[gmx-users] periodic boundary conditions

[JAVAD MAHMOUDZADEH]

Dear Gromacs developers and usersI have three questions. I appreciate it if you answer them:
1)I am doing MD simulation on a bilayer. The problem I have encounteredis that either I do reimaging on the trajectory using –pbc mol –ur compact ornot, the number of Na ions in each leaflet is shown different when I check thetrajectory through VMD every 1000ps. I have 30 Na ions in my simulation boxwhich at the beginning of MD are distributed 15 on upper leaflet and 15 onlower leaflet. After 1 ns from the start of MD simulation, there is 9 Na ionson upper leaflet and 21 on lower leaflet. After 2 ns from the start of MDsimulation, there is 10 Na ions on upper leaflet and 20 on lower leaflet. Thisis while when I check the movement of Na ions in VMD during the course ofsimulation, none of the Na ions cross the bilayer. Once I tracked the movementsof one of Na ions, I saw suddenly it disappeared from upper leaflet and appearedin lower leaflet. Therefore, I guess the problem should be related to the periodicboundary conditions (PBC). Why I see this uneven distribution of Na ions while Ihave used –pbc mol –ur compact?
2) As far as I know, when PBC is being used for MDsimulation, no reimaging is necessary (mandatory) to be done on the trajectorybefore the analysis are performed on the trajectory (unless if we want to havea nice image when we look in the system through VMD). On the other hand I knowthat what is written in the trajectory while MD simulation is being done, isthe coordinate of the atoms inside primary box and not atoms in periodic boxeseven those atoms which the atoms in primary cell are interacting with them. Howdoes Gromacs know the correct position of atoms which have already left the boxwhile the coordinates related to them (which have written to trajectory)belongs to their periodic images? For example, when you look at a trajectory inVMD (before reimaging), you see long unusual bonds from one side of the box toanother side. The reason is that for example one hydrogen atom of a watermolecule has left the box from right hand side and its periodic image hasentered from left hand side and since the position of this newly enteredhydrogen is written to trajectory we see long unusual bond across the boxbetween oxygen and hydrogen.  In case ofmy example on hydrogen atom, from the topology, Gromacs knows that for exampleO number 20 is bonded to Hydrogen 21 so when reimaging is done the programknows that should bring back the H21 next to O20. But what about Na ions? Sincethey are not covalently bond to anywhere, how the program should know thatwhether the Na number 1233 (as an example) belongs to the upper leaflet orlower leaflet? 3- We know that to get correct data from analysis, every atomin each frame should have its correct position otherwise the analysis which aredone, produce wrong data. In one hand we know that reimaging is not necessary(mandatory) to be done on the trajectory before the analysis are performed on it.On the other hand we know that the coordinates written to the trajectory filefor the atoms which have left the box belongs to their periodic images that haveentered the box from other side and these coordinates do not show the realposition of those atoms taken into consideration while calculation of forcesand energy. Does the Gromacs itself do reimaging as part of each analyses beforethe analysis (for example calculation of number of Hbonds) is done? 
With RegardsJavad