Reagrding Computation of betaP from hoomd.hpmc.compute.sdf

[hrithiknandy]

Dear Users,

I am trying to find pressure of a hard particle system using hoomd.hpmc.compute.sdf. Here I considered a system which had reached equilibrium using NPT simulation at a constant pressure betaP=40. But when I am running NVT simulation with that equilibrium state as my initial state and logging the betap , it is deviating too much from 40 . I must mention the packing fraction of my system is around 0.70. For relatively low pf (around 0.44 to .60) for the same system I am having correct values of betaP (with which I ran NPT and equilibrate the system). I tried to set xmax and dx such that the area under the s_comp(x_k) is near 0.5 as documented. It seems like betaP is deviating for higher pf for my case.

Could you kindly provide some suggestions regarding this issue?

Thanks

[joaander]

At constant pressure, the density fluctuates. At constant volume, the pressure fluctuates. To make a meaningful comparison, you need to run many replicate simulations and compute mean values of these quantities along with the standard error of the mean. For examples, see the tests that I use to validate HOOMD-blue: https://github.com/glotzerlab/hoomd-validation/

Should you find that the sampled mean pressure in NVT is more than several standard errors of the mean different than the set pressure, then your system is most likely not in equilibrium. This is very likely at high packing fractions where local trial moves are no longer ergodic (also known as kinetically arrested).