[gmx-users] g-msd diffusion of molecules

[Jun Soo Kim]

Hello,

I am trying to calculate the self diffusion coefficient of water  
molecules with tip5p water model.
Based on the manual, it seems that there can be two ways to calculate  
the diffusion of water molecules, not atoms :

1. Select the group of ( OW, HW1, HW2 ) of tip5p water in index.ndx  
to calculate the MSD. Because -mw is set as 'yes' by default, the MSD  
calculated is for the center of mass of tip5p water molecules.

2. Select the group of molecule numbers in index.ndx to calculate the  
MSD. This time, use option -mol to calculate the MSD of molecules.

Could anyone tell me if they are ok and, if so, which one is more  
reasonable?

And the method 2 above seems to be reasonable, but I am afraid that  
the mass of virtual sites OL1 and OL2 may also be counted when it  
calculate the center-of-mass MSD. Actually, this problem happens when  
we use genbox to generate the water box with tip5p. The density of  
water box is way too low for tip5p when it is generated by genbox  
because genbox counts the masses of OL1 and OL2 when it calculates  
the density. Is it still safe to use the method 2 to calculate the  
MSD for tip5p water model?

Thanks in advance.

Jun Soo