simulation npt apparent divergence of com (dt?)

Dear users,

as far as you have spare time and experience, I would like to ask for your opinion regarding a npt-simulation.

Code is attached. Problem is: at some point it’s like if something went wrong, i.e. from being arrested, particles suddenly gain a huge acceleration and the velocity of the center of mass does strange things. By my experience, it seems that perhaps DT (time integration step ) is too high. Nevertheless, I tried to lower it but it didn’t help.

Notes:

  • code dumps .png as visual output -> a “movie.avi” file can be made with the following command -> mencoder “mf://*.jpg” -o movie.avi -ovc lavc -lavcopts vcodec=mjpeg
  • file xcm.dat contains at 4, 5 column the x_com, y_com; column 2 is kinetic energy

I would appreciate any help

I apologise if asking for this kind of help doesn’t pertain to the list-netiquette. In this case I will not post similar entries in the future.

Best regards,

Mario

in.pour (1.44 KB)

Dear users,

as far as you have spare time and experience, I would like to ask for your
opinion regarding a npt-simulation.

Code is attached. Problem is: at some point it's like if something went
wrong, i.e. from being arrested, particles suddenly gain a huge acceleration
and the velocity of the center of mass does strange things. By my
experience, it seems that perhaps DT (time integration step ) is too high.
Nevertheless, I tried to lower it but it didn't help.

Notes:

- code dumps .png as visual output -> a "movie.avi" file can be made with
the following command -> mencoder "mf://*.jpg" -o movie.avi -ovc lavc
-lavcopts vcodec=mjpeg
- file xcm.dat contains at 4, 5 column the x_com, y_com; column 2 is kinetic
energy

I would appreciate any help

what you are experiencing is the so-called "flying icecube syndrome".
the topic comes up frequently on the list. thus please see the mailing
list archive for details.

BTW: using an npt integrator for a discrete element model is most
unusual and it is a major contributor to creating the flying ice cube,
as your compression forces the system into some kind of "phase
transition" and then the thermostat algorithm has no alternative, but
accelerate the COM of the whole system, after it has "crystallized".

axel.

Thank you very much, Axel. Very interesting, actually.

I’m gonna work on it

Best,

Mario