Background

The field of computational atomic, molecular and optical (AMO) physics is one of the oldest applications of computer programming. Since the 1960s researchers have been developing and applying software to solve some of the most fundamental problems in science- determining the electronic structure of atoms and molecules and modelling the interaction of these fundamental particles with light. In the modern era, people still perform these calculations, although with modern computing power, it has become possible to model extreme forms of matter, such as highly ionised atomic and molecular species that exist in distant stars or nuclear fusion reactors. Recently AMO calculations have been extended into the time-domain, so that researchers can model the dynamics of electrons and nuclei which drive chemical reactions such as photosynthesis.

The range of applications of this software is vast and diverse. As well as the astrophysics, fusion energy and biological impacts already mentioned, AMO calculations underpin our understanding of quantum theory and can be used to model radiation damage, industrial plasma techniques, drug design and more. A large reason for the success of software developed in AMO physics is its slow, hierarchical development. By paying attention to the complex details of a physical system, and describing them precisely with robust numerical approaches, we may build, layer on layer, to create software with truly unique capabilities.

This power comes at a price however, and that price is code complexity. All of the code suites developed and applied by the UK-AMOR high-end computing consortium, for instance, are written in a mix of Fortran 77/90/2003, with parallelisation realised through GPU acceleration, OpenMP (shared memory) and MPI (distributed memory) techniques. At present, to make use of the software requires a user to be expert in running several packages, using HPC facilities, performing complex data analysis, as well as being sufficiently familiar with atomic physics to set up realistic calculations and interpret the results. With many of the leading figures in computational AMO physics nearing retirement, there is an urgent need to modernise the codebase, create sustainable development practices, and increase the usability of the software.