Rigid bodies breaking apart during simulation

[annaashki]

Hi all,
The python script for the simulation I'm running is attached, the input parameters are nb = 21, lc = 21, ang = 135, betaP = 0.8. In the original configuration file (test2.gsd) everything looks fine, but as soon as I start running simulations, some of the constituent particles start breaking apart from the rigid body they're in (some of this can, I think, be seen in the screenshot) in the following way: when one of the coordinates starts to approach one of wall, the sign of this coordinate flips, i.e. it is placed near the opposite wall. This is a clumsy explanation, so an example (where f is one of the frames in the trajectory file) might help clear that up:
In: f.configuration.box
Out: array([51.01068, 57.3066 , 65.29509, 0. , 0. , 0. ],
dtype=float32)
In: f.particles.body[131937:131957] #including this to show that the particles in question all belong to the same rigid body)
Out: array([5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997,
5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997, 5997])
In: f.particles.position[131937:131957]
Out: array([[ 24.574509, 27.277618, -14.045145],
[ 24.832594, 27.259283, -15.011026],
[ 25.052984, 27.24666 , -15.986291],
[ 25.235336, 27.23977 , -16.969435],
[ 25.379372, 27.238625, -17.958942],
[ 25.484865, 27.243223, -18.953285],
[-25.459023, 27.253561, -19.950935],
[-25.431036, 27.26962 , -20.95035 ],
[-25.441898, 27.291376, -21.94999 ],
[-25.49159 , 27.318798, -22.948315],
[ 25.430641, 27.35184 , -23.94378 ],
[ 25.303574, 27.390451, -24.934858],
[ 25.138083, 27.434576, -25.920017],
[ 24.934425, 27.484142, -26.897738],
[ 24.692913, 27.539076, -27.866514],
[ 24.41392 , 27.59929 , -28.82485 ],
[ 24.097874, 27.664696, -29.771269],
[ 23.745264, 27.73519 , -30.704311],
[ 23.356636, 27.810661, -31.62254 ],
[ 22.932587, 27.890997, -32.524536]], dtype=float32)

The same thing happens with the y and z coordinates. Does anyone know what is causing this and how I can fix it? Many thanks!
simulation_full.txt

[erjank]

Hello annashki, The simulation volumes in HOOMD (and other engines) avoid problems with surfaces--and enable the representation of a bulk material with a finite number of simulation elements--using *periodic boundary conditions*: Particles are supposed to come in from the left after exiting to the right. Allen and Tildesley or Frenkel and Smit are two great texts to dig into for more details. Best, Eric

[annaashki]

hi! ahh okay thank you, my first thought was periodic b.c.'s as well but the particles nominally belonging to the same rigid body threw me off. accounting for this should resolve the issue I was having with my calculations, I will try in the morning and update!
UPD: yes that was it, thanks again!