Hybrid overlay of morse, coulombic, and repulsive core

Hello LAMMPS users,

If I want to run a simulation with the short-range morse potential, long range coulombic, and repulsive core of C/r^12.

I think I have down the Coloumbic and morse function - I can use hybrid/overlay:

pair_style hybrid/overlay morse 10.0 coul/long 10.0

How can I just add the C/r^12 term to create a repulsive core? I need this to model the interaction at high temperature and pressure.

Does anyone know which potential will add this term?

Thank you for your help.

Ben

How about using LJ potential with cutoff at 2^(1/6).

Hello LAMMPS users,

If I want to run a simulation with the short-range morse potential, long
range coulombic, and repulsive core of C/r^12.

I think I have down the Coloumbic and morse function - I can use
hybrid/overlay:

pair_style hybrid/overlay morse 10.0 coul/long 10.0

How can I just add the C/r^12 term to create a repulsive core? I need this
to model the interaction at high temperature and pressure.

Does anyone know which potential will add this term?

wouldn't it be simpler to just tabulate this potential?

axel

I thought about just using LJ potential, but then I would need to define epsilon and sigma. I just need a C/r^12 of which I have the numerical value of C so I could just input that parameter.

Ben

Axel, I have not tabulated a potential before. Do you think that would be easier? I am guessing there is no easy to way to just add on the C/r^12 term?

Ben

Axel, I have not tabulated a potential before. Do you think that would be
easier?

depends on how many atom types you have. the best way to learn working
with tables is to take a few of the existing input examples and try to
reproduce them with tables. there are some issues where you need to be
careful, if you are after high temperature, high pressure systems (not
sure if an r^^-12 term is very good for that actually, that one was
originally chosen for LJ because of computational efficiency as it
allows to avoid the (at that time more than now) expensive square
root).

in any case, you need to play around with the settings until you can
reproduce the analytic functions well.

I am guessing there is no easy to way to just add on the C/r^12
term?

you can use a table for that, too. but remember that with
hybrid/overlay, you have to walk the neighbor list multiple times thus
a table should be 3x faster.

axel.

p.s.: as a bonus, pair style table is fully supported in the GPU
package, whereas hybrid pair styles cannot use the gpu enabled
neighbor list builds.

Ok Axel, thanks.

I’ll give that a go.
I am just using the r^-12 term as I am just trying to reproduce the results of a paper and the potential they used had that term in it.

Ben

C = 4 epsilon / sigma^12

that’s all it takes.

Steve

C = 4 epsilon / sigma^12

that's all it takes.

yet, that will not give you a pure r**-12 term with lj/cut, even if
you set the cutoff at the minimum, it will always include the r**-6
term.

axel.