There is a strange phenomena when I do Langevin dynamics. See the
Figure attached, The center of mass of my studying object shift from
postive to negative abruptly though the volecity of c.m. is postive.
Why? Is there a bug in fix langevin command?
Note, I have checked the efield is small sufficiently not to change
the size of radius of gyration of object.
The jump is due to the CM passing thru a periodic boundary.
I'm at home (snow storm) so I can't check the source code,
but the xcm() function in variables probably always puts
the CM inside the periodic box.
I was wrong. The xcm() command in a variable unwraps the
atoms via the image flags to compute the COM . So whatever
value it is printing is where the COM of those atoms are after
(possibly) passing thru the box many times. If you do a dump
custom with those atoms and use xu,yu,zu you will see the
unscaled coords (via image flag) and can verify for yourself
that the COM is correct.