A BPHZ convergence proof in Euclidean position space

Two BPHZ convergence theorems are proved directly in Euclidean position space, without exponentiating the propagators, making use of the Cluster Convergence Theorem presented previously. The first theorem proves the absolute convergence of arbitrary BPHZ-renormalized Feynman diagrams, when counterterms are allowed for one-linereducible subdiagrams, as well as for one-line-irreducible subdiagrams. The second theorem proves the conditional convergence of arbitrary BPHZ-renormalized Feynman diagrams, when counterterms are allowed only for one-line-irreducible subdiagrams. Although the convergence in this case is only conditional, there is only one natural way to approach the limit, namely from propagators smoothly regularized at short distances, so that the integrations by parts needed to reach an absolutely convergent integrand can be carried out, without picking up short-distance surface terms. Neither theorem requires translation invariance, but the second theorem assumes a much weaker property, called “translation smoothness”. Both theorems allow the propagators in the counterterms to differ, at long distances, from the propagators in the direct terms. For massless theories, this makes it possible to eliminate all the long-distance divergences from the counterterms, without altering the propagators in the direct terms. Massless theories can thus be studied directly, without introducing a regulator mass and taking the limit as it tends to zero, and without infra-red subtractions.