Off-shell behavior of relativistic NN effective interactions and charge symmetry breaking

We examine in detail the suggestion of Iqbal et al. for calculating the class-four charge symmetry breaking amplitude in n-p scattering. By simplifying to a model problem, we show explicitly that the approximation scheme is unreliable if a phenomenological, effective nucleon-nucleon T matrix is used. Our results have wider implications for observables calculated in relativistic impulse approximation calculations. They reinforce the observation made in the literature that the procedure of fitting only positive energy matrix elements can lead to an NN interaction whose off-shell behavior is incorrect.