Existence and summability of the renormalized perturbation series of four-dimensional quantum electrodynamics (the surviving core of Dyson's 1949 convergence question)

In his 1949 paper establishing the renormalized S-matrix perturbation theory of quantum electrodynamics (QED), F. J. Dyson explicitly set aside “the problems of extending the treatment to include bound-state phenomena, and of proving the convergence of the theory as the order of perturbation itself tends to infinity”. Dyson's paper acknowledges six limitations; five of the six — bound states, infrared divergences, overlapping-divergence subtractions, the gauge-invariant photon-mass constant, and explicit fourth-order checks — were subsequently resolved by the Bethe–Salpeter/NRQED bound-state program, the Bloch–Nordsieck/Yennie–Frautschi–Suura/Kinoshita–Lee–Nauenberg infrared formalism, the Bogoliubov–Parasiuk–Hepp–Zimmermann renormalization theorem, the Ward–Takahashi identity, and explicit multi-loop calculations through tenth order. The surviving open core is the convergence question itself, sharpened by 75 years of partial results: Dyson's 1952 argument indicates the renormalized series in the fine-structure constant α diverges, and local Borel summability of the Schwinger-function series has been proven (Feldman–Hurd–Rosen–Wright, 1988); but it is not known whether the full series is Borel summable to well-defined observables, whether its expected renormalon/Landau-pole ambiguities cancel, or whether any exact nontrivial continuum four-dimensional QED exists of which the series is the asymptotic expansion. The problem is to settle these questions rigorously, in either direction: construct the non-perturbative theory, or prove that only a trivial (zero-charge) or ill-defined continuum limit exists.

Background

Quantum electrodynamics, the quantum field theory of electrons, positrons, and photons, describes scattering through the S-matrix — the unitary operator mapping freely prepared incoming states to outgoing states at asymptotic times. Dyson's 1949 paper synthesized the contemporaneous covariant formulations of Tomonaga, Schwinger, and Feynman into general diagrammatic rules for computing S-matrix elements as a power series in the fine-structure constant α = e²/(4πħc) ≈ 1/137, and showed that the ultraviolet (short-distance) divergences at every order can be absorbed into mass and charge renormalization — the re-expression of formally infinite bare parameters in terms of finite observed ones. The n-th order of the series (the “2n-loop” order in the modern idiom, a loop being one closed cycle of virtual particles in a Feynman diagram) receives contributions from factorially many diagrams.

Dyson's paper closed with an explicit list of what it did not do. Its abstract states: “Not considered in this paper are the problems of extending the treatment to include bound-state phenomena, and of proving the convergence of the theory as the order of perturbation itself tends to infinity.” The body adds that the expansion in powers of an external potential fails for bound states and low-energy scattering; that infrared divergences (divergences from arbitrarily soft photons emitted in scattering) are merely avoided by a cutoff λ, with Bloch–Nordsieck-type treatments deferred; that the proof that overlapping divergences (divergent subgraphs neither nested in nor disjoint from one another) multiply finite parts by the same renormalization constants Z rests on “a detailed argument too long to be given here”; that the photon mass-type constant A′ is set to zero by appeal to gauge invariance, verified only at lowest order; and that the whole finiteness proof “makes no pretence of being complete and rigorous”, asking to be “supplemented by an explicit calculation of at least one fourth-order radiative effect, to make sure that no unforeseen difficulties arise in that order.”

Five of these six limitations are textbook-settled. Bound states: Gell-Mann and Low, and Salpeter and Bethe (both 1951), extracted the relativistic two-body bound-state equation (the Bethe–Salpeter equation) directly from the S-matrix formalism; effective field theories such as NRQED (nonrelativistic QED, the systematic low-energy effective-Lagrangian framework of Caswell and Lepage) turn bound-state corrections into a controlled expansion; the review of Eides, Grotch, and Shelyuto documents agreement between bound-state QED and hydrogenlike spectroscopy at the limits of experimental precision. Infrared divergences: they cancel in inclusive (soft-photon-summed) cross sections (Bloch and Nordsieck; the systematic exponentiation of Yennie, Frautschi, and Suura; the theorems of Kinoshita and of Lee and Nauenberg on sums over degenerate states; the charged asymptotic states of Kulish and Faddeev). Overlapping divergences: the Bogoliubov–Parasiuk renormalization theorem, proved by Hepp and completed by Zimmermann's forest formula, gives a rigorous all-orders subtraction scheme. Photon mass: the Ward–Takahashi identity (the all-orders relation between the electron-photon vertex and the electron self-energy, Ward 1950, Takahashi 1957) forces the vacuum-polarization tensor to be transverse, so no photon-mass term survives renormalization at any order. Fourth-order checks: Sommerfield's fourth-order computation of the electron anomalous magnetic moment g−2 (the deviation of the electron's magnetic moment from the Dirac value) found no unforeseen difficulty, and the program has since been carried through tenth order by Aoyama, Hayakawa, Kinoshita, and Nio.

