MBX—Many-Body Exchange, an optimized Python package for generalized 4-point electron-hole scattering matrix element.

MBX (Many-Body Exchange) is a user-friendly, flexible, and efficient Python library designed for the computation of quantities relevant for many-body perturbation theory calculations, such as the evaluation of bare and screened Coulomb interactions relevant for solving the Bethe-Salpeter equation (BSE), and other processes involving excitons. This enables computations of multiparticle excitations beyond excitons—such as trion, biexciton, and exciton-Fermi polaron—as well as access to exciton dispersions in a computationally efficient manner. We leverage modularity and object-oriented programming paradigm, allowing users to easily compute and modify each component that goes into the construction of various matrix elements, facilitating exploration of novel many-body interactions. Additionally, MBX comes with many toolkits, including analysis scripts for excitonic properties and their associated wavefunctions, making it suitable for both end-users and developers. Despite being written in a high-level language, MBX achieves, for many problems, performance similar to or exceeding that of existing, traditional compiled codes due to its different parallelization strategy and support for stochastic compression algorithms. Our library provides an accessible, extensible platform for both method development and high-throughput excited-state calculations, and provides a particularly competitive platform for prototyping new first-principles formalisms.