Multi-leg One-loop Massive Amplitudes from Integrand Reduction via Laurent Expansion

TL;DR

The paper presents a Laurent-expansion based integrand reduction for one-loop amplitudes, enabling diagonal coefficient extraction and reducing the need for subtractions at the integrand level. Implemented in the Ninja library and interfaced with GoSam, this approach improves numerical precision and speed for massive multi-leg amplitudes. It demonstrates precision benchmarks and applies the framework to a wide array of processes with up to eight external legs, including heavy quarks and Higgs bosons, highlighting practical NLO QCD applications. The work underscores the potential of automated high-multiplicity one-loop computations for advancing phenomenology at colliders.

Abstract

We present the application of a novel reduction technique for one-loop scattering amplitudes based on the combination of the integrand reduction and Laurent expansion. We describe the general features of its implementation in the computer code NINJA, and its interface to GoSam. We apply the new reduction to a series of selected processes involving massive particles, from six to eight legs.

Figures (3)

• Figure 1: Correlation plot based on $10^4$ points for the process $u \bar{d} \to W b \bar{b} g$ with massive bottom quarks
• Figure 2: Precision Plot for $g g \to t \bar{t} H g$: the distributions are obtained using $5 \cdot 10^4$ randomly distributed phase space points.
• Figure 3: Precision plot for $u \bar{u} \to H u \bar{u} g g$ in VBF: the distributions are obtained using $10^5$ randomly distributed phase space points.