Non-perturbative spectral gravity measure in the Hilbert-Schmidt Gaussian completion: pro-torsor structure and the obstruction to canonical expectations

Assumes 1–2 courses of background. Domain terms may appear without definition.

This paper tackles a fundamental problem in quantum gravity: how to define a well-behaved path integral for spacetime geometries using the spectral action—a function of the Dirac operator that encodes the geometry. The naive path integral weight, \(\exp(-\operatorname{Tr} f(D^2/\Lambda^2))\), is not “coercive” and gives a divergent integral, so the authors cure this by adding a Gaussian reference measure that tames the fluctuations. This yields a finite, well-defined measure for each fixed background geometry (a “sector”), and these sectorial measures are compatible as you increase the resolution of the calculation. However, a deep obstruction appears: measures coming from different backgrounds are mutually singular—they live on completely different sets of configurations—so they cannot be combined into a single, background-independent probability measure. The global structure that emerges is a “pro-torsor,” a collection of local measure classes that are linked by symmetries but lack a canonical overall choice. The authors prove a no-go theorem: without adding extra structure, there is no way to pick a distinguished, scalar probability measure from this collection—any attempt either reduces to a tautology or fails. Remarkably, despite this obstruction, the one-loop predictions (such as those for particle interactions) are shown to be universal, independent of which Gaussian reference is used. In essence, the work rigorously identifies a fundamental limitation of constructing a background-independent quantum gravity path integral in this framework, while preserving the testable low-energy results.

AI-generated (deepseek-v4-flash) · created 2026-05-19