<html><body><div style="color:#000; background-color:#fff; font-family:times new roman, new york, times, serif;font-size:12pt"><div>Hi everyone!</div><div><br></div><div>I have a 20 ns simulation (2fs timestep but coordinates saved every 0.2 ps) and I was able to calculate the mean-square-displacement of the oxygen atoms of my water solvent. The entire MSD plot looks very linear to me. I need to fit this set of MSD data to obtain the diffusion coefficient. However, I am not sure if I will have to use the entire MSD points or may be just to consider a few picoseconds at the latter part of the simulation time. In papers I have read, they use up to a 100 ps, although no mention were made as to where those set of points came from. Please help.</div><div><br></div><div>This post seems to suggest a nanosecond window for the analysis of the MSD
curve</div><div><br></div><div>http://lists.gromacs.org/pipermail/gmx-users/2012-January/067487.html</div><div><br></div><div>Although, this other one suggests using the latter 40 ps of his MSD</div><div><br></div><div>http://lists.gromacs.org/pipermail/gmx-users/2009-November/046990.html</div><div><br></div><div>Your help is greatly appreciated.</div><div><br></div><div>Thanks,</div><div><br></div><div>Bernard<br></div><div><br></div><div></div></div></body></html>