Bond atoms missing with Angle potential under shear using deform with remap v

I am using a version from July 19th of 2011.

My neighbor list settings are:

neighbor 1 bin
neigh_modify delay 0
communicate multi vel yes # required for dpd/tstat

this looks proper. perhaps you can experiment
with a larger neighbor skin.

You bring up a good point. Since the angle potential is a 3-particle
potential, does this mean the neighbor skin should be extended to reach
twice the cut-off of any two bonded atoms?

under normal circumstances, that is not a problem,
since bonded interactions are stored on the processors
where the corresponding reference atom is local
(the middle atom for angles) and non-bonded cutoffs
are typically much larger than bonds.

if you suspect this is a problem in your setup, you can
augment the default behavior using the "communicate
cutoff" option.


This could be the problem if you are using
a common repulsive-only FENE cutoff
of 1.12 and a small neighbor skin. Then
you don't have enough neighbors to always
find the bond partners. Depends on your
bond length and the newton bond setting.

You can either increase the skin or use
the comm cutoff command Axel mentioned
to get around this.