From yubofan:~at~:guomai.sh.cn Tue Jul 8 15:56:16 1997 Received: from gm-email.guomai.sh.cn for yubofan |-at-| guomai.sh.cn by www.ccl.net (8.8.3/950822.1) id PAA17128; Tue, 8 Jul 1997 15:21:11 -0400 (EDT) Received: from aopen-k6-166 ([210.0.0.3]) by gm-email.guomai.sh.cn (8.8.5/8.8.5) with ESMTP id EAA02270 for ; Wed, 9 Jul 1997 04:17:13 +0800 Message-Id: <199707082017.EAA02270*- at -*gm-email.guomai.sh.cn> From: "Yubo Fan" To: Subject: Summary: Why does G94W's SCF option make errors? Date: Wed, 9 Jul 1997 03:18:47 -0000 X-MSMail-Priority: Normal X-Priority: 3 X-Mailer: Microsoft Internet Mail 4.70.1161 MIME-Version: 1.0 Content-Type: text/plain; charset=HZ-GB-2312 Content-Transfer-Encoding: 8bit Hi, I got many help from the CCL. Every advice is important for me, and some of them cannot be read from books or Jounals. Thank you very much. Here's the summary. Hi, ccl'ers, I calculated a molecule with 3 Phenyls. I use BLYP/3-21G**. This time, I try "restart", a SCF option that I didn't use before. I kill the job after L502.exe had been running 4 cycles, then I put "restart" after "SCF=". I ran the calculation again. L502.exe did 2 cycles and the result was gotten. Again I did the job from the very original start. After 6 cycles had been done, another result was gotten. I list them below: :With SCF=(save, restart) Convergence on energy, delta-E=1.12D-05 SCF Done: E(RB-LYP) = -692.644057005 A.U. after 2 cycles Convg = 0.2054D-03 -V/T = 2.0090 S**2 = 0.0000 KE= 6.864357999909D+02 PE=-3.864671343753D+03 EE= 1.355127785196D+03 :With SCF=save SCF Done: E(RB-LYP) = -692.644408415 A.U. after 6 cycles Convg = 0.4070D-05 -V/T = 2.0094 S**2 = 0.0000 KE= 6.861718246427D+02 PE=-3.864945652472D+03 EE= 1.355665717853D+03 Why do they have error I think it considerable great? Thanks a lot Y. FAN =================================================== FAN, Yubo Department of Chemistry Fudan University Shanghai, 200433 P. R. China Voice(86-21)65492222-4294 =================================================== --- Administrivia: This message is automatically appended by the mail exploder: CHEMISTRY ^at^ www.ccl.net: Everybody | CHEMISTRY-REQUEST ^at^ www.ccl.net: Coordinator MAILSERV #at# www.ccl.net: HELP CHEMISTRY or HELP SEARCH | Gopher: www.ccl.net 73 Anon. ftp: www.ccl.net | CHEMISTRY-SEARCH- at -www.ccl.net -- archive search Web: http://www.ccl.net/chemistry.html --- Re your message about BLYP/3-21G**, in case you didn't get an answer yet : Here, you did a Single-Point calculation, I presume. The Cut-Offs for these calcs are quite loose, thus the _small_ difference in energy (.027 mH, .04 kcal/mol). Start it from yet a different position, you'll get a third, also diff. number. This problem is I think more significant with DFT than with HF. If you want 7-8 significant digits, use the option scf(nosp) or scf(tight) when you do single points. Nest of luck Jeremy 1. This is not a SCF calculation, it is at BLYP (DFT) level. 2. Are you sure that molecular geometryies are identical? ================================================================= Dr. Buyong Ma buyong-: at :-ibmnla.chem.uga.edu Computational Center for Molecular Structure and Design Department of Chemistry University of Georgia Athens, Georgia 30602 USA Voice (706) 542-2044 ================================================================= On Tue, 24 Jun 1997, ~{76S}~}#~{( ~}wrote: > > Hi, ccl'ers, > > > I calculated a molecule with 3 Phenyls. I use BLYP/3-21G**. This time, I > try "restart", a SCF option that I didn't use before. I kill the job after > L502.exe had been running 4 cycles, then I put "restart" after "SCF=". I > ran the calculation again. L502.exe did 2 cycles and the result was gotten. > Again I did the job from the very