Two-loop instanton correction to the large-order behavior of the 2M-point Green functions of φ⁴ theory in d = 2 and 3, and its use in Borel–conformal resummation of critical exponents

The perturbative series of the massive Euclidean φ⁴ theory in d = 2 and 3 dimensions are divergent asymptotic series whose factorial large-order growth is governed, via Lipatov's dispersion-relation method, by the instanton saddle that dominates the unstable theory at negative coupling. Malatesta, Parisi and Rizzo (Nucl. Phys. B 922 (2017) 293–318) computed the two-loop (order-g) instanton correction — the 1/k term — for the partition function and the free energy, and explicitly left open the physically central case: the 2M-point Green functions, from which critical exponents are extracted by Borel–conformal resummation. The open problem is to compute the renormalized two-loop instanton correction to the large-order behavior of the 2M-point Green functions of φ⁴ theory in d = 2 and 3 — including the zero-mode, renormalization, and continuous-spectrum subtleties absent in the quantum-mechanical case — an input whose payoff, as the source paper motivates, is a more precise Borel–conformal resummation of the d = 3 series (known through seventh order) and hence of critical-exponent estimates. The problem is open: the d = 1 quantum-mechanical analogue was computed and numerically verified in 2020–2022, and high-precision instanton integrals for the d = 2, 3 field theory were computed in 2024 as ingredients, but the Green-function computation itself has not been performed.

Background

Perturbation theory in an interacting quantum field theory produces series whose coefficients grow factorially with the order $k$ , so the series diverge and are meaningful only as asymptotic expansions. For the Euclidean scalar φ⁴ theory — the Landau–Ginzburg model underlying the theory of critical phenomena, whose single-component $d = 3$ instance describes the Ising universality class (the set of systems that share the same critical behavior, such as the liquid–gas transition and the uniaxial ferromagnet near their critical points) — this divergence has a sharp physical origin. For negative coupling $g < 0$ the potential is unstable, the observables develop an imaginary part, and a dispersion relation converts that imaginary part into the large-order growth of the perturbative coefficients. Lipatov's method evaluates the imaginary part semiclassically: it is dominated by an instanton, a finite-action classical solution of the Euclidean equations of motion localized in space (here the spherically symmetric solution of $(-\Delta+1)\Phi_c - \Phi_c^3 = 0$ after the scaling $\varphi_c = \Phi_c/\sqrt{-g}$ ), and expanding the path integral around this saddle yields the asymptotic form of the coefficients. At one loop this gives, for the coefficients $G_K$ of the $n$ -point Green function of the $D$ -dimensional $O(N)$ -symmetric theory ( $N$ the number of field components; the problem below concerns the single-component case $N = 1$ in $d = 2$ and $3$ ), a law of the form

$G_K = \frac{c(N,D)}{\pi}\left(\frac{1}{A}\right)^{(n+N+D-1)/2}\left(-\frac{1}{A}\right)^{K}\Gamma\!\left(K + \frac{n+N+D-1}{2}\right)\left[1 + O(1/K)\right],$

where $A$ is the instanton action, $c(N,D)$ is a constant coefficient fixed separately for each $N$ and $D$ , and $\Gamma$ is the Euler gamma function (which extends the factorial).

This asymptotic knowledge is not merely structural. The standard way to extract numbers from a divergent asymptotic series is Borel summation — transforming the series so that the transformed series has a finite radius of convergence and integrating the Borel transform back — and its accuracy is greatly improved by a conformal mapping of the Borel plane that exploits the known position of the leading Borel singularity, which is exactly what the one-loop instanton calculation supplies. Borel–conformal resummation of this kind underlies the most precise field-theory determinations of critical exponents — the universal power-law indices, such as the correlation-length exponent $\nu$ and the anomalous dimension $\eta$ , that govern behavior near a continuous phase transition.

The first correction to the leading large-order law — the two-loop term, of relative order $1/k$ — is much harder, and this is where the open problem lies. Expanding around the instanton breaks translational invariance: the fluctuation operator (the second functional derivative of the action at the saddle, whose eigenvalue product forms the one-loop determinant) has $d$ zero modes — translations of the instanton center, directions along which the saddle costs no action — that must be separated by collective coordinates (the instanton center, promoted to an integration variable), which adds Jacobian vertices to the Feynman rules, and every internal propagator is the Green function of a space-dependent operator rather than the free one. In $d = 2$ and $d = 3$ the theory is superrenormalizable — only a mass counterterm $\delta m^2$ is needed — but that counterterm must cancel ultraviolet divergences order by order inside the instanton expansion, and the one-loop fluctuation determinant involves a continuous spectrum whose high-eigenvalue tail is estimated by Thomas–Fermi-type asymptotics (a semiclassical estimate of the density of large eigenvalues). In quantum mechanics ( $d = 1$ ) the propagator is explicit and the two-loop correction is known analytically. Malatesta, Parisi and Rizzo (Nucl. Phys. B 922 (2017) 293–318) computed it for the $d = 2$ and $d = 3$ field theory for the simplest observable, the partition function $Z$ and the free energy density $F = -(1/V)\ln Z$ : for the free-energy coefficients they obtained

$F_k \approx -C(-A)^{-k}\,\Gamma(k+d/2)\left[1 - \frac{I_4}{4k}\left(\xi - \tfrac{3}{4} V G_0^2(0)\right) + O(1/k^2)\right],$

with the finite two-loop combination $\xi - \tfrac{3}{4} V G_0^2(0)$ evaluated numerically as 0.6553(2) in $d = 2$ and 3.4988(9) in $d = 3$ (here $C$ is the one-instanton prefactor containing the renormalized fluctuation determinant, $I_4$ a moment of the instanton profile, $V$ the volume, and $G_0(0)$ the free propagator at coinciding points).

