Hi Anirban Paul

Why do you say that the two molecules cannot be distant more that half the box length? PBCs are a construct to simulate an 'infinite' system, like to obtain the behavior of bulk fluid by avoiding any surface effect. But in principle the original molecules of your box should be free to move outside of the box!

Hello Nicola Piasentin ,

Yes, PBCs are used to construct infinite system. And this is done as follows: Consider my system is within simulation box of 10 nm, and now it is replicated with periodicity of 10 nm. Thus , an atom placed at 12 nm should be looked as it is at 2nm. Right?

Please do comments. You can see some PBC related issues in Gromacs here:

https://www.researchgate.net/post/Is_05_nm_distance_between_protein_surface_and_boxedge_is_accepaable_in_Gromacs

https://www.researchgate.net/post/Can_two_ends_of_a_protein_artificially_touch_each_other_by_crossing_box_boundaries

Hi Anirban Paul

I think that this

https://www.researchgate.net/post/Is_05_nm_distance_between_protein_surface_and_boxedge_is_accepaable_in_Gromacs

is a problem related to cutoff distances, that is the box is too small and the protein can 'see itself', and that generates artifacts.

The second problem

https://www.researchgate.net/post/Can_two_ends_of_a_protein_artificially_touch_each_other_by_crossing_box_boundaries

is similar, the box was ok at the beginning but then the protein unfolded and interacted with itself, that is again a problem of auto-interactions that can be solved by a larger simulation box.

I am not sure that I am getting your problem, but in principle the original molecules in your box can move outside the box. PBCs will make so that another molecule is coming inside, and this is important so that the interactions can be calculated properly, but the two molecules are labelled differently, otherwise you wouldn't be able to calculate quantities such as MSD or similar. Look at PBC description here

http://manual.gromacs.org/documentation/2018.4/user-guide/terminology.html

Hi Nicola,

Thank you so much for your suggestion. Let me give an example of what kind of problem I was facing.

Consider I have a simulation box (1d for simplicity, say along x axis, in that sense it is not a box but a boundary) of length 10 unit ,extends from -5 to +5. Now I have two particles located at -4 and +4. Now any simple code will give the inter-particle distance to be 8 unit. But it should be 2 unit as per PBC.

I think the problem was in the post processing script ,where we try to find out distance between two atoms. We must use minimum image convention so that it gives the result 2 , not 8.

Please do comment and suggest.

Anirban

Hi Anirban Paul

I would say that it depends on what you are measuring. If you are interested into the distance between the two particles in the real space that the distance is 8, otherwise quantities such as the mean square displacement would have no meaning anymore. The minimum image is useful to compute interactions, and PBC ensure that every molecule is coming back in the original box, but no box is really existing so no surfaces are really there, molecules are free to diffuse outside.