When one obtains the mean-field Hamiltonian of a (classical or quantum) spin system and then needs to find the mean-field parameters by minimizing the expectation value of the Hamiltonian, does one neglect the constant term in the mean-field Hamiltonian that depends on the mean-field paraters themselves? If yes, why is it possible?