From owner-nwchem-users@emsl.pnl.gov Thu Oct 21 14:08:03 2004 Received: from odyssey.emsl.pnl.gov (localhost [127.0.0.1]) by odyssey.emsl.pnl.gov (8.12.10/8.12.10) with ESMTP id i9LL835J024839 for ; Thu, 21 Oct 2004 14:08:03 -0700 (PDT) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.12.10/8.12.10/Submit) id i9LL83oP024838 for nwchem-users-outgoing; Thu, 21 Oct 2004 14:08:03 -0700 (PDT) Date: Thu, 21 Oct 2004 14:07:56 -0700 From: "Andrew G. Stack" Subject: optimize with eigenvectors To: nwchem-users@emsl.pnl.gov Message-id: <47CEB31C-23A5-11D9-B3F0-000A9568D86E@ucdavis.edu> MIME-version: 1.0 (Apple Message framework v619) X-Mailer: Apple Mail (2.619) Content-type: multipart/alternative; boundary=Apple-Mail-9--173407687 Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk --Apple-Mail-9--173407687 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; format=flowed Hello, We have run into a problem and would like to know if anyone in the NWChem User's group has had experience with this. Its a DFT frequency calculation (using the LDA and the DZVP basis set) with nwchem 4.6 on mpp2. The system is a polyoxometallate with a first hydration shell included and a large number of hydroxyls (101 atoms total). We optimized using the "tight" keyword in driver and then calculated frequencies, but this results in many large negative frequencies (>= -530 cm-1). We know that in nwchem, it is possible to follow a particular mode to a transition state using the "track" keyword and the stepper optimizer. Our question for the users' group is, is it possible to do the reverse: Can we use the track or moddir keywords in nwchem to follow the negative modes to the minimum instead of the saddle point while optimizing? We ask this because it appears that in this system, optimizing using energy gradients is not working and that optimizing by following the eigenvectors might be a better approach (of course, using an gradient-optimized geometry as the starting point). In the NWChem manual, it says that using the track keyword is not valid for a minimization walk because it doesn't give the absolute minimum, so perhaps we're on the wrong track and there is a way to tweak the optimization criteria to give a better minimum? We're not sure. Thank you very much for you suggestions, we're very grateful! Regards, Andrew Stack Jim Rustad -- Andrew G. Stack Postgraduate Researcher Dept. LAWR University of California Davis, CA, USA 95616 Phone(work): +1 (530) 752-2146 Fax: +1 (530) 752-1552 E-mail: astack@ucdavis.edu --Apple-Mail-9--173407687 Content-Transfer-Encoding: 7bit Content-Type: text/enriched; charset=US-ASCII Hello, We have run into a problem and would like to know if anyone in the NWChem User's group has had experience with this. Its a DFT frequency calculation (using the LDA and the DZVP basis set) with nwchem 4.6 on mpp2. The system is a polyoxometallate with a first hydration shell included and a large number of hydroxyls (101 atoms total). We optimized using the "tight" keyword in driver and then calculated frequencies, but this results in many large negative frequencies (>= -530 cm-1). We know that in nwchem, it is possible to follow a particular mode to a transition state using the "track" keyword and the stepper optimizer. Our question for the users' group is, is it possible to do the reverse: Can we use the track or moddir keywords in nwchem to follow the negative modes to the minimum instead of the saddle point while optimizing? We ask this because it appears that in this system, optimizing using energy gradients is not working and that optimizing by following the eigenvectors might be a better approach (of course, using an gradient-optimized geometry as the starting point). In the NWChem manual, it says that using the track keyword is not valid for a minimization walk because it doesn't give the absolute minimum, so perhaps we're on the wrong track and there is a way to tweak the optimization criteria to give a better minimum? We're not sure. Thank you very much for you suggestions, we're very grateful! Regards, Andrew Stack Jim Rustad -- Andrew G. Stack Postgraduate Researcher Dept. LAWR University of California Davis, CA, USA 95616 Phone(work): +1 (530) 752-2146 Fax: +1 (530) 752-1552 E-mail: astack@ucdavis.edu --Apple-Mail-9--173407687--