Solid-Liquid coexistence + fluctuations

Dear Aidan,

I have a question regarding the fluctuations when calculating melting point using solid-liquid coexistence. I calculated the melting point of silicon given by Tersoff potential, and I could get correct result (by comparing the result with those calculated by Morris et. al). However, when i run my script for another potential function to calculate melting point of silicon, I cannot get correct result. My computed melting temperature is lower (up to 300 K) than that computed by the developers.

I wonder if you could tell me whether this deference comes from the fluctuations of a specific potential?

Best regards,

Different Si potentials give different melting points. Some are more
accurate than others. The melting temperature satisfies the
thermodynamic equilibrium condition T = (PE_l - PE_s)/(S_l - S_s)
where PE is average potential energy, S is the entropy (can not be
directly measured), and the subscripts l and s refer to liquid and
solid at the same temperature and pressure. Since potentials are
usually quite accurate for PE_s and S_s, the different melting points
are usually due to variations in the values of PE_l and S_l.

Also, for potentials that exhibit very different melting temperatures,
care should be taken to avoid unexpected phases changes, either to a
different solid phase, a glass, or a vapor. Finally, it is important
to verify that both phases are at the same pressure, with no shear