The sixth item — convergence — survives, transformed. A power series $\sum_n a_n\alpha^n$ is asymptotic to a function $F(\alpha)$ if the remainders after $n$ terms vanish faster than $\alpha^n$ as $\alpha\to 0$, even when the series diverges; a divergent asymptotic series can still determine $F$ uniquely if it is Borel summable: the Borel transform $B(t)=\sum_n a_n t^n/n!$ must converge, be analytically continuable along the positive axis with suitable growth, and then

$$F(\alpha)=\int_0^\infty e^{-t}\,B(\alpha t)\,dt.$$

Dyson himself (1952) argued that the QED series cannot converge: if F(e²) were analytic at e²=0, the theory would also be well defined for e²<0, where like charges attract and the vacuum disintegrates by spontaneous pair creation; hence a singularity at the origin and zero radius of convergence. Dyson stated the argument was “lacking in mathematical rigor”. Later physics analysis sharpened the expectation: the running of the coupling (the positive QED beta function, dα/d ln μ > 0, which drives the effective charge α(μ) to a pole — the Landau pole — at astronomically high energies) generates factorially growing coefficients with an expected singularity of the Borel transform on the positive axis — the mechanism whose counterpart in quantum chromodynamics (QCD, the strong-interaction gauge theory) is the renormalon analysis reviewed by Beneke — so the series is expected not to be Borel summable, with a nonperturbative ambiguity of relative order exp(−const/α).

On the rigorous side, constructive quantum field theory — the program that constructs interacting quantum field theories as mathematically defined objects, for example satisfying the Osterwalder–Schrader axioms (Euclidean axioms whose verification yields a Wightman theory in Minkowski space; Schwinger functions are the Euclidean correlation functions) — has succeeded only in spacetime dimension below four and in two-dimensional asymptotically free models; no interacting four-dimensional theory has ever been constructed, and the sibling problem for pure Yang–Mills theory carries a Clay Millennium Prize (as the survey of Summers records). For QED specifically, Feldman, Hurd, Rosen, and Wright proved renormalizability to all orders and local Borel summability (analyticity of the Borel transform in a disk about the origin) of the Schwinger-function perturbation series; but, in Summers's words, “It is not known if there exists an exact model to whose Schwinger functions this series is asymptotic”, and although it is “widely believed that since QED is not asymptotically free in four spacetime dimensions it cannot be defined as a mathematically rigorous theory … this matter has not been settled.” Triviality here would mean that the cutoff-removed continuum limit forces the renormalized charge to zero, so the limit is a free theory — proven for the four-dimensional Ising/φ⁴ model by Aizenman and Duminil-Copin, but for no gauge theory. Recent physics proposals to evade the Landau pole — an interacting ultraviolet fixed point with a Pauli spin-field coupling (Gies and Ziebell), next-to-leading-order effective-field-theory corrections (Djukanovic, Gegelia, and Meißner), and a nonperturbative “everlasting-interaction” polarization summation (Evans and Rafelski) — provide evidence and model-building, but neither a construction of a continuum limit nor an impossibility proof.

Problem Statement

The problem is to settle, with mathematical rigor, the convergence question that Dyson's 1949 paper explicitly set aside — “proving the convergence of the theory as the order of perturbation itself tends to infinity” — for the renormalized perturbation series of four-dimensional quantum electrodynamics. Seventy-five years of partial results sharpen it into two coupled questions joined by one objective: to give the renormalized α-series a non-perturbative meaning, or to prove that it has none.

