GROMACS version: gromacs/2020.1
GROMACS modification: No
Here post your question
I am using Gromacs to model the evaporation of the ethanol film. The film is 10nm in z direction and was placed in the center of box of 60 nm in z direction, pbc in xyz. I use the NVT in 500K. However, during the simulation, the center of the mass of film always shifting in z direction although I have set up the “comm_mode = Linear; nstcomm = 1; comm_grps = System”. I want the film stay at the center of the box all the time. Any suggestion is appreciated. I attached the snapshot of my box at 0 ns and 3 ns in this post.
Many thanks & best regards
The center of the box is an arbitrary construct and you can place your system within this visualisation box where ever you want, once the simulation has completed. Some post processing with
gmx trjconv is what is required.
Many thanks for your suggestion.
I actually have tried to use
- gmx trjconv -f orig.xtc -o mol.xtc -pbc mol …
- gmx trjconv -f mol.xtc -o mol.clus.xtc -pbc cluster …
- gmx trjconv -f mol.clus.xtc -o mol.clus.center.xtc -center yes …
But it still does not work. The center of mass still shift in z direction
-pbc mol won’t do what you want, that simply puts the COM of the individual molecules into the box, and your molecules are very small.
-pbc cluster may work, what does the output of that look like? With something dispersed like you have, it may or may not do what you want.
-center will only work if you have an appropriate group to select to put into the center of the box. The entire system is unlikely to be appropriate here.
Better solution may be something like creating an index group of all the molecules that are in contact/close together in the final frame, then centering the trajectory/frame onto that group.
While the trjconv options are nice to modify things in previously-generated trajectories, it can still be important not to have excessive drift in the simulations, since at some point you’ll end up with an ice cube flying through space and leading to rounding and/or integration errors.
In your case, you seem to have quite a bit of evaporation from your film, which is likely correct if you simulate at 500K (although you might need to take some care, since few if any force fields are parametrised for 500K). This makes it hard to say whether the center of mass actually has changed; you can check that by removing the PBC jumps from the trajectory (according to the instructions above), and then plotting the centre-of-mass of the system as a function of time.
If you still have a drift in that case, get back and we might have to look into it.