Dear LAMMPS users ,

I want to calculate T (∂p / ∂T)ρ on my system ! Temperature multiply to [ pressure variation with respect to temperature variation at constant density] .

Which ensemble do you think is suitable for that ?

How can I make the density to be constant in my simulation?

Thanks for your attention.

Regards,

I forgot to tell you that I use NVT ensemble but I can’t get good results .

because of Pressure fluctuation sometimes (∂p / ∂T)ρ is positive and sometimes negative !

Regards,

What are you doing in LAMMPS?

Are you performing separate simulations at constant density, or running a single simulation and taking instantaneous values of p and T?

Jackson ,

I run a single simulation and taking instantaneous T and P .

In that case, you are not measuring the thermodynamic quantities T and

P, but rather the statistical mechanical fluctuations delT and delP.

The quantity <delP/delT> probably has some interesting properties, but

this is not the thermodynamic derivative you are looking for. The most

straightforward to get (dP/dT)_rho is to run two or more

*equilibrated* NVT simulations at different values of T and then use

the finite changes in adjacent (<P>,T) points to estimate dP/dT. The

optimal spacing of the points in T is determined by a trade off in

statistical errors in delta<P> and systematic errors due to curvature

d/dT(dP/dT).

Aidan