Hello!
I have to use some older version of Gromacs (3.3.1 which is installed on the cluster I used for run simulations). When I applied pbc (pbc=xyz in mdp file) I noticed that the output .gro file contains the coordinates which are extended outside the simulation box. In other words my molecules are not "broken" due to pbc but rather are kept in one piece (that's why they extend the simulation box). What is more in the log file I noticed the phrase: "Removing pbc first time" and
"Done rmpbc". However apart of the visual aspect (my system is not a box during visualisation) the results of the calculation behave reasonable as pbc was still switch-on.
I had also possibility to run some limited number of calculations on different cluster with Gromacs 4.5.3 and when applied pbc=xyz (and periodic-molecules=yes
Old versions of GROMACS like 3.3.1 never write "broken" molecules. With the introduction of domain decomposition in 4.0, mdrun does write broken molecules in the trajectory. GROMACS has always removed periodicity effects (e.g. "made molecules whole") at the outset of the MD simulation to facilitate importing structures from other programs. In 3.3.1, a structure generated in GROMACS, which is whole, has nothing done to it in this initial periodicity "removal."
Using "pbc = no" means you're effectively simulating a droplet, so your observation of ball formation is intuitive. You're observing edge artifacts. The "periodic_molecules" keyword serves a completely different function (molecules of infinite length, e.g. polymers, CNT, etc). One does not necessarily need to specify "periodic_molecules" as a part of normal PBC usage.
Justin, thank you for the answer. So probably I made misteake in newer version when add "periodic-molecule tag into mdp file. Indeed in this case mz finall structures are broken to be fit into a box and I need some additional effort to re-make molecules to be in one piece again.
Alternatively, if you want you can also check:
1. The distance between periodic images, option is there in GROMACS. This should be within the twice of your box periphery from the solute surface, as you provided in "-d" option of editconf.
2. you can also do trjconv to converge all trajectory. This will remove the visualization artifacts due to pbc and give you the predefined box shape.
Hope this help
Aditya
It is OK to have atoms outside the box. There will be a corresponding void in the position that the atom "would have occupied", translated one box-length away. This is a normal phenomenon resulting from the use of PBC and the symmetry properties associated with that.
- Badges
- Science topic