Energy Vs. Temperature


I started using LAMMPS for my research at the start of this year, and doing energy minimization at 0k has worked well. I am currently moving to doing some heating simulations and I want to look at the Energy vs. Temperature to see phase changes. My understanding of looking through the calculations, the temperature will be calculated via the total kinetic energy per degree of freedom. If I am looking at the calculated etotal vs. temperature, one is inferred from the other so the straight line E vs. T graphs make sense as they are being calculated from each other. This would not allow me to see a phase change in an E vs T graph. My question is, is there a way to get temperature or KE values that are not inferred from one another?

A little background into what I am running;
20 carbon atom “isolated-cluster”
heating using nvt from 1 - 3500 K
LCBOP potential used

Temperature is not computed from etotal, but from ke (the kinetic energy).

However, the problem with phase transitions from MD simulations is that most are activated processes requiring suitable nucleation which in turn requires compatible fluctuations, which are limited by finite size, short simulation time, and idealized composition of typical MD systems. So you would have a massive hysteresis when looking at those graphs compared to experiments.