# Lennard Jones units and pair_coef

Hello,

When units are defined as lj, all quantities are scaled with epsilon and sigma. But then what must be the unit of epsilon and sigma when parsed through the pair_coef command?

John

Hello,

When units are defined as lj, all quantities are scaled with epsilon and
sigma. But then what must be the unit of epsilon and sigma when parsed
through the pair_coef command?

they have no unit. that is the whole point.

axel.

The way I understand it, the quantities are not scaled to an epsilon and sigma you supply with pair_coeff. Rather, some arbitrary reference epsilon and sigma are defined and set to 1. All epsilons and sigmas you supply with pair_coeff are multiples of those reference energy and length scales.

This is also why you can also use e.g. a Morse potential in Lennard-Jones units.

Just like how in SI units you express lengths with respect to some arbitrary length scale (the metre), you now express lengths in terms of an arbitrary length scale sigma. The units only really mean something if you have multiple things of the same units. So if you have only one type of particle and a thermostat, then you can set either k_B T or the epsilon of your potential to 1 without loss of generality. For sigma, you already have as length scale the simulation box, whose size is also specified as multiples of this arbitrary reference sigma, which is not a sigma you put into pair_coeff.

I hope this clears it up and is not too confusing, it took a while for me before I understood it myself.

2014-11-21 14:26 GMT+01:00 John <[email protected]...>:

Hello,

When units are defined as lj, all quantities are scaled with epsilon and
sigma. But then what must be the unit of epsilon and sigma when parsed
through the pair_coef command?

They *are* the unit definitions. That's how reduced units work.

Kristof