(1) Summability of the series. Determine the large-order behavior and summability status of the renormalized QED perturbation series for well-defined observables (for example, the Schwinger functions of ultraviolet-renormalized, infrared-regulated QED, or infrared-safe inclusive quantities such as the electron anomalous magnetic moment). Does the series have zero radius of convergence, as Dyson's 1952 vacuum-instability argument suggests? Is its Borel transform analytic along the positive real axis so that the series is Borel summable — only analyticity in a disk about the origin is currently proven — or does the positive beta function / Landau pole obstruct summability? If nonperturbative ambiguities exist, do they cancel in physical observables, or do they represent genuine nonperturbative contributions to QED?

(2) Existence of the theory. Construct, or rule out, a nontrivial continuum four-dimensional QED: a mathematically well-defined model (satisfying the Osterwalder–Schrader or Wightman axioms, or an explicitly stated equivalent) obtained as the cutoff-removed limit of a regulated four-dimensional QED, at fixed finite renormalized charge α ≈ 1/137 and electron mass, whose correlation functions have the renormalized perturbation series as their asymptotic expansion. Ruling it out means proving that every such continuum limit is necessarily trivial (zero renormalized charge, as holds for the four-dimensional Ising/φ⁴ model) or does not exist.

The two questions are two faces of one objective: standard Borel logic connects the summability of the series to the existence of a theory it expands. A complete answer resolves the objective from one direction — a constructive answer supplies the model and, with it, the summability status of its series; an impossibility answer (proven non-summability of the series for well-defined observables, or proven triviality/non-existence of the continuum limit) shows the series has no non-perturbative sum of the standard kind. The regulated framework (lattice, momentum cutoff, or other), gauge fixing, and infrared treatment are the solver's to choose, provided they cover the continuum limit of four-dimensional QED with standard renormalization conditions and are stated precisely. Heuristic arguments such as Dyson's, numerical evidence, functional-renormalization-group studies, and computations to any finite loop order do not close the problem.

