From owner-chemistry $#at#$ ccl.net Mon Mar 27 18:54:01 2006 From: "Wirawan Purwanto wirawan0_._yahoo.com" To: CCL Subject: CCL: Question on NWChem run (UHF, singlet) Message-Id: <-31340-060327172544-13364-jhp3COcxmibKdeZMGmBYbQ() server.ccl.net> X-Original-From: Wirawan Purwanto Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=iso-8859-1 Date: Mon, 27 Mar 2006 14:25:38 -0800 (PST) MIME-Version: 1.0 Sent to CCL by: Wirawan Purwanto [wirawan0 a yahoo.com] Hi, I am a new member of this list, and I appreciate the opportunity to join CCL. I have a question concerning running NWChem for UHF calculations for singlet states. Let's take, for example, a water molecule (H2O) or a singlet dimer (such as Li2, P2, etc.). Naive setting scf uhf; singlet; end task scf energy would simply give the RHF result, so let's not discuss that. I got some suggestion to use scf uhf; singlet vectors swap alpha 4 5 swap beta 5 6 end for a water molecule (STO-6G basis), but this does not get to the lowest UHF energy. I have been trying swapping different states, it would not yield any positive progress. Prior to this, I used GAMESS, and I could run the singlet/UHF run easily (I'm not for or against one package or the other, though). What I did with GAMESS was to use the RHF final configurations as the initial "alpha" states, and random configurations as the initial "beta" states -- usually I'm lucky this way, and got the UHF total energy lower than the RHF total energy. Now, with NWChem, I could not find a way to fully "customize" the initial configuration (i.e. specify the vector elements directly). The only way I know is to use *.movecs file generated priorly. I don't have time or energy to create my own *.movecs, though. Does anyone has any suggestion? Or if you need more detail to answer my question, I will gladly provide it. Thanks, Wirawan __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com