Attempting to remove infinites from divergent series: Hardy will hardly help

Abstract

The consequences of adopting other definitions of the concepts of sum and convergence of a series are discussed in the light of historical and epistemological contexts. We show that some divergent series appearing in the context of renormalization methods cannot be assigned finite values in a form consistent with Hardy's axioms without at the same time equating one to zero, thus destroying the mathematical building. We show that if the replacements for the concept of sum of a series are required to be associative, to be invariant under finite permutations of the terms and dilution, further restrictions emerge. We finally discuss the epistemological costs of accepting these practices in the name of instrumentalism.