how to choose reduced time-step

I want to carry out simulations in LAMMPS for a metallic system.

I am confused as to how to choose the reduced time step.
The unit of time, tau, is tau=sqrt(masssigmasigma/epsilon).
If I want a time-step of 1 femto-second, I will get
different values of the reduced time-step depending on different
values of tau.

For, example, if I choose mass=1 amu, sigma =1 Angstrom
and epsilon =1 ev, I get a reduced time-step 0.1 for 1 femtosecond.

However, if I choose the actual mass of the metal (copper), i.e
63.546 amu, and if I chose sigma to be the value at which the potential
crosses the zero axis, i.e. about 2 Angstrom, and epsilon to be
about 0.1ev, then I get a much lower value for the reduced time-step,
which corresponds to 1 femtosecond.

Is there any rule on how to choose the reduced time-step.

Best Regards,
Ronald

Reduced units are used to do material independent simulations. In the case of a Lennard-Jones potential for example you can use a single equation of state when you work with reduced units and by plugging in a specific material parameters you’ll back out the actual values. In the first case you get a tau that is ~10 femtosecond but in the second you’ll get a ~510 femtosecond.
Usually you choose m, sigma and epsilon to be one for your reference material then by using actual parameters you can estimate your tau and adjust the timestep accordingly e.g. 0.1tau for the first case or 0.002tau for the second one if you need a 1femtosecond time stepping in real units.

HTH,

Kasra.

Dear Kasra,
sorry for the delay in response. Many thanks for your help. It is indeed very helpful.

Ronald