From owner-chemistry@ccl.net Sat Mar 14 12:00:00 2015 From: "Sergio Manzetti sergio.manzetti||outlook.com" To: CCL Subject: CCL: Gaussian error Message-Id: <-51111-150314091902-8863-6sf86EDJXnJ6JI0ARq/now||server.ccl.net> X-Original-From: "Sergio Manzetti" Date: Sat, 14 Mar 2015 09:19:00 -0400 Sent to CCL by: "Sergio Manzetti" [sergio.manzetti(~)outlook.com] Dear all, while trying to do BD(TQ) calculations on a system I get the Message: Warning: this job cannot use analytic gradients and so will do many energy evaluations. Post-SCF densities or gradients only with Real MP2, MP3, MP4SDQ, CI, CCD, and QCI. Error termination via Lnk1e in /home/apps/g09/g09-dsgroup/g09/l1.exe a What is this limitation caused by? Thanks From owner-chemistry@ccl.net Sat Mar 14 16:59:00 2015 From: "Nathan Houtz nhoutz#purdue.edu" To: CCL Subject: CCL: Is DSF valid to use for simulating nucleation? Message-Id: <-51112-150314165808-22183-WNBesjF82lBxC+xl+eWFgA*|*server.ccl.net> X-Original-From: "Nathan Houtz" Date: Sat, 14 Mar 2015 16:58:06 -0400 Sent to CCL by: "Nathan Houtz" [nhoutz:+:purdue.edu] Hello, I will be trying to simulate nucleation of tetrolic acid (TTA, A.K.A 2-butynoic acid) in a few different solvents using a custom-built code on Xeon Phi Coprocessors. I'd like to use DSF to deal with Coulombic forces, but I have recently been told on the LAMMPS users list (I'm using LAMMPS to compare results from DSF and Ewald simulations) that DSF is not valid for inhomogeneous solutions or solutions that include interfaces. This is a problem for nucleation of course. However, I've found examples of using damp-shifted forces to simulate nucleation in the literature before. Here is a paper by Milek and Zahn that simulates the nucleation of Ag in ethylene glycol using DSF: http://pubs.acs.org/doi/abs/10.1021/nl502503t Background: If you'd like to view the message thread in the LAMMPS users list, it's here: http://sourceforge.net/p/lammps/mailman/message/33597041/ . I was originally asking about something else that I don't think matters here. This is the relevant comment from Dr. Axel Kohlmeyer: > re accuracy: a good RDF by itself is not a good justification. or > rather it is difficult to see differences that are relevant. the rdf > has large variations at short range, but for liquids, its long range > structure is quickly averaged out. so that it will require very long > simulations to make any correlations visible over the noise. > furthermore, it only looks at structure averaged over spherical > shells. whether using approximations to the long-range coulomb create > large errors or not, depends a lot on the kind of system you simulate > and the kind of properties you want to compute. if you have a > homogeneous bulk system with a close to cubic simulation cell, then > the approximate coulomb causes the least errors. do you have > inhomogeneities or even an interface, then you are toast and have huge > errors because your coulomb approximation assumes a spherically > symmetric environment. So I'm wondering what your thoughts are on using DSF to simulate liquids consisting of more than one kind of molecule and especially liquids in which nucleation is taking place. Is there a difference between the simulation in the paper I linked and my own simulations that would make mine invalid? Additionally, do you agree that showing negligible difference between RDF's produced by DSF and Ewald is an adequate justification for using DSF? (I was also attempting to compare energies, but there was a separate issue there, which is what the conversation in the LAMMPS-users-list thread was about.) If this is not adequate, what other comparisons could I make to decide whether DSF is an adequate replacement for an Ewald sum? Thanks very much for your help, Nathan Houtz From owner-chemistry@ccl.net Sat Mar 14 22:26:01 2015 From: "Salter-Duke, Brian James - brian.james.duke#,#gmail.com" To: CCL Subject: CCL: Gaussian error Message-Id: <-51113-150314170955-8642-cAKo4cL0exNZdJx2Vyy2nQ-*-server.ccl.net> X-Original-From: "Salter-Duke, Brian James -" Content-Disposition: inline Content-Type: text/plain; charset=us-ascii Date: Sun, 15 Mar 2015 08:09:42 +1100 MIME-Version: 1.0 Sent to CCL by: "Salter-Duke, Brian James -" [brian.james.duke#,#gmail.com] As it says - "this job cannot use analytic gradients". There is no code in the program for analytic gradients for the level of theory you have selected. Brian. On Sat, Mar 14, 2015 at 09:19:00AM -0400, Sergio Manzetti sergio.manzetti||outlook.com wrote: > > Sent to CCL by: "Sergio Manzetti" [sergio.manzetti(~)outlook.com] > Dear all, while trying to do BD(TQ) calculations on a system I get the Message: > > > Warning: this job cannot use analytic gradients > and so will do many energy evaluations. > > Post-SCF densities or gradients only with > Real MP2, MP3, MP4SDQ, CI, CCD, and QCI. > Error termination via Lnk1e in /home/apps/g09/g09-dsgroup/g09/l1.exe a > > > What is this limitation caused by? > > Thanks> -- Brian Salter-Duke (Brian Duke) Brian.Salter-Duke^-^monash.edu Adjunct Associate Professor Monash Institute of Pharmaceutical Sciences Monash University Parkville Campus, VIC 3052, Australia