A proposal for new Hamiltonian constructor

[mariya-nasution]

Currently, we must pass a list of pseudopotentials to Hamiltonian constructor as positional parameters:

Ham = Hamiltonian( atoms, pspfiles, ecutwfc )

If we want to use all-electrons calculation, we use the following constructor (with pspfiles removed):

Ham = Hamiltonian( atoms, ecutwfc )

I think it is good to use ecutwfc as the first positional argument and use keyword parameters to specify other arguments. I also think it is very convenient to have keyword pspset = "GTH-PADE" or similar and let the constructor choose pseudopotential based on list of atomic species.

Ham = Hamiltonian( ecutwfc, atoms, pspset="GTH-PADE" )

Having to specify pspfiles manually while we have a standardized set of pseudopotentials is rather tedious for me.

[f-fathurrahman]

I think atoms::Atoms will still be the first positional argument.

Meanwhile ecutwfc might become keyword argument as it is only particular to plane wave basis set. This will apparent when we extend Hamiltonian to other basis sets.

I also think that using pspset keyword is a good idea.