparse storage techniques) seems adequate. Have you looked at "Numerical Recipes"? It gives references to the stuff it leaves out. In the most general case, you're screwed anyway because the problem gets so badly ill-conditioned. My personal favorite matrix book is Golub and Van Loan (title "Matrix Computations"); but it doesn't go into extreme practical details. Steve ---------------------------------------- From: rossi@t1.chem.umn.edu (Ivan Rossi) Keith, I think the absolute best around are the LAPACK routines from J. Dongarra They use BLAS level 3 wherever possible, and if you use a workstation or a Cray BLAS come optimized for maximum performance (libblas on IBM, SGI; libsci on Cray C90) . You can get the source from netlib send a message to netlib@ornl.gov or netlib.att.com saying send index from lapack or search the WWW site at http://gams.nist.gov/ happy computing Ivan Ivan Rossi | Department of Chemistry | EXPERT : (n) University of Minnesota | someone who avoids all the 207 Pleasant St. SE,Minneapolis, MN 55455 | little errors, going straight Tel. +1-612-624-6099 | towards the catastrophe e-mail : rossi@t1.chem.umn.edu | | Intel Inside : the warning label ---------------------------------------- From: Bob Funchess Dear Keith, There's a fairly fast method for sparse matrices called the "Lanczos Algorithm"; I'm not at my reference papers right now so I can't give you the exact reference. The original method had some numerical instabilities, but those are correctable... I know that the book "Biological Magnetic Resonance, vol. 8: spin labeling: theory and applications" by academic press (L. Berliner, ed.; pub. date 1988 or 1989) contains an EPR simulation program which uses this method. The book comes with a DOS disk that has the source code to the program, and the chapter which discusses the program contains the full references to the original algorithm and the correction techniques. Regards, Bob Funchess -- Dr. Robert B. Funchess bobf@msi.com Senior Support Scientist Tel (617) 229-9800 x202 Molecular Simulations Inc. FAX (617) 229-9899 16 New England Executive Park http://www.msi.com/~bobf/bobf.html Burlington, MA 01803 ---------------------------------------- From: Robert Fraczkiewicz Keith, I use Fortran routines DSYEV.F and DSPEV.F from LAPACK package available from NETLIB (a repository of mathematical software). If you have a WWW client try the following URL: ftp://netlib.att.com/netlib/master/readme.html.Z Eventually you can ftp anonymously to netlib.att.com; directory netlib/lapack. Some machines have specifically optimized LAPACK library; using it is much better than transferring source code. On UNIX machine you would type command "man lapack" to see if your sysadm installed one. I am also interested in high-speed diagonalization; if you get better answers, please, let me know and/or post the summary to CCL. Thanks in advance ! Happy computing, -- +[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+ + |\+_+/\+_+/| + + just Robert Fraczkiewicz |=\ /||\ /=| robert@roman.chem.uh.edu + + Deparment of Chemistry |--\-||-\--| chem86@jetson.uh.edu + + University of Houston |=/_\||/_\=| (713) 743-3236 + + |/+ +\/+ +\| + +[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+ + + + "It is also a good rule not to put too much confidence in experimental + + results until they have been confirmed by Theory" + + + + Sir Arthur Eddington + + + +[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+[]={}-()~|+ From S.R.Kilvington@soton.ac.uk Mon Dec 12 06:19:19 1994 Received: from beech.soton.ac.uk for S.R.Kilvington@soton.ac.uk by www.ccl.net (8.6.9/930601.1506) id FAA05973; Mon, 12 Dec 1994 05:21:09 -0500 Received: from vision.soton.ac.uk (vision.soton.ac.uk [152.78.128.74]) by beech.soton.ac.uk (8.6.9/relay-8.9) with ESMTP id KAA05138 for ; Mon, 12 Dec 1994 10:20:58 GMT Received: (from srk@localhost) by vision.soton.ac.uk (8.6.9/server-8.7) id KAA21694 for CHEMISTRY@ccl.net; Mon, 12 Dec 1994 10:20:59 GMT Message-Id: <199412121020.KAA21694@vision.soton.ac.uk> Subject: Mol2 file format To: CHEMISTRY@ccl.net Date: Mon, 12 Dec 1994 10:20:58 +0000 (GMT) From: "Simon Kilvington" X-Mailer: ELM [version 2.4 PL23] Content-Type: text Content-Length: 140 Dear all, does anyone know where I can get hold of the definition of the "Mol2" file format? Si. --- Simon Kilvington, srk@soton.ac.uk From KATEY@psipsy.uct.ac.za Mon Dec 12 06:22:14 1994 Received: from ucthpx.uct.ac.za for KATEY@psipsy.uct.ac.za by www.ccl.net (8.6.9/930601.1506) id FAA06144; Mon, 12 Dec 1994 05:36:39 -0500 Received: by ucthpx.uct.ac.za (/\==/\ Smail3.1.28.1 #28.43) id m0rH86b-000KGkC; Mon, 12 Dec 94 12:35 SAST Received: From ITS/WORKQ1 by charongw1.uct.ac.za via Charon-4.0A-VROOM with IPX id 100.941212123218.352; 12 Dec 94 12:38:32 -0200 Message-ID: To: chemistry@ccl.net From: KATEY@psipsy.uct.ac.za Date: 12 Dec 94 12:36:23 SAST-2 Subject: min. dist. between protons X-pmrqc: 1 Priority: normal X-mailer: Pegasus Mail v2.3 (R4). I am working with Macromodel, and have the problem that it generates structures in which the approach between two non-bonded hydrogens can be as little as 1.76 Angstroms, and the energy of such interactions is reported to be relatively low. I have used both the Amber force field and the MM2 force field. I don't understand how such a close interaction can have such a low energy. Can anyone please offer me an explanation? Kate Sunter. From IGOR@novelty.tau.ac.il Mon Dec 12 09:19:21 1994 Received: from decscc.tau.ac.il for IGOR@novelty.tau.ac.il by www.ccl.net (8.6.9/930601.1506) id JAA07547; Mon, 12 Dec 1994 09:10:45 -0500 Received: from charon.tau.ac.il by decscc.tau.ac.il with SMTP id AA05147 (5.67b/IDA-1.5 for ); Mon, 12 Dec 1994 16:07:53 -0200 Received: From NOVELTY/WORKQUEUE by charon.tau.ac.il via Charon-4.0A-VROOM with IPX id 100.941212161137.512; 12 Dec 94 16:09:55 -0200 Message-Id: From: "Igor - 9597" Organization: Tel-Aviv University To: CHEMISTRY@ccl.net Date: Mon, 12 Dec 1994 16:11:33 GMT+2 Subject: CCL:"ETIME" command in LINUX (f77) Return-Receipt-To: "Igor - 9597" Priority: normal X-Mailer: Pegasus Mail/Windows (v1.22) I have installed LINUX on my pentium in order to run some molecular dinamics writen in FORTRAN for unix. When I tryed to compile the programs I have found that it uses external function ETIME ( for cpu run time) but on LINUX that function doesnt seem to exist. During my investigation of a problem I have found up that LINUX doesnt realy have a f77 compiler. All it has is F77 script that uses f2c translator to compile F-programs. After searching some ftp sites I have come up with an f77.c file BUT in that ftp site there were no proper headers to compile it ( I mean .h files) so it turned out to be useless. I would be very gratefull if any of you could help me with one of the following problems: 1. What is LINUX'es external function for reading the CPU time in fortran ( i.e. ETIME) 2. Where could I find an f77.c together with all needed headers to compile it on LINUX. thank you in advance Igor Derzy. From gregory@wucmd.wustl.edu Mon Dec 12 10:00:07 1994 Received: from wugate.wustl.edu for gregory@wucmd.wustl.edu by www.ccl.net (8.6.9/930601.1506) id KAA08931; Mon, 12 Dec 1994 10:00:07 -0500 Received: by wugate.wustl.edu (5.67b/WUSTL-0.3) with SMTP id AA05265; Mon, 12 Dec 1994 09:00:02 -0600 Received: by wucmd (931110.SGI/911001.SGI) for @wugate.wustl.edu:chemistry@ccl.net id AA25665; Mon, 12 Dec 94 07:03:11 -0600 Date: Mon, 12 Dec 1994 06:58:15 -0600 (CST) From: Gregory Nikiforovich Subject: Re: CCL:min. dist. between protons To: KATEY@psipsy.uct.ac.za Cc: chemistry@ccl.net In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII On 12 Dec 1994 KATEY@psipsy.uct.ac.za wrote: > I am working with Macromodel, and have the problem that it generates > structures