Should we rescale the density for final NVT run?

[ltalirz]

@Gitdowski pointed out that density fluctuates in NpT and we simply take the last NpT structure for the final NVT run.

One could in principle average over a typical fluctuation time period and rescale coordinates accordingly

Thoughts @Atilaac ?

[Atilaac]

That could be done. Although at low temperatures, it has minimal or no effect on the results/properties.

The density fluctuation almost disappears for large systems. We can perform tests to assess the intensity of these fluctuations and include them in the parametric study as a proposed sample size for negligible density fluctuations.

I am suggesting doing it this way because we have a glass, which means the structure is already heterogeneous, and the rescaling will be affine, so artifacts may be introduced if the scaling factor is large.

[ltalirz]

Thanks for the quick reply, @Atilaac !

The density fluctuation almost disappears for large systems. We can perform tests to assess the intensity of these fluctuations and include them in the parametric study as a proposed sample size for negligible density fluctuations.

Yes, let's do that and decide based on the outcome