Diffusion of Center of Mass

I am working on a system and I would like to calculate the diffusion coefficient using the mean-squared deviation (MSD) method. I have 343 acetone molecules in my box (3430 atoms). I have several questions:

1. I would like to calculate the MSD for each molecule I have. I am unsure what the manual means about the command:

compute 3 all msd com yes

Does this refer to the center of mass? Will this command compute the MSD for each molecule or each atom?

1. Can I group the atoms into molecules and refer to the center of mass MSD for molecules? How would I do that?

Thanks,

Josh

I am working on a system and I would like to calculate the diffusion
coefficient using the mean-squared deviation (MSD) method. I have 343
acetone molecules in my box (3430 atoms). I have several questions:

1) I would like to calculate the MSD for each molecule I have. I am
unsure what the manual means about the command:

compute 3 all msd com yes

Does this refer to the center of mass? Will this command compute the MSD for
each molecule or each atom?

the "com yes" flag would remove a center of mass drift of the whole
system. for your kind of system, you should not have a significant
drift of that kind, as that would result in other properties, e.g.
temperature being off (in the most extreme case, you get the so-called
flying icecube syndrome).
it will compute the MSD averaged over each atom. but in the long time
limit where the einstein relation applies, that is used here, there is
no difference.

2) Can I group the atoms into molecules and refer to the center of mass
MSD for molecules? How would I do that?

no. and there really is not need for that. see above.

axel.

There is also a compute msd/chunk command which can calculate

an MSD per molecule. But I agree with Axel that it is probably

not needed for what you want. Just compute msd should be fine.

Steve