[lammps-users] How to combine tabulated potential with Ewald summation

Dear LAMMPS users:

Thank you all for your previous help. I finally get reasonable result about thermal expansion of uranium dioxide. But I am troubled again while taking the next step.

My question can be described as how to combine tabulated potential with Ewald summation technique for long range and short range interactions.

Previously, I tabulated the customized potential with inner cutoff distance 1.2 Angstrom and with columbic term excluded. For pair style command I used “pair_style hybrid/overlay table linear 1200 coul/long 10” so that the columbic term can be added back to the tabulated potential and in the mean while the damping parameter for Ewald summation technique could be generated by LAMMPS. Using this strategy to combine tabulated potential with Ewald summation seems work, because the Lattice parameter of uranium dioxide as a function of temperature is in reasonable agreement with the published result.

For the next step, I attempt to extend the potential file to cover the distance smaller than 1.2 Angstrom, which is required to simulate radiation induced displacement cascade. According to the literature, people usually use ZBL potential for this type of short range interactions (0.01~0.5Angstrom). I used fifth order polynomial to smoothly connect ZBL potential and the customized potential during the range of 0.5 to 1.2 Angstrom.

Now, my problem is I can’t use the previous strategy to combine tabulated potential with Ewald summation scheme because there is no columbic term that I can exclude for the short range (<1.2Angstrom). But I tried to use the same method anyway, just deducting the value of columbic energy from tabulated potential. So the new tabulated potential keeps all the value beyond 1.2 Angstrom but with update to cover shorter distance. Interesting thing is simulation result using new potential file is completely mess even though all others in the input file for thermal expansion simulation are remained same.

In my opinion, old and new potential files should give the same result, because inter-atomic distance should not be possible to be smaller than 1.2Angstrom and two files are identical for the range greater than 1.2Angstrom. In addition, if the distance is smaller than 1.2Angstrom, LAMMPS would have no potential data to refer to if the old potential is used.

I wonder if anybody could give me a hint about solving my problem. Any suggestion or comment would be deeply appreciated.

Thank you all for your time and help.

Bin Wu

Have you printed out (via pair_write) and plotted the tabulated
potential LAMMPS is using after it did a fit to your input table?
Splines are tricky and if you have kinks or mis-matches in
how you've joined different distance regions in your table then
the resulting fit may have oscillations or other undesirable
features. Always use pair_write to high precision (lots of points)
and look at the plot carefully.