2015-05-29 12:57 GMT-06:00 Vitalie Botan <[email protected]>: > I have checked the exclusions and they seem to be ok, at least it works > pretty well with the same mapping on the atomistic trajectory. I suspect > it's purely numerical stability and convergence problems. In that respect I > would like to ask you how the convergence criteria is actually defined in > VOTCA, which functional form it takes (I couldn't find this information in > the manual) and what is that damping parameter (eq. 2.15 of the manual) > when doing updates, can it be controlled in the same spirit as Picard > iteration scheme? I'm not sure what is the best way to iterate over > multicomponent system: currently I update only single potential every single > step, so that means that for one iteration only one convergence parameters > should be checked. Is there also a way to control iterations with this > parameter, i.e. to reject those updates, which lead to bigger convergence > parameter? Eq 2.15 refers to scaling the update with a constant factor. This factor and also the update order for multicomponent system is very specific to the system. I would either update all interactions at the same time and scale them with 1/n or update one at the time and don't scale them at all.
We have recently extended the post-add convergence calculation (see postadd_convergence.sh for details), one can now calculated the weighted sum of multiple difference of e.g. distribution, potential with respect to the target or the last staep using L1 or L2 norm. (Defaults are L1 norm of dist.new - dist.tgt with weight 1). Using the post-add accumulated convergence script one can track the convergence over steps. See settings.xml of the urea-water/kb-ibi tutorial for details how to use that. To do what you proposed above you will need to write you own custom post-add script (see the customization section of the manual), but the overwrite post-add script (postadd_overwrite.sh) seems like a good starting point. Christoph > > --Vitalie > > On Thursday, May 21, 2015 at 7:45:33 PM UTC+2, Christoph Junghans wrote: >> >> >> It seems like there maybe some exclusions missing. Remember VOTCA used >> Gromacs-like exclusion scheme with nrexcl=1, meaning all atom involved in an >> bond, angle or dihedral are excluded. >> >> You can check the exclusions with >> $ csg_dump --top XXX --cg YYY --excl >> and add more bonds to the mapping file to exclude more atom. >> >> Christoph > > -- > 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. -- Christoph Junghans Web: http://www.compphys.de -- 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.
