Uniqueness, all-orders unitarity, and true high-energy behavior of the momentum-space supergraph construction for nonpolynomial Lagrangian field theories

The 1970 Salam-Strathdee paper introduced a momentum-space supergraph (superpropagator) method for nonpolynomial nonderivative scalar interactions - the form taken by chiral pion Lagrangians, Stueckelberg-type weak couplings and, in its ultimate motivation, Einstein gravity - and stated explicitly that the method was not rigorous, left a residual finite ambiguity, established unitarity only at the level of plausibility, and could not decide the high-energy (timelike) behavior, which at fixed order in the major coupling grows unacceptably and requires a controlled summation in that coupling to assess. The open problem is whether this construction (or an equivalent definition of the summed nonpolynomial theory) can be completed, for the motivating classes in four spacetime dimensions, into an S-matrix that is simultaneously (i) mathematically well defined as distributions including the massive case, (ii) unique, with the residual ambiguity (a real constant $b$, more generally ambiguity functions $a^{N}$) fixed by unitarity and higher-order consistency or else shown to be irreducible physical freedom, (iii) unitary to all orders in the major coupling, and (iv) of acceptable high-energy behavior after summation. In the follow-up literature (Lehmann-Pohlmeyer minimal-singularity constructions to fourth order; Keck-Taylor order-by-order schemes; the Isham-Salam-Strathdee gravity series; constructive triviality of the regularized exponential interaction; effective-field-theory and asymptotic-safety reframings; the 2024-2026 revival of Borel-finite non-polynomial theories), each facet has been settled at most for restricted subclasses, while the general uniqueness question, all-orders unitarity, and the high-energy behavior of the summed amplitudes remain open.

Background

Many Lagrangians of physical interest are nonpolynomial in the fields: the chiral $SU(2)\times SU(2)$ pion Lagrangians, Stueckelberg-type representations of intermediate-boson weak couplings (reformulations of massive-vector interactions in gauge-like form), and the Einstein gravitational Lagrangian written in terms of the contravariant metric components. Taylor-expanding such an interaction produces vertices of arbitrarily high power of the field, so the theory is nonrenormalizable: each order in the coupling brings new ultraviolet divergences requiring counterterms with ever more derivatives, and - the second classical difficulty, emphasized since Heisenberg - even after subtraction the finite parts grow unacceptably with external momenta, the growth increasing with the order of approximation.

Three lines of attack predate the source paper. First, the rigorous subtraction procedures of renormalizable theory (Dyson-Salam, Bogoliubov-Parasiuk-Hepp, analytic renormalization) can be extended to nonrenormalizable theories, but this leaves an infinity of renormalization constants whose organization and physical meaning are unset. Second, the position-space approach of Efimov and Fradkin for theories with rational nonpolynomial Lagrangians: writing the interaction function as $V(\phi)$, one defines its Dyson index $D$ by the large-field growth $V(\lambda)\sim\lambda^{D}$ and Borel-sums, in $x$ space, the divergent series in the minor coupling $f$ (Borel summation reconstructs a function from its divergent power series through the Laplace integral of the series' Borel transform; the minor coupling $f$ is the constant whose powers label the expansion in powers of the field - the coefficient of the $n$-th field power being proportional to $f^{n}$ - while the overall interaction strength is set by the major coupling $G$ in which the amplitude expansion is organized); for $D\le 4$ only a few Borel sums are ultraviolet divergent and for $D<2$ none are, suggesting that low-index nonpolynomial theories are effectively super-renormalizable. Third, Lee and Zumino studied the momentum-space Fourier transform of the second-order two-point function of the exponential model (the prototype with interaction function of exponential type, $V(\phi)\sim e^{f\phi}$) and found that the Borel sum possesses the requisite analyticity and unitarity properties but is not polynomially bounded, and is defined only up to the addition of entire functions of the momentum. What was missing was a method that works directly in momentum space, summing the infinite tower of field powers while keeping the analytic structure visible; this is what Salam and Strathdee built in the source paper, following an earlier suggestion of Volkov and the classical distributional theory of Fourier transforms of complex powers of $x^{2}$.

