Chemical potential of ring molecules is not consistent

[skvaraj2]

Expected behavior

when calculating the chemical potential of ring molecules (e.g. the cyclohexane in the example folder) as a mixture of two identical components (e.g. species 1 - 200 molecules of cyclohexane, species 2 - 200 molecules of cyclohexane), the chemical potential of both species should be the same. This test works for each molecule in the test folder except the cyclohexane (presumably because of the ring fragment)

Current behavior

In the simple test, these chemical potentials differ significantly. For molecules without ring fragments, I did not notice any deviations. This disagreement of chemical potential differences disappears when the molecule is considered to be rigid.
These chemical potentials cannot be used for GCMC simulations.

Steps to reproduce

Steps to reproduce the behavior:

1. run NPT simulation with 2 identical species of molecule with ring fragment (species 1 - 200 molecules of cyclohexane, species 2 - 200 molecules of cyclohexane),)
2. At T=403 K and P=1 bar the chemical potentials from short runs are about 88.8 kJ/mol for species 1 and -56.6 kJ/mol for species 2

Possible solution (optional)

I suppose there is a problem with ring fragment energy. But I wasn't able the identify the origin of this problem.

Additional context

This "bug" complicates the simulations of molecules with very similar structure (e.g. chiral mixtures)

[emarinri]

Hello @skvaraj2, thank you very much for reporting this.

I am traveling at the moment and will not get a chance to take a look at this until a later in the week.

[skvaraj2]

This is interesting. If I set the system to be considered a ternary mixture of three identical species, the first two chemical potentials are sort of okay. But the third one is completely off the scale. To run the GCMC calculation, I can set the third chemical potential to an extremely high value (absolute value) and then I can reproduce the system density with 2 equally distributed species.