From owner-nwchem-users@emsl.pnl.gov Thu Oct 30 11:47:01 2008 Received: from odyssey.emsl.pnl.gov (localhost.pnl.gov [127.0.0.1]) by odyssey.emsl.pnl.gov (8.14.1/8.14.1) with ESMTP id m9UIl0PW022972 for ; Thu, 30 Oct 2008 11:47:01 -0700 (PDT) Received: (from majordom@localhost) by odyssey.emsl.pnl.gov (8.14.1/8.14.1/Submit) id m9UIl0Uo022971 for nwchem-users-outgoing-0915; Thu, 30 Oct 2008 11:47:00 -0700 (PDT) X-Authentication-Warning: odyssey.emsl.pnl.gov: majordom set sender to owner-nwchem-users@emsl.pnl.gov using -f X-Ironport-SG: OK_Domains X-Ironport-SBRS: 2.3 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: AtECAN2dCUmp6IYDe2dsb2JhbACUCwEBFiKuNoYvhAoBg1A X-IronPort-AV: E=Sophos;i="4.33,517,1220252400"; d="scan'208";a="85481382" Message-ID: <490A011C.5080906@chem.ucla.edu> Date: Thu, 30 Oct 2008 14:46:52 -0400 From: "Steven E. Wheeler" User-Agent: Thunderbird 2.0.0.17 (Windows/20080914) MIME-Version: 1.0 To: nwchem-users@emsl.pnl.gov Subject: [NWCHEM] DFT geometry optimizations with more than 100 atoms Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: owner-nwchem-users@emsl.pnl.gov Precedence: bulk It looks like this fix just addresses the hard-limit of 100 atoms for numerical gradients. Gianfranco raised the larger issue of why NWChem switched to numerical gradients in the first place. We have had similar problems, and it appears to have little to do with the number of atoms. Often an optimization that runs fine (using analytic gradients) one time will run a second time using numerical gradients, without any changes to the input file! To date, we have found no pattern, which is mysterious. If the same job is re-run several times it will eventually run correctly (ie: using analytic gradients), but that is hardly a satisfactory solution. Any suggestions? Sincerely, Steven Wheeler --- Hi Gianfranco, Please make the following changes in task_num_grad.F (in /task) and = recompile. In the subroutine=20 subroutine fd_make_vectors(rtdb, geom, natoms, natoms3, $ coords, temp, $ vectors, nvec, oprint) Change the following: 1.=20 Change: logical oactive(maxat) To: logical oactive(natoms) 2. Change: double precision work(maxat3, nwork), eval(maxat3) To: double precision work(natoms3, nwork), eval(natoms3) 3. Comment out the following lines if (natoms.gt.maxat) call errquit('fd_make_vec: hard dim', natoms, & INPUT_ERR) Regards, -Niri=20 -----Original Message----- From: owner-nwchem-users@emsl.pnl.gov = [mailto:owner-nwchem-users@emsl.pnl.gov] On Behalf Of Gianfranco = Ercolani Sent: Tuesday, October 28, 2008 9:23 AM To: nwchem-users@emsl.pnl.gov Subject: [NWCHEM] DFT geometry optimizations with more than 100 atoms Hello, I am a new user of NWChem. I am trying to perform DFT geometry = optimizations with more than 100 atoms, but all calculations ended with = the error: > NWChem Numerical Gradients Module > --------------------------------- > > > = ------------------------------------------------------------------------ > fd_make_vec: hard dim 108 > =20 > ---------------------------------------------------------------------- > -- > = ------------------------------------------------------------------------ > current input line : > 129: task dft optimize > =20 > ---------------------------------------------------------------------- > -- > ------------------------------------------------------------- =20 > fd_make_vec: hard dim 108 > =20 > ---------------------------------------------------------------------- > -- > = ------------------------------------------------------------------------ > current input line : My input specification is > > dft > direct > xc m06-L > end > > task dft optimize I have read that "...You get that error for using more than 100 atoms in = a numerical geometry optimization...", but I have never specified that I = want numerical gradients. I think that NWChem switches to numerical gradients when the number of = atoms is higher than a fixed limit. How can I specify that I want = analytical gradients even with 108 atoms? Thanks in advance for any help/suggestion Gianfranco _____________________________________ Prof. Gianfranco Ercolani Dipartimento di Scienze e Tecnologie Chimiche Universit=E0 di Roma Tor = Vergata Via della Ricerca Scientifica 00133 Roma - Italy Phone: (+39) 0672594383 Fax: (+39) 0672594328 E-mail: ercolani@uniroma2.it=20