Key points are not available for this paper at this time.
Are the electromagnetic scalar and vector potentials dispensable? Vaidman Phys. Rev. A 86, 040101(R) (2012) has suggested that local interactions of gauge-invariant quantities, e.g., magnetic torques, suffice for the description of all quantum electromagnetic phenomena. We analyze six thought experiments that challenge this suggestion. All of them have explanations in terms of local interactions of gauge-dependent quantities, and, in addition, some have explanations in terms of nonlocal interactions of gauge-invariant quantities. We claim, however, that two of our examples have no gauge-invariant formal description and that, in general, no local description can dispense with electromagnetic potentials.
Aharonov et al. (Mon,) studied this question.