Originally Posted By: lennart
It is not clear to me why you don't need PBC (support for which is limited in COOR ANAL).

Somewhere I read that in an constant pressure simulation the box size is taken from the trajectory and there is no need to add "xbox @x ybox @y zbox @z"

The simulation itself was run with PBC.

Originally Posted By: rmv
If the lipids were subject to image centering during the simulation, then you must account for that; but that's not the biggest problem.

I don't believe COOR ANAL is suitable for lipid diffusion. The rigorous approach to compute diffusion is to use the center-of-mass (COM), and for lipids, that's often done as a 2D diffusion, not 3D. COOR ANAL is not set up to use the COM to compute r^2 vs t, or to compute r in 2D instead of 3D.

Also, coupling between neighbors leads to size effects; see

Dynamical motions of lipids and a finite size effect in simulations of bilayers
Jeffery B. Klauda, Bernard R. Brooks, and Richard W. Pastor
J. CHEMICAL PHYSICS 125(14) Article No. 144710 (2006)

See also this thread

Note that the finite size correction discussed in that thread only applies to cubic systems at present, and would not help with the coupling problem.

I see. So I have to come up with an own code to get the proper diffusion rate comparable to lab experiments.

But lets make the assumption I only want to observe, if the lipid diffusion constant changes during the simulation. Selection a single atom from the headgroup and calculating the diffusion constant for this atom - could this reflect those changes, couldn't it? The quick and dirty way to see if all the afford of coming up with a code to calculate the proper diffusion constant pays off.

Thanks for the link and the paper.