in which the approach between two non-bonded hydrogens can > be as little as 1.76 Angstroms, and the energy of such interactions > is reported to be relatively low. I have used both the Amber force > field and the MM2 force field. I don't understand how such a close > interaction can have such a low energy. Can anyone please offer me an > explanation? Looks like your protons are connected to the same carbon (CH2 group). They could interact in force fields you have mentioned as so called "i - i+2" atoms, which have an artificial energy potential, having nothing similar to a real situation. This is an inherent flaw for any force field using flexible valence geometry. Do not pay attention to this "interproton energy" at all. Good luck. ############################################################################## Gregory V. Nikiforovich Phone (314) 935-4677 Research Professor Fax (314) 935-4979 Center for Molecular Design E-mail address: Washington University gregory@wucmd.wustl.edu Lopata Hall, Box 1099 St.Louis, MO 63130 ############################################################################## From pueyo@pinon.ccu.uniovi.es Mon Dec 12 11:19:28 1994 Received: from opalo for pueyo@pinon.ccu.uniovi.es by www.ccl.net (8.6.9/930601.1506) id KAA09482; Mon, 12 Dec 1994 10:30:17 -0500 Received: from pinon.ccu.uniovi.es by etsiig.uniovi.es (PMDF V4.3-10 #4541) id <01HKK9I5QSTS8WY0PQ@etsiig.uniovi.es>; Mon, 12 Dec 1994 16:30:15 +0100 (GMT) Received: by pinon.ccu.uniovi.es (1.37.109.10G/16.2) id AA179686070; Mon, 12 Dec 1994 16:27:50 +0100 Date: Mon, 12 Dec 1994 16:27:50 +0100 From: Lorenzo Pueyo Subject: Re: CCL:"ETIME" command in LINUX (f77) To: CHEMISTRY@ccl.net, chemistry-request@ccl.net Message-id: <01HKK9I6FW1U8WY0PQ@etsiig.uniovi.es> Content-transfer-encoding: 7BIT Here you have a routine etime.c that compiles with no problem on a Linux box and can be called from any fortran program. On the other hand, I have been developing programs on Linux and porting them effortless to Convex (and nearly effortless to HP's) for the last year. The standard route is to use the f77 shell script directly available in the Slackware distribution of Linux. F77 uses f2c in a transparent way to translate your f77 code to c, and then it uses the gcc compiler. The process is quite fast and clean. The only problem with this method is that using the gdb debugger is cumbersome. Best regards, Victor Lua~na >>>>>>>>>>>>>>>>>>> etime.c <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< /* etime implementation, by Kevin Dowd, 'High Performance Computing', (O'Reilly, Sebastopol, CA, 1993) */ #include #define INTERVAL 60 float etime_ (tarray) struct { float user; float system; } *tarray; { struct tms local; times (&local); tarray->user = (float) local.tms_utime/INTERVAL; tarray->system = (float) local.tms_stime/INTERVAL; return (tarray->user + tarray->system); } +--------------------------------------------+ +---^---/ / ! Victor Lua~na ! | ~ / Just in case ! Departamento de Quimica Fisica y Analitica ! | | you don't ! Universidad de Oviedo, 33006-Oviedo, Spain ! < / remember ! e-mail: victor@hobbit.quimica.uniovi.es ! | / where Oviedo ! phone: (34)-8-5103523 ! |____ ___/ is ;-) ! fax: (34)-8-5103480 ! \/ +--------------------------------------------+ From frits@rulglj.LeidenUniv.nl Mon Dec 12 11:24:32 1994 Received: from rulway.LeidenUniv.nl for frits@rulglj.LeidenUniv.nl by www.ccl.net (8.6.9/930601.1506) id LAA10393; Mon, 12 Dec 1994 11:19:14 -0500 Received: from rulglj.LeidenUniv.nl by rulway.LeidenUniv.nl with SMTP id AA27418 (5.65c+/IDA-1.4.4 for <@RULWAY.LEIDENUNIV.NL:chemistry@ccl.net>); Mon, 12 Dec 1994 17:18:21 +0100 Received: by rulglj.LeidenUniv.nl (920330.SGI/920502.SGI) for @RULWAY.LEIDENUNIV.NL:IGOR@novelty.tau.ac.il id AA20904; Mon, 12 Dec 94 17:19:14 GMT Date: Mon, 12 Dec 94 17:19:14 GMT From: frits@rulglj.LeidenUniv.nl (Frits Daalmans) Message-Id: <9412121719.AA20904@rulglj.LeidenUniv.nl> To: "Igor - 9597" Subject: Re: CCL:"ETIME" command in LINUX (f77) Cc: frits@rulglj.leidenuniv.nl, chemistry@ccl.net (This reply does not seem to bear any direct importance to computational chemistry, but may be useful for people wanting to compile comp.chem. programs in Fortran under Linux.) Fortran compilers under Linux that I know of: - You can use the tandem: fortran code ---------------> (unreadable) C code -------------> executable f2c translator gcc compiler This is the only workable option at this time, AFAIK. Both f2c and gcc are obtainable from all main Linux ftp sites, and included in packages like the Slackware distribution. Although this route may seem cumbersome, *it works*. f2c comes with a small shellscript called f77 what does exactly what you'd expect (i.e. you don't see the intermediate C code any more). - Or, buy the NAGWare f90 compiler from the Numerical Algorithms Group; this one has been announced in the NetNews group comp.os.linux.announce recently (begin November?); I take the freedom to spread this information further. The advert said that more information can be obtained by mailing their informationdesk, infodesk@nag.com. - Or, wait for, or better still help with, the development of the Gnu g77 compiler. I don't know very much about it, except that it is not yet stable. Information can be gotten by using finger: finger fortran@gate.gnu.ai.mit.edu Concerning the need for some system functions that your Fortran program expects the Linux f2c library (libf2c.a or libF77.a, libI77.a) to have: If you can't find any Fortran timing routines, you can tinker around with C library functions. Those are in chapter 3 of the (online) manual. If I had time to try some compiles at home, I'd be able to give you more information :-(. I only have a general idea of some workarounds right now. Sorry. Hopefully this information is sufficiently useful to enough Computational Chemists to justify posting it here. Good luck with it, Frits Daalmans Disclaimer: all opinions stated here are mine and not necessarily those of my employer(s). Frits Daalmans OIO Conformational Analysis Gorlaeus Laboratoria Leiden, The Netherlands E-mail: frits@rulglj.leidenuniv.nl Tel: [+31] (0)71-274505 From hinsenk@ERE.UMontreal.CA Mon Dec 12 13:20:05 1994 Received: from condor.CC.UMontreal.CA for hinsenk@ERE.UMontreal.CA by www.ccl.net (8.6.9/930601.1506) id LAA11328; Mon, 12 Dec 1994 11:55:28 -0500 Received: from eole.ERE.UMontreal.CA by condor.CC.UMontreal.CA with SMTP id AA17543 (5.65c/IDA-1.4.4 for chemistry@ccl.net); Mon, 12 Dec 1994 11:10:50 -0500 Received: from cyclone.ERE.UMontreal.CA by eole.ERE.UMontreal.CA (940406.SGI/5.17) id AA09060; Mon, 12 Dec 94 11:10:49 -0500 Received: by cyclone.ERE.UMontreal.CA (940406.SGI/5.17) id AA24927; Mon, 12 Dec 94 11:10:48 -0500 Date: Mon, 12 Dec 94 11:10:48 -0500 From: hinsenk@ERE.UMontreal.CA (Hinsen Konrad) Message-Id: <9412121610.AA24927@cyclone.ERE.UMontreal.CA> To: chemistry@ccl.net Subject: PDB format: the answers Thanks to all who sent me information about where to find the PDB format definition. Since I received so many replies with information and an almost equal number of requests for forwarding it, I decided to send a public reply to all instead of spending the rest of the day answering individual mails... For those who are also looking for information on the PDB format, here the answers: The official documentation is available in PostScript format by ftp from pdb.pdb.bnl.gov as pub/format.desc.ps (which is where I got it from). It is also said to be available from www.ccl.net in the directory /pub/chemistry/documents/PDB/PDB_file_format (which I didn't check). Public domain code in C and C++ to read and write PDB is available by ftp from cgl.ucsf.edu as /pub/libpdb.shar and /pub/libpdb++.shar. I will have a look at the C++ version, it might be all I