Fix/halfmode
This MR deals with a couple of bugs in the halfmode formulation. First, in the current implementation the two PhysDeriv functions are using opposite signs, and therefore are not consistent. Also, all fields are being defined as real, when w should be imaginary.