The Magnus expansion in relativistic quantum field theory

TL;DR

This paper develops a novel method to compute the Magnus expansion in relativistic quantum field theory, enabling more efficient calculations of classical observables by relating loop amplitudes to tree-level structures.

Contribution

It introduces direct methods for calculating Magnus amplitudes, expressing loop-level results in terms of tree-level phase-space integrals, and establishes relations between loop and tree amplitudes in quantum field theory.

Findings

Magnus amplitudes at tree level are expressed with retarded and advanced propagators.

One-loop Magnus amplitudes are determined by phase-space integrals of forward limits of tree amplitudes.

Magnus diagrams contributing to the classical limit are conjectured to be forward limits of trees.

Abstract

We investigate the Magnus expansion of the $N$-operator in relativistic quantum field theory, which is related to the $S$-matrix via $S = e^{iN}$. We develop direct methods to compute matrix elements of the $N$-operator, which we refer to as Magnus amplitudes, bypassing scattering amplitudes entirely. At tree level, Magnus amplitudes are expressed in terms of retarded and advanced propagators, with each diagram weighted by factors that we identify as Murua coefficients. At loop level this structure is augmented by the Hadamard cut function, and we establish remarkable relations between loop- and tree-level Magnus amplitudes. Among these, we find that $n$-point one-loop Magnus amplitudes are entirely determined by phase-space integrals of forward limits of $(n{+}2)$-point tree-level amplitudes, and hence related to Murua coefficients, and we generalise this to a class of higher-loop contributions. Furthermore, in the case of heavy particles interacting via massless mediators, we conjecture that Magnus diagrams that contribute to the classical limit are always given by forward limits of trees, and we show this explicitly in a one-loop example. We derive these results studying theories of scalar fields with cubic interactions, but our methods are applicable to general theories as well as to integral functions appearing in gravitational-wave computations. Given that Magnus amplitudes are free of hyper-classical terms, and the known relations between Magnus amplitudes and the radial action, our results lay the groundwork for systematic and efficient calculations of classical observables from quantum field theory.