[lammps-users] why by simulation is becoming bigger

Hi all,

i have read in a previous thread that someone can find data from http://webbook.nist.gov/chemistry/ for densities of several sites. I compute mass density for specific values of temperature and pressure. I transformed it to number density and then choosing arbitrary the number of molecules and finally through number density i calculate the volume. Now i use a npt integrator and all of a sudden i see after the visualization that my simulation box is becoming bigger. Assuming that my calculations are correct why is this happening? From a physical point of view.
As a note i put the molecules randomly in the box. In several papers i read bulk density or bulk water for example.Is there any difference in bulk and in the usual density let’s say for a liquid or a gas?

Thank you in advance
Nelson

There are any number of reasons this could be happening:

(1) The units aren’t consistent.
(2) The number of molecules you’re using is too small to determine an accurate result.
(3) The potential you’re using isn’t accurate enough to predict the bulk density accurately.
(4) You’ve made a mistake in setting up the simulation—charges or pair potentials are incorrect, etc.
(5) The long-range interactions have not been correctly handled.

Remember that potentials do not reproduce all of the properties of a real material with perfect accuracy. They only reflect the attempt to get as close as possible on many different thermodynamic parameters (sometimes for one material, sometimes for a whole class of materials). In addition, potentials are sensitive to the conditions in which they were parameterized. For instance, errors in density of several percent are entirely possible depending on the value of the cutoff parameter for the potential and electrostatics.

–AEI

Try equilibrating with a langevin thermostat for a hundred picoseconds or so before using NPT. If you use a random configuration to start NPT the box will explode.
Matt

Quoting Nelson Jordan <[email protected]>:

i used the lj potential , long range coulomb terms. the number was ranging from 5000-10000 molecules. you think it is better to use something else or "play " with the cuttof distances?
You mentioned only bulk density. is there any difference with the usual density?

I have minimized my system and then equilibrate it for 100ps and then i use the npt.

i used the lj potential , long range coulomb terms. the number was ranging from 5000-10000 molecules. you think it is better to use something else or "play " with the cuttof distances?

OK, so it’s not likely to be a system-size effect. The question then is what is the potential that you used? Where did it come from? Did the authors of the original potential include the density in their measurements? If so, how did it compare? What values for the cutoff did they use? Are your conditions similar to theirs?

(A lot of potentials don’t reproduce the density accurately; this is particularly true of the many water models out there.)

You mentioned only bulk density. is there any difference with the usual density?

No. Bulk density is the “usual” density.

–AEI

The question then is what is the potential that you used? Where did it come from? Did the authors of the original potential include the density in their measurements? If so, how did it compare? What values for the cutoff did they use? Are your conditions similar to theirs?

I haven’t any paper to compare it with. My main question is why with the given data from nist webbook why the box is becoming bigger when it was suposed to be near the initial values even with an npt. so i guess i have to change something in potentials. ?Because the coefficients are taken form papers.

(A lot of potentials don’t reproduce the density accurately; this is particularly true of the many water models out there.)

You mentioned only bulk density. is there any difference with the usual density?

No. Bulk density is the “usual” density.

–AEI

Okay, this doesn’t make any sense. If you’ve taken the coefficients from papers, then someone has to have done the comparison with experimental results—that’s how they choose the parameters in the first place! If the papers you’re citing in turn cite other papers, then perhaps those papers have the data you’re looking for. But it has to be somewhere.

At any rate, based on what you’re saying, it’s most likely one of three situations:

(1) If the box is increasing just a tiny amount, then it probably means that the potential, as it’s parameterized, doesn’t reproduce the density exactly. (It happens; simulations almost never agree perfectly with experimental results.)

(2) Have the oscillations in the volume damped down? If not, then you should wait for the changes to become small before determining how much the box has grown (if any).

(3) If the box has relaxed, and the increase in size is large, that means there’s probably an error in your script.

–AEI

If you send us your input file we can be more help.
Matt

Quoting "Ahmed E. Ismail" <[email protected]>:

dimension 3
boundary p p p
units real
neighbor 0.3 bin
neigh_modify delay 4 every 1
atom_style full
bond_style harmonic
angle_style harmonic
pair_style lj/cut/coul/long 8 10
pair_modify mix arithmetic
kspace_style pppm 1e-4
read_restart data_restart.restart
pair_coeff 1 1 0.1569 3.05
pair_coeff 2 2 0.0536 2.8
bond_coeff 1 986 1.16
angle_coeff 1 10000 180.0
fix 1 all shake 1e-4 400 0 t 1 2
fix 2 all npt temp 400 400 100.0 iso 2.0 2.0 1000.0
#equilibration
timestep 0.5
run 200000
thermo 1000
log produc.log
thermo_style custom etotal ke temp pe press vol
thermo_modify line multi
timestep 1.0
dump 1 all atom 1000 Post.lammpstrj
dump_modify 1 image yes
run 1000000

this is the simulation script.there is one more with the minimization but with or witout it i have the same results

Lammps_input.data (1.09 MB)

Is the atom type 2 the hydrogen atom? I'm not familiar with this water model. Where did you get the parameters? What is the name of the model? Was it designed to use at 400K?
Matt

Quoting Nelson Jordan <[email protected]>:

no it is CO2 model. 3 site model by Pottof

If you run NPT and the box becomes bigger, the reason is
simple. The pressure is larger than your target pressure,
so the box expands to reduce the pressure. So the real
question is why you have a large pressure. If you have
random atoms, there will be overlaps and thus a high pressure,
at least initially.

Steve