The resulting supergraph calculus joins vertices pairwise by superlines. A superline carries the superpropagator $\Delta^{z}$ - the analytic continuation in a complex variable $z$ (initially in the strip $0\le\mathrm{Re}\,z\le 2$) of powers of the free Feynman propagator $\Delta$ - and the sum over field powers is replaced by a Watson-Sommerfeld-type contour integral (the standard technique that replaces a discrete sum by a contour integral encircling the integers, weighted by an analytic interpolation $v(s)$ of the expansion coefficients). Euclidicity - the Wick rotation to imaginary time - is not derived but postulated in the Symanzik region (the region of spacelike external momenta where the position-space integrands are real and positive), other momentum regions being reached by analytic continuation. The method reproduces the Lee-Zumino result at second order, yields spacelike asymptotics directly, and yields conjectural discontinuity formulas of the standard unitarity-cut type, in which the imaginary part of an amplitude across a threshold cut is produced by on-shell intermediate states.

The paper itself states what remains unresolved. The limit in which an auxiliary complex variable $\lambda$ tends to $1$ is not uniquely defined: the authors define it by an average over phases with arbitrary parameters, and the requirement that the imaginary part obey the unitarity relation reduces the freedom, at second order, to a single arbitrary real constant $b$ ('Possibly we could take $b=0$'), and at higher orders to ambiguity functions $a^{N}(z_{ij})$ resembling the signature ambiguity of Regge theory, the standard framework for high-energy scattering; whether unitarity and higher-order consistency fix these is left open, and the paper notes that a good asymptotic-behavior argument 'may be a good reason for taking $b=0$' - a selection criterion from outside unitarity. The analytic manipulations are rigorous only in the zero-mass case; products of superpropagators acquire prescribed distributional singularities (principal values and delta functions) only in the cases treated; the unitarity discussion is a plausibility argument. Finally, at second order the four-particle amplitude in the timelike region grows like $G^{2}e^{s}$ (with $s$ the squared center-of-mass energy), which is physically unacceptable, and the authors conclude that any verdict on the true high-energy behavior requires first summing the series in the major coupling $G$ - just as summing selected diagram chains in renormalizable theories reveals Regge behavior invisible at any fixed order.

The follow-up literature settled pieces of this for restricted classes - notably the Lehmann-Pohlmeyer minimal-singularity distributions for exponential Lagrangians, selected by the criterion of mildest singular behavior among the admissible definitions (explicitly to fourth order) and order-by-order finite unitary schemes for massless nonderivative Lagrangians - while a rigorous constructive result later showed the regularized exponential interaction itself to be trivial above two dimensions (its Euclidean correlation functions, the Schwinger functions, converge to those of the free field), and the mainstream reframed the whole subject through effective field theory and asymptotic safety (the program seeking a nontrivial ultraviolet fixed point of the renormalization group that would make gravity nonperturbatively renormalizable). The revived 2024-2026 work on Borel-finite non-polynomial potentials confronts the same uniqueness and ultraviolet-behavior questions again.

Problem Statement

For nonpolynomial interactions of real scalar fields in four-dimensional Minkowski space - at minimum the nonderivative exponential and rational prototype Lagrangians to which the method applies directly, together with the chiral Lagrangians and the derivative-coupled gravitational interaction as the motivating cases the program was designed to reach - decide whether the momentum-space supergraph construction of Salam and Strathdee (or an equivalent definition of the summed nonpolynomial theory) can be completed into a physically acceptable and fully specified S-matrix. Concretely, the following jointly define the target:

1. Well-definedness. Give a definition of all supergraph amplitudes as distributions - including products of superpropagators at arbitrary vertices and the massive-field case - with rigorous justification of the interchanges of summations, of the $z$-contour deformations, of the auxiliary-parameter (Borel-type) integrals, and of the Fourier transforms on which the construction rests.

