I am running MD simulations using the OPLS-AA force field. I created my system using gmx insert-molecules to pack solvent and salt molecules into a simulation box. Energy minimization (EM) converges normally:
Fmax < 100
Potential energy ~ -5.9 × 10⁴ kJ/mol
No warnings or instabilities
However, when I run NVT equilibration, I observe visible empty regions / low-density pockets in the box, which do not look physical. These regions appear only during NVT, but after NPT equilibration the system becomes fully uniform and the density looks normal.
My questions are:
Why do these empty spaces appear during NVT?
Is it normal when using gmx insert-molecules + OPLS-AA?
Are these voids simply due to the fixed box volume during NVT before the system reaches the correct density?
Is there a recommended way to avoid these visible voids during NVT, or are they harmless as long as NPT fixes them?
Do these voids indicate any issue with force-field parameters or system preparation?
Any references or explanations would be very helpful. Thank you!
If you used official parameters of OPLS-AA then I would tend to exclude any force field problem, but this is most likely due to your point 3, that is, the fixed volume is generating these artifacts.
The NPT should fix any void bubble. Since this is a simple solvent/salt simulations, I would not expect problems later on due to this happening during NVT (while if you have bilayers/proteins etc then things are more delicate). In general my understanding is that gmx insert-molecules approximates very roughly the filling of the space, so this behaviour is kind of expected. For sanity check, take a look at the density of you box and see how far you are from the experimental values!
Yes, I’m working with salt/ solvents. I’m just asking as I have fixed volume and using force fields from LIGPARGEN. After EM, these are the results of nvt, but after npt, it gets fixed because of the density and volume changes… is there any way to avoid these empty spaces??
NPT simulation or start with a better estimate of the density, that is, insert more molecules in the starting configuration. If you plan is to run NVT I would anyway still run EM > NVT > NPT to relax the box and then revert to NVT once the volume stabilized.
i calculated molecules according to box size, cant increase or decrease by myself. how do you do simulation, EM>NVT>NPT>PRODUCTION NVT?? or by which way??
You can change the number of molecules to insert with the -nmol flag of gmx insert-molecules. Anyway, you should check the final density of your box and compare it with the closest experimental setup you can find, even the pure solvent, just to eyeball any more serious cause of the empty bubbles you see.
One of the most standard simulations protocols is to start with energy minimization, then thermalize in NVT, and then move to NPT (in case NPT is the final ensemble that one wants to simulate). You said you did energy minimization followed by NVT. You can now switch to NPT and relax the pressure/box dimensions. If you intend to run in NVT you can then switch back once you stabilize the volume/density. Starting from NVT and not equilibrating the box is very dangerous if your starting densities are very much off (like your case seems to be).
Yes, i did it. em,nvt, npt and production run via nvt ensemble. But when we do not system relaxes and become more realistic…i want to know, do these empty spaces effects the system?
I do not fully understand your question, but if you mean that the presence of the bubble affects the results with respect to another NVT simulations where first you relaxed with an NPT simulations then yes, in principle it does. You can already see from your snapshot that the system is most likely collapsing on itself and trying to maximize the intermolecular interactions. I would expect this to have major repercussions on properties such as a radial distribution function, or the diffusion coefficient. The bubble you have seems enormous also, so I really thing the starting volume was enormous with respect to the expected density of your solute at that temperature!
actually i started with EM, then i did nvt and this is the snapshot of nvt, but when i did npt after the nvt, the system becomes more realistic and normal. so thats what i’m asking like in nvt its occurrence is okay?
