[gmx-users] g_msd Utility - (Unexpected) Different Results Depending on Output Flag

[Solomon Berman]

Greetings friends!

I have a question this morning about the g_msd utility related to the output generated by different flags.

I'm working with an LJ Liquid and am trying to determine the diffusion constant for the system.  I'm employing both the Einstein formulation and the Green Kubo formula.  Obviously, for the Einstein equation, I need the mean square displacement versus time.  The box I am using is cubic, with a length of a side of 3.602 nm. 

Now, when I pass the flag -o for an output (to get the data of the MSD of the system versus time) with g_msd, and plot the results in xmgrace, I find that the slope of the resulting graph continuously increases linearly once in the diffusion regime until the end of the simulation time.  However, if I pass the flag -mol (to get the average MSD per atom) alongside -o, the curve saturates at two-thirds of the maximum mean square displacement and the slope goes to zero.

I would like to inquire as to why I receive two different results between "g_msd -o" and "g_msd -o -mol"?  In particular, is it possible that the boundary conditions are treated differently for each output, where -o ignores the periodic boundary conditions and -mol uses the periodic boundary conditions?  I have tried combing the internet, and the man pages, and haven't been able to find anything on the reason for the difference in results.

Thank you, again, for the help everyone is so willing to provide a neophyte like myself!

Best,

Solomon Berman
Chemistry Department
Boston University