2. Uniqueness of the ambiguity. Determine whether the residual freedom in the construction - the arbitrary real constant $b$ at second order and the ambiguity functions $a^{N}(z_{ij})$ of the superline continuation variables at higher orders - is fixed by unitarity together with higher-order consistency conditions, or instead labels physically distinct theories; in the latter case, characterize precisely what additional physical input (renormalization conditions, matching to experiment, or a minimality principle) is needed to select a unique amplitude.

3. Unitarity to all orders. Prove, or disprove, that the amplitudes so defined satisfy the unitarity relations - discontinuity and factorization formulas with the correct on-shell intermediate states - to all orders in the major coupling $G$, not merely at the plausibility level of the source paper.

4. High-energy behavior. Determine the true asymptotic behavior of the amplitudes at large timelike external momenta after a controlled summation of the series in $G$: in particular, whether the exponential growth seen at fixed low order in $G$ is an artifact of truncation, and whether the summed amplitudes are polynomially (Froissart) bounded, as physical two-particle amplitudes of local theories must be.

A claimed construction must be non-degenerate: it must define a genuinely interacting S-matrix and must not reduce to the free theory. In particular, the known Euclidean triviality of the regularized exponential interaction above two dimensions is a property of that regularization and does not by itself supply any of the four properties for the supergraph construction.

This question spans the nonderivative exponential and rational prototype Lagrangians together with the chiral Lagrangians and the derivative-coupled gravitational interaction, and items 2-4 concern the all-orders and summed theory.

