Hi Jan, Some times this issue occurs due to large update size for the CG potentials. This could be avoided by using a scaling parameter of less than 1.0, which would effectively scale down the update size and avoids the unphysically large update to the CG potential and hence, blow up of MD simulations. So I suggest you to try out using a scale parameter between 0.0 - 1.0 which can be specified as inverse.post_update_options.scale for each CG interaction, i.e.,
<cg> <nonbonded> ..... <inverse> ... <post_update_options> <scale>0.7</scale> </post_update_options> </nonbonded> </cg> I hope it helps. Sikandar On Wed, Jun 10, 2015 at 6:15 AM, Jan Henning Peters <[email protected]> wrote: > Hi, > > I have run several ibi calculations to get single-bead cg models for > differnt liquids (alkanes from hexane to docosane). While initial attempts > were quite sucessful, I now have run into the following problem. First of > all, none of my ibi calculations finished the number of iterations > (iterations_max was set to 300 without a convergence check that would have > ended the calculation earlier) I set, but all exited during the GROMACS > simulation, apparently due to an "exploding system". When I looked at the > results of the iterations that did finish, I saw that the rdf of the system > did quickly converge towards the target rdf (usually during the first 20-30 > iterations), but then started to diverge again. > > I have added two plots from an ibi-run that crashed after 164 iterations. > As can be seen, the sum of squared differences between the iteration rdf > and the target rdf first quickly decreases (it reaches a minimum at > iteration 38) but then increases again until the simulation crashes. The > other plot shows a comparison of the target rdf, the best rdf and the one > of the last iteration that finished simulating. > > I'm using the development version of Votca (the calculations shown were > performed on git commit dd7c19fc4fa8392294168e897a52d65ad6fdb7cf from Tue > Apr 28 10:17:49 2015 -0600). > > Any ideas ? > > Regards, > Jan > > -- > You received this message because you are subscribed to the Google Groups > "votca" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/votca. > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "votca" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/votca. For more options, visit https://groups.google.com/d/optout.
