The standard approach to quantum field theory is perturbative: expand the path integral around a free Gaussian, generate Feynman diagrams, compute. For electrodynamics and the electroweak theory this works extraordinarily well. For quantum chromodynamics it works in the high-energy regime where the coupling is small. For quantum gravity it is a disaster.
The Einstein–Hilbert action expanded around flat spacetime gives an infinite series of interaction vertices, each contributing Feynman diagrams whose number grows factorially with loop order. At one loop, the graviton self-energy requires evaluating tens of thousands of diagrams. At two loops, hundreds of thousands. The series is not merely difficult to compute; it is not Borel summable, meaning the perturbative expansion does not converge to the physical answer even in principle. This is not a technical deficiency to be overcome by better computers. It is a signal that the perturbative, diagram-based language is not the natural language of quantum gravity.
The Natural Language
The celestial sphere perspective suggests a different basis. A scattering amplitude is not a function of momenta but of conformal primaries on the two-sphere at null infinity. In this basis, the amplitude is a correlator of a two-dimensional conformal field theory — and conformal field theory has powerful non-perturbative tools: the operator product expansion, crossing symmetry, the conformal bootstrap.
More concretely, the shadow discontinuity mechanism replaces Feynman diagrams entirely at loop level. A one-loop amplitude is obtained not by summing diagrams but by computing a single discontinuity:
This is not an approximation. It is exact. The shadow pole at Δ1 + Δ2 = 2 encodes the exchange of a virtual particle and its time-reversed image across the T boundary — which is precisely what a loop is. The optical theorem, reread at the celestial boundary, is the loop formula.
The Speedup
For four-graviton scattering at one loop, the conventional approach requires evaluating on the order of 105 Feynman diagrams after gauge fixing and ghost contributions. The shadow discontinuity method requires a single discontinuity of a six-point tree amplitude. The tree amplitude is known in closed form from the Penrose–Ward correspondence. The speedup is not a matter of clever programming — it is structural.
At L loops the conventional count grows as (2L)! while the shadow method requires an L-fold discontinuity of a fixed tree amplitude. The advantage becomes astronomical at high loop order. This suggests that the extraordinary difficulty of quantum gravity calculations in the conventional approach is not an intrinsic property of the theory but an artefact of the wrong choice of basis.
What This Means
The implication is not that perturbation theory is wrong. It is that the natural expansion parameter for quantum gravity is not the coupling constant but the loop order in the shadow discontinuity expansion — and the natural basis functions are not plane waves but conformal primaries on the celestial sphere.
Feynman's diagrams are a monument to physical intuition. But intuition developed for a theory of pointlike particles in Minkowski space may not be the right guide for a theory of gravitons whose natural home is the two-sphere at the boundary of spacetime.
Loading…