From: Mercer, Brian S. (mercer9_at_llnl.gov)
Date: Mon Jun 22 2015 - 10:14:28 CDT

Hi Axel,

Thanks for your reply.

1) You are right about this, however it looks like I get lucky in my particular case. In all my system's "type A B" pairs (that I care about), the A-A and B-B pairs are either never bonded, or always bonded, and the cutoffs I would use aren't high enough to create falsely visualized bonds. So the visualization ends up working well enough for me with this method.

2) I actually looked at bond order versus bond length LAMMPS output for the bonds in my system and saw a near-perfect 1 to 1 relation between the two, which is why I was thinking I could single out distance as a breaking criteria. At any rate, it is a good point that VMD won't pick up bonds across periodic boundaries. I'll work on using the reax/c/bonds output from LAMMPS to do the bond counting. I was hoping to use VMD to see what happens to bond counts when changing the bond distance cutoff for breaking, but I suppose I could get the bond count numbers by changing the bond order cutoff criteria, using LAMMPS "rerun", and processing those results.

Brian

--
Brian Mercer
Lawrence Livermore National Laboratory
University of California Berkeley
Lawrence Graduate Scholar and PhD Candidate
-----Original Message-----
From: Axel Kohlmeyer [mailto:akohlmey_at_gmail.com] 
Sent: Friday, June 19, 2015 2:12 PM
To: Josh Vermaas
Cc: Mercer, Brian S.; vmd-l_at_ks.uiuc.edu
Subject: Re: vmd-l: Visualizing and counting broken bonds in VMD
On Fri, Jun 19, 2015 at 3:55 PM, Josh Vermaas <vermaas2_at_illinois.edu> wrote:
> Hi Brian,
>
> 1.) Can you use three separate representations? For instance if you 
> had atoms of type A, B, and C, you could have one representation for 
> "type A B", one for "type B C", and one for "type A C". Its perfectly 
> possible that it won't look right, but at the very least each of the 
> representations can be set to have a difference distance cutoff, and 
> that might be what you are looking for.
there is a known problem with this approach: with "type A B" you not only capture A-B bonds but also trigger A-A and B-B and similar for the others. if the cutoff is particularly long, you'll get a lot of bonds.
> 2.) Aren't you basically asking the question of "are these two things 
> close together?" a whole bunch of times? I would use the "measure contacts"
> (http://www.ks.uiuc.edu/Research/vmd/current/ug/node138.html) command 
> to get me the full list of atoms of a certain type that are within 
> some cutoff of one another (if pbc is an issue, exwithin will probably 
> need to be used, and the scripting gets more involved). These lists 
> would shrink with time as the number of bonds shrinks, and that would 
> I think be the easiest way of getting a numerical answer as to how 
> many bonds have been broken. You could also keep track of which are 
> the bonds that break this way by analyzing the lists. The caveat here 
> is that measure contacts works on things that aren't bonded, so you'd 
> need to clear the automatically generated bondlist first (with topotools for instance) to have it do what you want.
the problem with ReaxFF calculations is that what is considered a bond is not directly determined by a distance cutoff, but rather by the bond order. there is some correlation between the two, yet if one wants reasonably accurate and consistent information, one needs to consider the parameters that are also used as input in the ReaxFF simulation. as i mentioned on my other response, LAMMPS has such a facility built in. so this analysis is mostly a matter of writing a perl/python/whatever script to post-process the file with the bonding information generated during the simulation.
axel.
> In general, VMD's programming assumes that bonds are not dynamically 
> made and broken, so its going to be unwieldy to go around changing 
> what was meant to be an unchanging topology on a per frame basis.
>
> -Josh Vermaas
>
>
> On 6/19/15 1:43 PM, Mercer, Brian S. wrote:
>
> Hi VMD users,
>
>
>
> I am doing ReaxFF simulations of a crystalline polymer undergoing 
> fracture, and am seeking a way to visualize the bond-break behavior, 
> as well as track which bonds have broken over time. I am new to using 
> VMD but it seems like it has a lot of the tools in place to do this, 
> though I’m currently stuck on a few fronts.
>
>
>
> I’m currently doing simulations in LAMMPS and dumping the trajectory 
> to an xyz file to read into VMD.
>
>
>
> 1) Is there a way to assign different bond cutoff lengths to different 
> types of bonds in the visualizer? DynamicBonds works well for 
> visualizing except that the bond cutoff is universal. I am interested 
> in 3 distinct bond types in the system, each of which breaks at a 
> different length. I see that you can assign bond topology explicitly 
> using topotools – can perhaps I make the visualizer draw bonds 
> according to that topology, and “delete” bonds that exceed the cutoff criteria?
>
>
>
> 2) In addition to visualizing, I’d like to keep track of how many 
> bonds (and of which type) have broken over the course of the 
> simulation. Similar to how the hbonds.tcl script can be used to plot 
> the number of hydrogen bonds present in each frame. I’m again thinking 
> this could be done by assigning an initial bond topology and removing 
> bonds from the system as they exceed a specified distance criteria 
> based on their type. Or, does VMD already provide such a functionality?
>
>
>
> Please let me know if you have any thoughts on how I might proceed.
>
>
>
> Thank you,
>
>
>
> Brian
>
>
>
>
>
>
>
> --
>
> Brian Mercer
>
> Lawrence Livermore National Laboratory
>
> University of California Berkeley
>
> Lawrence Graduate Scholar and PhD Candidate
>
>
>
>
--
Dr. Axel Kohlmeyer  akohlmey_at_gmail.com  http://goo.gl/1wk0 College of Science & Technology, Temple University, Philadelphia PA, USA International Centre for Theoretical Physics, Trieste. Italy.