I hope there is someone out there with a solid grasp of renormalization in quantum electrodynamics that can answer this question. It bothers me that the integrals that arise from loop Feynman diagrams involve a limit of infinite energy of virtual particles. But, according to the uncertainty principle a particle with infinite energy has a zero lifetime. It appears this consideration is not taken into account in these integrals. The position 4-vector integrals in the S matrix terms, evaluated before the 4-momentum integrals, are taken over all space and time. So how can the 4-momentum integrals take the uncertainty principle into account?