need. ------------------------------------------------------------------------------- Konrad Hinsen | E-Mail: hinsenk@ere.umontreal.ca Departement de Chimie | Tel.: +1-514-343-6111 ext. 3953 Universite de Montreal | Fax: +1-514-343-7586 C.P. 6128, succ. A | Deutsch/Esperanto/English/Nederlands/ Montreal (QC) H3C 3J7 | Francais (phase experimentale) ------------------------------------------------------------------------------- From AZHARI@FRCU.EUN.EG Mon Dec 12 13:41:30 1994 Received: from FRCU.EUN.EG for AZHARI@FRCU.EUN.EG by www.ccl.net (8.6.9/930601.1506) id LAA10947; Mon, 12 Dec 1994 11:46:58 -0500 From: Received: from FRCU.EUN.EG by FRCU.EUN.EG (PMDF V4.2-11 #3805) id <01HKKE9IJPS0006GWY@FRCU.EUN.EG>; Mon, 12 Dec 1994 18:46:31 O Date: Mon, 12 Dec 1994 18:46:31 +0000 (O) Subject: quadrupole coupling contstants To: chemistry@ccl.net Message-id: <01HKKE9IM4KY006GWY@FRCU.EUN.EG> X-VMS-To: IN%"chemistry@ccl.net" MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII Content-transfer-encoding: 7BIT Dear CCL: Dose anyone know how to calculate the quadrupole coupling constants in MHz using the following data from an ab initio job? Thank you very much for your assisstance. Adel El-Azhary Cairo University =========================================================================== Dipole Moment in atomic units x y z Electronic -25.22565938 0.00000000 41.14626548 Nuclear 26.16293578 0.00000000 -41.43691737 Total 0.93727640 0.00000000 -0.29065188 Dipole Moment 0.98130809 In Debyes x y z Electronic -64.11726324 0.00000000 104.58342815 Nuclear 66.49958343 0.00000000 -105.32219193 Total 2.38232019 0.00000000 -0.73876377 Dipole Moment 2.49423764 Second Moments and Quadrupole Moments are calculated relative to the origin of the coordinate frame Second Moments - in atomic units XX YY ZZ XY XZ YZ Electronic -114.12783 -29.06291 -149.51676 0.00000 37.19446 0.00000 Nuclear 95.35667 11.17119 133.49805 0.00000 -36.93342 0.00000 Total -18.77116 -17.89172 -16.01871 0.00000 0.26104 0.00000 Quadrupole Moments -- in atomic units XX YY ZZ XY XZ YZ Electronic -24.83799 102.75938 -77.92138 0.00000 55.79170 0.00000 Nuclear 23.02205 -103.25616 80.23412 0.00000 -55.40013 0.00000 Total -1.81595 -0.49679 2.31273 0.00000 0.39156 0.00000 Quadrupole Moments -- in E.S.U. units (10**-26 ESU CM**2) XX YY ZZ XY XZ YZ Electronic -33.40797 138.21497 -104.80700 0.00000 75.04179 0.00000 Nuclear 30.96546 -138.88316 107.91770 0.00000 -74.51512 0.00000 Total -2.44251 -0.66819 3.11071 0.00000 0.52667 0.00000 Electric fields at nuclei (atomic units) Atom Ex Ey Ez N -0.0843679 0.0000000 0.0354536 C -0.0072302 0.0000000 0.0005697 C 0.0028301 0.0000000 -0.0022156 C -0.0000986 0.0000000 -0.0138588 H 0.0013261 0.0000000 0.0090170 H 0.0040023 -0.0075692 -0.0014053 H 0.0040023 0.0075692 -0.0014053 H -0.0032318 -0.0080626 -0.0032587 H -0.0032318 0.0080626 -0.0032587 From mercie@mail.med.cornell.edu Mon Dec 12 14:19:29 1994 Received: from mail.med.cornell.edu for mercie@mail.med.cornell.edu by www.ccl.net (8.6.9/930601.1506) id NAA13174; Mon, 12 Dec 1994 13:51:57 -0500 Received: (from mercie@localhost) by mail.med.cornell.edu (8.6.9/ech1.75) id NAA15591; Mon, 12 Dec 1994 13:51:56 -0500 Date: Mon, 12 Dec 1994 13:51:55 -0500 (EST) From: Gustavo Mercier To: CCL Subject: Reference for Hyperchem 2.0 in SGI Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Hi! I have not been successful communicating with hyperchem people so... Does anybody have the reference for the now defunct Hyperchem v. 2.0 for SGI? I need it for a manuscript! Thanks! Sincerely, Gustavo A. Mercier, Jr. mercie@mail.med.cornell.edu From hurst@hyper.hyper.com Mon Dec 12 17:19:50 1994 Received: from quartz.sentex.net for hurst@hyper.hyper.com by www.ccl.net (8.6.9/930601.1506) id QAA16145; Mon, 12 Dec 1994 16:46:40 -0500 Received: from hyper.hyper.com (Uhyper@localhost) by quartz.sentex.net (8.6.5/8.6.9) with UUCP id QAA18866; Mon, 