This paper addresses a core problem in one approach to quantum gravity. The spin foam program builds spacetime amplitudes by summing over discrete pieces of geometry, usually given by triangulations. The hard question is how to remove that discrete regulator and recover a true continuum theory. The authors study this problem in a model-independent, structural way. They introduce an axiomatic framework inspired by Atiyah’s Topological Quantum Field Theory (TQFT) axioms to make precise what a continuum limit would mean for spin foams.

At the discrete level the authors assign a Hilbert space to each triangulated spatial boundary and a vector (an amplitude) to each bulk triangulation that fills a boundary. They organize the set of triangulations with natural orderings so one can speak of refining them and taking limits. This mirrors TQFT ideas, but with a key difference: in standard TQFT the cylinder over a spatial slice gives an amplitude that acts as the identity on the boundary Hilbert space. In spin foam truncations no such identity appears at finite triangulation. The paper asks which notion of convergence of these truncated amplitudes would produce a sensible continuum theory.

A striking result they prove is a no-go theorem: if one requires a sufficiently strong form of convergence, the only possible continuum limit is a topological theory. In plain terms, too-strong convergence collapses the theory so that it loses the infinite number of degrees of freedom expected of gravity and behaves like a TQFT, which depends only on topology and not on local geometry. This shows that the usual, strongest convergence criteria are incompatible with recovering a non-topological gravitational theory from spin foam sums.

To avoid this obstruction the authors propose a weaker notion of limit. They adopt a distributional point of view, motivated by Refined Algebraic Quantisation (RAQ). In this picture the amplitude of the cylinder does not become the identity but instead defines a rigging map. A rigging map is a generalized projection that builds the physical Hilbert space from kinematical states when ordinary projectors do not exist. With this assumption the continuum amplitudes act as well-defined distributions on the resulting space of physical states. Thus the spin foam path integral is given a clear mathematical role: it produces the projection onto solutions of the constraints and yields physical inner products in a precise sense.