Critical exponents, however, are extracted from correlation functions, not from the partition function: it is the perturbative series of the 2M-point Green functions that are Borel–conformal resummed, and for those only the one-loop (leading) large-order behavior is known. Malatesta, Parisi and Rizzo explicitly leave this extension to future work, observing that in $d = 3$ the perturbative series is available through seventh order, so that the two-loop large-order input is the computationally accessible route to improving resummation precision. The quantum-mechanical analogue of the correlation-function computation has since been carried out and numerically verified, but the $d = 2, 3$ field-theory computation has not.

Problem Statement

For the massive single-component Euclidean φ⁴ theory in dimensions $d = 2$ and $d = 3$ , with action

$S[\varphi] = \int d^d x \left[ \tfrac{1}{2}(\nabla\varphi)^2 + \tfrac{1}{2}\varphi^2 + \tfrac{g}{4}\varphi^4 + \tfrac{\delta m^2}{2}\varphi^2 \right],$

where $\delta m^2$ is the mass counterterm that makes the theory finite, compute the renormalized two-loop (order- $g$ ) instanton correction to the large-order behavior of the 2M-point Green functions $G^{(2M)}$ ( $M \ge 1$ an integer; only even-point functions are nonzero because the action is even in $\varphi$ ) — the $1/k$ correction to the Lipatov asymptotics of the coefficients of their perturbative expansions in $g$ — generalizing the computation performed for the partition function and free energy by Malatesta, Parisi and Rizzo. Concretely, the required results are: (a) the renormalized expansion of the imaginary part of $G^{(2M)}$ at negative coupling to order $g$ around the instanton saddle, with $2M$ external-field insertions, including the collective-coordinate treatment of the $d$ translational zero modes and the order-by-order cancellation of ultraviolet divergences by the mass counterterm; (b) the dispersion-relation reconstruction of the $1/k$ term in the large-order coefficients of $G^{(2M)}$ ; and (c) numerically evaluated, error-controlled values of the required two-loop quantities in $d = 2$ and $d = 3$ — at least for the Green functions entering the critical-exponent analysis (the two-point function with its momentum dependence, and the four-point function at zero momentum) — checking the $d = 1$ limit against the known quantum-mechanical two-loop correlation-function results and the one-loop limit against the established Lipatov asymptotics. These components form one scientific objective: producing the first usable subleading large-order input for the series that are actually resummed to extract critical exponents. The source paper motivates this computation by its physical payoff: combining the resulting large-order input with the published high-order $d = 3$ perturbative coefficients (known through seventh order) in a Borel–conformal resummation would allow the change in precision and internal consistency of critical-exponent estimates for the $d = 3$ Ising universality class to be quantified; that resummation study is the stated motivation of the source's future-work program rather than a further required deliverable beyond (a)–(c).

References

E. M. Malatesta, G. Parisi, T. Rizzo. Two-Loop Corrections to Large Order Behavior of φ⁴ Theory. Nuclear Physics B 922 (2017) 293–318. DOI: 10.1016/j.nuclphysb.2017.07.011. arXiv:1704.04458. https://arxiv.org/abs/1704.04458 (source paper)
L. T. Giorgini, U. D. Jentschura, E. M. Malatesta, G. Parisi, T. Rizzo, J. Zinn-Justin. Two-Loop Corrections to the Large-Order Behavior of Correlation Functions in the One-Dimensional N-Vector Model. Physical Review D 101, 125001 (2020). DOI: 10.1103/PhysRevD.101.125001. arXiv:2005.01485. https://arxiv.org/abs/2005.01485
L. T. Giorgini, U. D. Jentschura, E. M. Malatesta, G. Parisi, T. Rizzo, J. Zinn-Justin. Correlation Functions of the Anharmonic Oscillator: Numerical Verification of Two-Loop Corrections to the Large-Order Behavior. Physical Review D 105, 105012 (2022). DOI: 10.1103/PhysRevD.105.105012. arXiv:2111.12765. https://arxiv.org/abs/2111.12765
L. T. Giorgini, U. D. Jentschura, E. M. Malatesta, T. Rizzo, J. Zinn-Justin. Instantons in φ⁴ Theories: Transseries, Virial Theorems and Numerical Aspects. Physical Review D 110, 036003 (2024). DOI: 10.1103/PhysRevD.110.036003. arXiv:2405.18191. https://arxiv.org/abs/2405.18191
A. J. McKane. Perturbation expansions at large order: Results for scalar field theories revisited. Journal of Physics A: Mathematical and Theoretical 52, 055401 (2019). DOI: 10.1088/1751-8121/aaf768. arXiv:1807.00656. https://arxiv.org/abs/1807.00656
R. Guida, J. Zinn-Justin. Critical exponents of the N-vector model. Journal of Physics A: Mathematical and General 31, 8103 (1998). DOI: 10.1088/0305-4470/31/40/006. https://doi.org/10.1088/0305-4470/31/40/006
M. Serone, G. Spada, G. Villadoro. λφ⁴ theory — Part I. The symmetric phase beyond NNNNNNNNLO. Journal of High Energy Physics 2018, 148 (2018). DOI: 10.1007/JHEP08(2018)148. arXiv:1805.05882. https://arxiv.org/abs/1805.05882
M. Borinsky, A. Favorito. Feynman integrals at large loop order and the log-Γ distribution. Journal of High Energy Physics 2025, 182 (2025). DOI: 10.1007/JHEP09(2025)182. arXiv:2503.07803. https://arxiv.org/abs/2503.07803