The quick answer is yes, forces will be calculated. The thing with automatic PME grid calculations is that the grid will be resized in accordance with your cell's size. Therefore, it will always contain your entire system in the calculations.

However, I believe you have somewhat misunderstood the concepts of periodic boundary conditions and NAMD's wrapping.

Since you are setting PBC options, the simulation itself will always handle the atoms periodic. In other words, even if atoms move beyond the cell, they will "reappear" in a mirror image of your cell. That's what periodic boundary conditions do, they create "copies" of your system around the original cell, essentially mimicking the overall structure of a crystal (where, again, particles are oriented periodically). On the other hand, wrapping is basically an "output" option, i.e., the DCD and coordinate output files. Therefore, it doesn't really affect force calculations, it's affects what you see in the final system.

Are you setting the PME gridsize manually, with PMEGridSizeX, Y and Z values? 

Try disabling them and use only PMEGridSpacing (default value is 1.0). By using only a grid spacing value, NAMD will calculate the grid size automatically, based on the grid spacing value and the cell size.

Oh, It would be helpful. Thanks you so much!

Then, when setting the PME gridsize automatic and the WrapAll off, does the PME processing consider the molecules which moved beyond walls to be in the cell, and calculate forces?

The quick answer is yes, forces will be calculated. The thing with automatic PME grid calculations is that the grid will be resized in accordance with your cell's size. Therefore, it will always contain your entire system in the calculations.

However, I believe you have somewhat misunderstood the concepts of periodic boundary conditions and NAMD's wrapping.

Since you are setting PBC options, the simulation itself will always handle the atoms periodic. In other words, even if atoms move beyond the cell, they will "reappear" in a mirror image of your cell. That's what periodic boundary conditions do, they create "copies" of your system around the original cell, essentially mimicking the overall structure of a crystal (where, again, particles are oriented periodically). On the other hand, wrapping is basically an "output" option, i.e., the DCD and coordinate output files. Therefore, it doesn't really affect force calculations, it's affects what you see in the final system.

Wow, thanks you so much again !

I can now definitely understand the meaning of 'wrapall' option. It is only one of the output options !

Your consice explanation directly helps me so much, thank you !

Well said Fotis. I would add that one legitimate reason to use the wrapall is if the subject of the simulation moves a lot and the simulation time is very long. In that case, the coordinates can get soo large that namd runs into problems. But this is a specialized case.

Thank you for your comments !

Jae,

NAMD fix grid spacing at the beginning of simulation an keep this distance fixed during MD.

So , if your box shrink too much in an NPT run, you need to perform a short simulation of your solvent and get a rough estimate of its density and then box length(or from experiment). With the equilibrium box length set PME grid size to (Box Length/ grid spacing) . Then simply specify grid spacing,  NAMD authomatically would determine PMEGridSizeX ,Y,Z.

By the way , your problem has nothing to do with PBC and wrapping options.

Thanks! I forgot answering your reply.

Hello Jae!!

Good luck !!

Bye!!  :)  :)  :)