Enable const float_X as flag type

[n01r]

In order to include effective potential models for ionization it is necessary to give the soon-to-come flags protonNumber and neutronNumber a float type.

[psychocoderHPC]

In PMacc and PIConGPU all particle/species flags must be classes. The reason is the C++ restriction that it is not allowed to use float in templates.
Therefor it is not possible to wrap a float value in a class like it is possible with integral values (e.g. boost::mpl::integral_c<uint32_t,42>)

If no compile time calculations are needed it is possible to wrap a float with the following workaround:

alias(protonNumber);
struct IonProtons
{
  static const float_X value = 2.0;
};

typedef bmpl::vector<protonNumber<IonProtons>, ... > IonFlagList;

[n01r]

As for why it can be non-integer and should therefore be float:

http://www.webelements.com/periodicity/eff_nuc_charge_clementi/

[ax3l]

thanks for looking up a reference! 👍

It's also explained in this wiki article.
(
E. Clementi and D.L.Raimondi, J. Chem. Phys. 1963, 38, 2686.
E. Clementi, D.L.Raimondi, and W.P. Reinhardt, J. Chem. Phys. 1967, 47, 1300.
)

[n01r]

I think with the introduction in #687 we can close this issue.