References

1. Abdus Salam and J. Strathdee, "Momentum-Space Behavior of Integrals in Nonpolynomial Lagrangian Theories", Physical Review D 1, 3296-3312 (1970). DOI: 10.1103/PhysRevD.1.3296, URL: https://doi.org/10.1103/PhysRevD.1.3296 (source paper)
2. Benjamin W. Lee and Bruno Zumino, "Some considerations on the Efimov-Fradkin method in non-linear field theories", Nuclear Physics B 13, 671-686 (1969). DOI: 10.1016/0550-3213(69)90200-4, URL: https://doi.org/10.1016/0550-3213(69)90200-4
3. G. V. Efimov, "On local quantum field theory without ultraviolet divergences", Physics Letters 4, 314-316 (1963). DOI: 10.1016/0031-9163(63)90490-6, URL: https://doi.org/10.1016/0031-9163(63)90490-6
4. E. S. Fradkin, "Application of functional methods in quantum field theory and quantum statistics", Nuclear Physics 49, 624-640 (1963). DOI: 10.1016/0029-5582(63)90126-3, URL: https://doi.org/10.1016/0029-5582(63)90126-3
5. Arthur M. Jaffe, "High-Energy Behavior in Quantum Field Theory. I. Strictly Localizable Fields", Physical Review 158, 1454-1461 (1967). DOI: 10.1103/PhysRev.158.1454, URL: https://doi.org/10.1103/PhysRev.158.1454 (defines the localizable-field classification used in the superpropagator literature)
6. M. K. Volkov, "Method of construction of the Green's functions in momentum space for unrenormalizable interactions", Annals of Physics 49, 202-218 (1968). DOI: 10.1016/0003-4916(68)90195-4, URL: https://doi.org/10.1016/0003-4916(68)90195-4
7. H. Lehmann and K. Pohlmeyer, "On the superpropagator of fields with exponential coupling", Communications in Mathematical Physics 20, 101-110 (1971). DOI: 10.1007/BF01646529, URL: https://doi.org/10.1007/BF01646529
8. K. Pohlmeyer, "Higher order perturbation theory for exponential Lagrangians: Third order", Communications in Mathematical Physics 26, 130-148 (1972). DOI: 10.1007/BF01645700, URL: https://doi.org/10.1007/BF01645700
9. K. Pohlmeyer, "Higher order perturbation theory of exponential Lagrangians: Fourth order", Communications in Mathematical Physics 35, 321-346 (1974). DOI: 10.1007/BF01646353, URL: https://doi.org/10.1007/BF01646353
10. B. W. Keck and J. G. Taylor, "On the finiteness of nonderivative nonpolynomial Lagrangians", Journal of Physics A: General Physics 5, 1473-1478 (1972). DOI: 10.1088/0305-4470/5/10/012, URL: https://doi.org/10.1088/0305-4470/5/10/012
11. J. G. Taylor, "Perturbation theory for nonderivative nonpolynomial Lagrangians", Journal of Mathematical Physics 14, 68-78 (1973). DOI: 10.1063/1.1666174, URL: https://doi.org/10.1063/1.1666174
12. C. J. Isham, Abdus Salam, and J. Strathdee, "Infinity Suppression in Gravity-Modified Electrodynamics. II", Physical Review D 5, 2548-2565 (1972). DOI: 10.1103/PhysRevD.5.2548, URL: https://doi.org/10.1103/PhysRevD.5.2548
13. C. J. Isham, Abdus Salam, and J. Strathdee, "Is quantum gravity ambiguity-free?", Physics Letters B 46, 407-411 (1973). DOI: 10.1016/0370-2693(73)90153-6, URL: https://doi.org/10.1016/0370-2693(73)90153-6
14. G. Lazarides and A. A. Patani, "Superpropagators as boundary values of analytic distributions", Journal of Mathematical Physics 15, 1281-1284 (1974). DOI: 10.1063/1.1666808, URL: https://doi.org/10.1063/1.1666808
15. Alan Chodos and Kenneth Lane, "Nonpolynomial Lagrangian Field Theory Without Asymptotic Expansions", Physical Review D 4, 1667-1681 (1971). DOI: 10.1103/PhysRevD.4.1667, URL: https://doi.org/10.1103/PhysRevD.4.1667
16. Z. Horváth and G. Pócsik, "High-energy summation of ladder diagrams in nonpolynomial field theories", Annals of Physics 74, 555-566 (1972). DOI: 10.1016/0003-4916(72)90151-0, URL: https://doi.org/10.1016/0003-4916(72)90151-0
17. S. N. Biswas and G. P. Malik, "Note on Regge behavior in a nonpolynomial Lagrangian theory", Physical Review D 12, 3360-3362 (1975). DOI: 10.1103/PhysRevD.12.3360, URL: https://doi.org/10.1103/PhysRevD.12.3360
18. R. Delbourgo, "How to deal with infinite integrals in quantum field theory", Reports on Progress in Physics 39, 345-399 (1976). DOI: 10.1088/0034-4885/39/4/002, URL: https://doi.org/10.1088/0034-4885/39/4/002
19. S. Albeverio, G. Gallavotti, and R. Høegh-Krohn, "Some results for the exponential interaction in two or more dimensions", Communications in Mathematical Physics 70, 187-192 (1979). DOI: 10.1007/BF01982355, URL: https://doi.org/10.1007/BF01982355
20. John F. Donoghue, "General relativity as an effective field theory: The leading quantum corrections", Physical Review D 50, 3874-3888 (1994). DOI: 10.1103/PhysRevD.50.3874, URL: https://doi.org/10.1103/PhysRevD.50.3874 (arXiv:gr-qc/9405057)
21. M. Reuter, "Nonperturbative evolution equation for quantum gravity", Physical Review D 57, 971-985 (1998). DOI: 10.1103/PhysRevD.57.971, URL: https://doi.org/10.1103/PhysRevD.57.971 (arXiv:hep-th/9605030)
22. Mikhail Shaposhnikov and Sebastian Zell, "Non-polynomial interactions as a path towards a non-renormalizable UV-completion", Physics Letters B 855, 138804 (2024). DOI: 10.1016/j.physletb.2024.138804, URL: https://doi.org/10.1016/j.physletb.2024.138804 (arXiv:2312.13359)
23. Thomas Curtright and Galib Hoq, "Analytic Scalar Field Theory", arXiv:2509.00686 (2025; latest version June 2026). URL: https://arxiv.org/abs/2509.00686