original start. After 6 cycles had been > done, another result was gotten. I list them below: > > :With SCF=(save, restart) > Convergence on energy, delta-E=1.12D-05 > SCF Done: E(RB-LYP) = -692.644057005 A.U. after 2 cycles > Convg = 0.2054D-03 -V/T = 2.0090 > S**2 = 0.0000 > KE= 6.864357999909D+02 PE=-3.864671343753D+03 EE= 1.355127785196D+03 > > :With SCF=save > SCF Done: E(RB-LYP) = -692.644408415 A.U. after 6 cycles > Convg = 0.4070D-05 -V/T = 2.0094 > S**2 = 0.0000 > KE= 6.861718246427D+02 PE=-3.864945652472D+03 EE= 1.355665717853D+03 > > > Why do they have error I think it considerable great? > > > Thanks a lot > > Y. FAN > > > > > > --- > Administrivia: This message is automatically appended by the mail exploder: > CHEMISTRY {*at*} www.ccl.net: Everybody | CHEMISTRY-REQUEST {*at*} www.ccl.net: Coordinator > MAILSERV[ AT ]www.ccl.net: HELP CHEMISTRY or HELP SEARCH | Gopher: www.ccl.net 73 > Anon. ftp: www.ccl.net | CHEMISTRY-SEARCH |-at-| www.ccl.net -- archive search > Web: http://www.ccl.net/chemistry.html > --- > > Dr. Fan, If you look at the convergence of the two cases you see that they agree to the convergence criteria of the restarted job. Based on the information you supplied it is not clear why G94W would have stopped at this point as the RMS change in the density, the value listed as Convg, is larger than the default criteria. If you ran these with #P you can look at the status at each iteration and perhaps we can get a clue. Also include the information about what revision of G94W you are running. > Hi, ccl'ers, > > > I calculated a molecule with 3 Phenyls. I use BLYP/3-21G**. This time, I > try "restart", a SCF option that I didn't use before. I kill the job after > L502.exe had been running 4 cycles, then I put "restart" after "SCF=". I > ran the calculation again. L502.exe did 2 cycles and the result was gotten. > Again I did the job from the very original start. After 6 cycles had been > done, another result was gotten. I list them below: > > :With SCF=(save, restart) > Convergence on energy, delta-E=1.12D-05 > SCF Done: E(RB-LYP) = -692.644057005 A.U. after 2 cycles > Convg = 0.2054D-03 -V/T = 2.0090 > S**2 = 0.0000 > KE= 6.864357999909D+02 PE=-3.864671343753D+03 EE= 1.355127785196D+03 > > :With SCF=save > SCF Done: E(RB-LYP) = -692.644408415 A.U. after 6 cycles > Convg = 0.4070D-05 -V/T = 2.0094 > S**2 = 0.0000 > KE= 6.861718246427D+02 PE=-3.864945652472D+03 EE= 1.355665717853D+03 > > > Why do they have error I think it considerable great? > > > Thanks a lot > > Y. FAN > > > > > > --- > Administrivia: This message is automatically appended by the mail exploder: > CHEMISTRY \\at// www.ccl.net: Everybody | CHEMISTRY-REQUEST \\at// www.ccl.net: Coordinator > MAILSERV %-% at %-% www.ccl.net: HELP CHEMISTRY or HELP SEARCH | Gopher: www.ccl.net 73 > Anon. ftp: www.ccl.net | CHEMISTRY-SEARCH#* at *#www.ccl.net -- archive search > Web: http://www.ccl.net/chemistry.html > --- > Douglas J. Fox Director of Technical Support help -8 at 8- gaussian.com Dr. Fan, There was a condition with Rev. B.3 where G94W could incorrectly decide that the SCF had converged because the energy increased. This has been corrected and may be the cause of your problem. While the SCF=Restart option can save you time you must continue to check the result to make sure it is reasonable, as you must with any computational procedure. > Dr. Fox, > > > The Calculation Package I use is Gaussian 94 Revision-B.3. > > I will do a big job later, so I must do a test for occasional poweroff or > system crack. I kill the job after 4 cycles have done, then I retake the > job with option restart. > > Thanks for your help. > > Y. FAN > Douglas J. Fox Director of Technical Support help%!at!%gaussian.com