Dielectric constant

[Anthoni-Alcaraz]

Dear all,

I was checking the compute for the dielectric constant and I figure it out that implementation is only for a epsilon where ε_{Reaction Field} =∞. According to Neumann theory (1983), this way to compute is missing +1. I think for large values like in liquid water is a good approximation.

https://github.com/leeping/forcebalance/blob/master/src/data/npt.py#L686:

However, what about other values? For example when ε_RF tends to 1 or another value.

I think we can add more cases without a lot of work. For example, dipole fluctuations could come from calc_eps0 function in a general way like this:
dipoleFluctuations = calc_eps0(None,**{'d_':Dips, 'v_':V})/3.0
or modify calc_eps0 function to something like this: name and return.

dipoleFluctuations = calc_dipFluc(None,**{'d_':Dips, 'v_':V})
where return prefactor*D2/bzavg(v_,b)/T/3.0

Thus we can add three cases: ε_RF = ∞, ε_RF = 1 and ε_RF = ε .

Case 1: Eps0 = dipoleFluctuations*3.0 + 1.0
Case 2: Eps0 = (dipoleFluctuations*2.0 + 1.0)/(1.0-dipoleFluctuations)
Case 3: .....

And this line calls the case selected:
https://github.com/leeping/forcebalance/blob/master/src/data/npt.py#L691

But I don't know if Eps0_err will change and analytic derivative too.
https://github.com/leeping/forcebalance/blob/master/src/data/npt.py#L693
https://github.com/leeping/forcebalance/blob/master/src/data/npt.py#L703

Thank you in advance,

Anthoni