Hi. Suppose you do the following in an input script:
fix 3 all recenter INIT INIT INIT units box
fix 4 all msd 100 diffsquared.1
run 250000
fix 5 all msd 100 diffsquared.2
run 250000
I am running a smallish system for a long time, so I need to average data over multiple time-offset msd fixes to get a good measurement of the diffusion. However, it just occurred to me that the above is a bad way to do this. The reason is, the recenter fix adjusts the coordinates to keep the center of mass (COM) the same as it is at step 0. MSD fix 4 therefore measures MSD with COM motion subtracted out, which is what I want. However, MSD fix 5 measures MSD relative to coordinates at step 250000.
(I think) this introduces an error, because part of the motion after step 250000 arises from the recentering fix called at step 0. Thus the output of MSD fix 5 has a term which does not appear in the output of MSD fix 4.
So the question is, what is the correct way of doing what I want? Would
fix 3 all msd 100 diffsquared.1 com yes
run 250000
fix 4 all msd 100 diffsquared.2
run 250000