12 Dec 1994 16:42:48 -0500 Received: by hyper.hyper.com.hyper.com (920330.SGI/890607.SGI) (for mail.med.cornell.edu!mercie) id AA28861; Mon, 12 Dec 94 16:54:22 -0500 Date: Mon, 12 Dec 94 16:54:22 -0500 From: hurst@hyper.hyper.com (Graham Hurst) Message-Id: <9412122154.AA28861@hyper.hyper.com.hyper.com> To: Gustavo Mercier Subject: Re: CCL:Reference for Hyperchem 2.0 in SGI Cc: chemistry@ccl.net Reply-To: hyper.com!info@hyper.hyper.com Gustavo Mercier writes: > Hi! > > I have not been successful communicating with hyperchem people so... We had trouble with email to hyper.hyper.com timing out for two weeks but it is now fixed. Info@hyper.com and support@hyper.com were unaffected. > Does anybody have the reference for the now defunct > Hyperchem v. 2.0 for SGI? I need it for a manuscript! You might check with journal submission guidelines or ask an editor if there is a standard format for citing commercial software. One format I've seen (that follows the usual book format) is: HyperChem for SGI: Release 2, Autodesk Inc., Sausalito CA, USA (1992) where the "HyperChem for SGI: Release 2" is in italics. For HyperChem Release 4 use: HyperChem for Windows: Release 4, Hypercube Inc., Waterloo, Ontario, Canada, 1994. We'd love to have reprints of any papers that mention using HyperChem! Cheers, Graham ------------ Graham Hurst, PhD Email to: hurst@hyper.com (until 16 Dec 1994) Until 16 Dec 1994 at: Hypercube Inc, 7-419 Phillip St, Waterloo, Ont, Canada N2L 3X2 (519)725-4040 Info requests to: info@hyper.com Support questions to: support@hyper.com Email group: Send "subscribe hyperchem" to hyperchem-request@hyper.com From huang@mazda.wavefun.com Mon Dec 12 18:19:28 1994 Received: from volvo.wavefun.com for huang@mazda.wavefun.com by www.ccl.net (8.6.9/930601.1506) id RAA16899; Mon, 12 Dec 1994 17:54:54 -0500 Received: from mazda.wavefun.com by volvo.wavefun.com via SMTP (931110.SGI/930416.SGI) for chemistry@ccl.net id AA06912; Mon, 12 Dec 94 14:54:42 -0800 Received: by mazda.wavefun.com (931110.SGI/930416.SGI) for workshop@wavefun.com id AA25554; Mon, 12 Dec 94 14:54:30 -0800 From: "Wayne Huang" Message-Id: <9412121454.ZM25551@mazda.wavefun.com> Date: Mon, 12 Dec 1994 14:54:28 -0800 Reply-To: huang@wavefun.com X-Mailer: Z-Mail (3.1.0 22feb94 MediaMail) To: chemistry@ccl.net Subject: Computation Workshop Cc: workshop@wavefun.com Content-Type: text/plain; charset=us-ascii Mime-Version: 1.0 Announcing ... ------------------------------------ | COMPUTATIONAL CHEMISTRY WORKSHOP | ------------------------------------ Intensive 3-day workshops focus on the application of modern electronic structure methods to chemistry. Lectures will describe underlying theory, assess the performance of modern electronic structure methods, outline practical strategies for doing calculations, apply these methods to diverse chemical problems and illustrate graphical models on description of molecular structure, properties, chemical reactivity and selectivity. Laboratories provide hands-on experience using a wide selection of experiments, as well as ample time to explore your own chemistry. Lecturer: Prof. Warren Hehre; Lab instructor: Dr. Wayne Huang The upcoming 1995 Workshop Schedule: January Workshop: January 11, 12 & 13, 1995 February Workshop: February 1, 2 & 3, 1995 March Workshop: March 29, 30 & 31, 1995 (This is right before Anaheim ACS Meeting, Anaheim is 20 minutes from Irvine) Enrollment is strictly limited to eight people per workshop. Small class size ensures each participant their own IBM RS/6000 37T workstation and plenty of individual attention. Workshops will held at Wavefunction, Inc. Irvine, California, USA To request a workshop brochure, please send e-mail to: workshop@wavefun.com or for more information, contact: Dr. Wayne Huang Wavefunction, Inc. 18401 Von Karman, Suite 370 Irvine, CA 92715 (714)955-2120, Fax:(714)955-2118 E-mail: huang@wavefun.com