Hi, everyone

I have optimized a Ca2SiO4 unit cell with Gulp, but after I converted it into Lammps, the minimize command gave me high pressure value of 100000 atomasphere for the unit cell. Even if I perform a NPT simulation for a while, the pressure is still high. I have check my input file for many times, all the pamarameters are converted correctly form Gulp to Lammps, but I don’t understand why Lammps’s results are so different from that of Gulp. Did I miss something? Thanks.

best,

Lei

Do you get an energy with GULP? If so, is the same

energy you get with LAMMPS for the identical snapshot?

If not, then the force fields in the 2 codes are not

setup identically and are not giving the same answer.

If they are not, then forces will not be the same,

and neither pressure. Does GULP give a pressure?

Again is it the same as LAMMPS computes for the

same snapshot?

Steve

Hi, everyone

I have optimized a Ca2SiO4 unit cell with Gulp, but after I converted it

into Lammps, the minimize command gave me high pressure value of 100000

atomasphere for the unit cell. Even if I perform a NPT simulation for a

while, the pressure is still high. I have check my input file for many

times, all the pamarameters are converted correctly form Gulp to Lammps, but

I don't understand why Lammps's results are so different from that of Gulp.

Did I miss something? Thanks.

how about a bad conversion of the geometry (including box dimensions)

leading to close or overlapping atoms.

also, are you sure that the units are properly user?

axel.

Hi, Steve

The force field I use is a simple lj/cut/coul/cut pair style. I have converted the lennard jone parameters correctly from GULP to LAMMPS considering the different unit and formular used in them. GULP doesn’t give pressure directly, but I calculated it myself with the dU/dε formular after applying a strain to the unit cell by changing the cell parameter a little. GULP would give a near-zero stress in each direction for the unit cell at equilibrium.

I played with cutoff parameter for lj/cut/coul/cut in LAMMPS, finding that the energy and pressure value are very sensative to it, which is same for a supercell structure. If I decrease the cutoff to a value of 2.6 Angstrom that is slightly bigger than the bond length like O-Si, O-Ca in my system, the pressure would approach zero. Then the total potential energy I got in LAMMPS for the unit cell is -9852 Kcal/mole, while the Total lattice energy in GULP is -249 eV (-5748.9 Kcal/mole). If the cutoff parameter is chosen as a bigger value of 3.5 Angstrom (the cell parameter are 7.21 5.88 9.25), the total potential energy is changed to -5669 Kcal/mole that is close to GULP, but the pressure in each direction will be 3~4 GPa.

best,

Lei

Hi, axel

Both in GULP and LAMMPS, after energy minimization of the unit cell, the cell parameters and atoms’ positions changed a little. And I have converted the lennard jone parameters from GULP to LAMMPS considering the different unit and formular used in them. I found that by playing with cutoff paramter in LAMMPS, the total potential energy could match the total lattice energy in GULP (with a cutoff of 3.5), or the pressure could approach zero (with a cutoff of 2.6). The results are very sensative to the cutoff. But either 2.6 or 3.5 is small since for lennard jones and coulomb normally the cutoff is around 10 for a supercell. If the unit cell is replicated into a 3*3*3 supercell, changing the cutoff from 2.6 to 10 would have a big effect on the energy and pressure result.

best,

Lei