Machine Learning Applications for Atomic and Molecular Collisions

The use of machine learning algorithms in the physical sciences has exploded in recent years, including many areas of physics such as high energy physics, quantum many body problems, quantum computing, molecular chemistry, and material science. However, despite their strong potential, these techniques have been slow to make their way into atomic collision physics. For fields such as plasma physics modeling, the success of the models relies, at least in part, on the accuracy and availability of electron scattering cross sections over a wide range of energies, target species, and collision processes. Unfortunately, the necessary data sets are often unavailable or incomplete due to the difficulty associated with detailed measurements and the challenges of widespread application of sophisticated theoretical models. Machine learning may be able to help fill the gap in available cross section data and could represent a major leap forward in the prediction of cross sections for complex atomic and molecular targets that are beyond the reach of existing theoretical models. Here, we review the current state of machine learning applications to problems in both atomic and molecular collision physics, as well as plasma modeling. We also present preliminary results for the calculation of collision cross sections using machine learning algorithms and address their potential to enhance and expand existing cross section data sets.