Support Exponential Hopping
Fakher (and others) approximate e^(k+V)
where k
is the hopping matrix and V
the HS interaction as e^k e^V
while in our code we do ~= 1+(k+V) ~= k + (1+V) ~= k + e^V
. It would be nice, for comparison's sake, to also have the fermion matrices with the exponential of the hopping matrix, rather than the linearized version.
(We also move k
to the on-time-diagonal, which would be best avoided for comparison purposes.)
This is not critical at all whatsoever.