References

1. F. J. Dyson, The S Matrix in Quantum Electrodynamics, Physical Review 75, 1736 (1949). DOI: 10.1103/PhysRev.75.1736, https://doi.org/10.1103/PhysRev.75.1736
2. F. J. Dyson, Divergence of Perturbation Theory in Quantum Electrodynamics, Physical Review 85, 631 (1952). DOI: 10.1103/PhysRev.85.631, https://doi.org/10.1103/PhysRev.85.631
3. M. Gell-Mann and F. Low, Bound States in Quantum Field Theory, Physical Review 84, 350 (1951). DOI: 10.1103/PhysRev.84.350, https://doi.org/10.1103/PhysRev.84.350
4. E. E. Salpeter and H. A. Bethe, A Relativistic Equation for Bound-State Problems, Physical Review 84, 1232 (1951). DOI: 10.1103/PhysRev.84.1232, https://doi.org/10.1103/PhysRev.84.1232
5. W. E. Caswell and G. P. Lepage, Effective Lagrangians for bound state problems in QED, QCD, and other field theories, Physics Letters B 167, 437 (1986). DOI: 10.1016/0370-2693(86)91297-9, https://doi.org/10.1016/0370-2693(86)91297-9
6. M. I. Eides, H. Grotch and V. A. Shelyuto, Theory of light hydrogenlike atoms, Physics Reports 342, 63 (2001). DOI: 10.1016/S0370-1573(00)00077-6, arXiv:hep-ph/0002158, https://doi.org/10.1016/S0370-1573(00)00077-6
7. F. Bloch and A. Nordsieck, Note on the Radiation Field of the Electron, Physical Review 52, 54 (1937). DOI: 10.1103/PhysRev.52.54, https://doi.org/10.1103/PhysRev.52.54
8. D. R. Yennie, S. C. Frautschi and H. Suura, The infrared divergence phenomena and high-energy processes, Annals of Physics 13, 379 (1961). DOI: 10.1016/0003-4916(61)90151-8, https://doi.org/10.1016/0003-4916(61)90151-8
9. T. Kinoshita, Mass Singularities of Feynman Amplitudes, Journal of Mathematical Physics 3, 650 (1962). DOI: 10.1063/1.1724268, https://doi.org/10.1063/1.1724268
10. T. D. Lee and M. Nauenberg, Degenerate Systems and Mass Singularities, Physical Review 133, B1549 (1964). DOI: 10.1103/PhysRev.133.B1549, https://doi.org/10.1103/PhysRev.133.B1549
11. P. P. Kulish and L. D. Faddeev, Asymptotic conditions and infrared divergences in quantum electrodynamics, Theoretical and Mathematical Physics 4, 745 (1970). DOI: 10.1007/BF01066485, https://doi.org/10.1007/BF01066485
12. N. N. Bogoliubow and O. S. Parasiuk, Über die Multiplikation der Kausalfunktionen in der Quantentheorie der Felder, Acta Mathematica 97, 227 (1957). DOI: 10.1007/BF02392399, https://doi.org/10.1007/BF02392399
13. K. Hepp, Proof of the Bogoliubov-Parasiuk theorem on renormalization, Communications in Mathematical Physics 2, 301 (1966). DOI: 10.1007/BF01773358, https://doi.org/10.1007/BF01773358
14. W. Zimmermann, Convergence of Bogoliubov's method of renormalization in momentum space, Communications in Mathematical Physics 15, 208 (1969). DOI: 10.1007/BF01645676, https://doi.org/10.1007/BF01645676
15. J. C. Ward, An Identity in Quantum Electrodynamics, Physical Review 78, 182 (1950). DOI: 10.1103/PhysRev.78.182, https://doi.org/10.1103/PhysRev.78.182
16. Y. Takahashi, On the generalized Ward identity, Il Nuovo Cimento 6, 371 (1957). DOI: 10.1007/BF02832514, https://doi.org/10.1007/BF02832514
17. C. M. Sommerfield, Magnetic Dipole Moment of the Electron, Physical Review 107, 328 (1957). DOI: 10.1103/PhysRev.107.328, https://doi.org/10.1103/PhysRev.107.328
18. T. Aoyama, M. Hayakawa, T. Kinoshita and M. Nio, Tenth-Order QED Contribution to the Electron g−2 and an Improved Value of the Fine Structure Constant, Physical Review Letters 109, 111807 (2012). DOI: 10.1103/PhysRevLett.109.111807, https://doi.org/10.1103/PhysRevLett.109.111807
19. M. Beneke, Renormalons, Physics Reports 317, 1 (1999). DOI: 10.1016/S0370-1573(98)00130-6, arXiv:hep-ph/9807443, https://doi.org/10.1016/S0370-1573(98)00130-6
20. I. M. Suslov, Divergent Perturbation Series, arXiv:hep-ph/0510142 (2005), https://arxiv.org/abs/hep-ph/0510142
21. J. S. Feldman, T. Hurd, L. Rosen and J. D. Wright, QED: A Proof of Renormalizability, Lecture Notes in Physics 312 (Springer, Berlin, 1988). DOI: 10.1007/3-540-50213-0, ISBN 9783540502135, https://doi.org/10.1007/3-540-50213-0
22. S. J. Summers, A Perspective on Constructive Quantum Field Theory, arXiv:1203.3991 (2012), https://arxiv.org/abs/1203.3991
23. M. Aizenman and H. Duminil-Copin, Marginal triviality of the scaling limits of critical 4D Ising and λφ₄⁴ models, Annals of Mathematics 194, 163 (2021). DOI: 10.4007/annals.2021.194.1.3, arXiv:1912.07973, https://doi.org/10.4007/annals.2021.194.1.3
24. D. Djukanovic, J. Gegelia and Ulf-G. Meißner, Triviality of quantum electrodynamics revisited, arXiv:1706.10039 (2017), https://arxiv.org/abs/1706.10039
25. H. Gies and J. Ziebell, Asymptotically safe QED, The European Physical Journal C 80, 607 (2020). DOI: 10.1140/epjc/s10052-020-8171-8, arXiv:2005.07586, https://doi.org/10.1140/epjc/s10052-020-8171-8
26. S. Evans and J. Rafelski, Everlasting interaction: polarization summation without a Landau pole, Physical Review D 110, 036012 (2024). DOI: 10.1103/PhysRevD.110.036012, arXiv:2311.00891, https://doi.org/10.1103/PhysRevD.110.036012