Dr. Mark Goodsell

As the LHC collects more data and the low-hanging fruit of new physics searches has been gathered, there are now numerous innovative searches being performed. These either look for unusual signatures, such as long-lived particles, or use novel techniques, such as neural networks for analysing the signals, or a statistical treatment beyond cut-and-count analyses. As new searches are published which do not fit the same mould as previous ones, this provides a challenge for theorists to reuse them to study new models, because it is impossible and undesirable to simply perform the same analysis steps as they did: we have to work out how to make good approximations. The best way to circumvent this problem is a close collaboration with the experiments. In this talk I will present examples involving axion-like particle searches, both prompt and long-lived, where in the one case ML 'surrogate models' are used, and in the other an atypical statistical treatment is called for, allowing the searches to be 'recycled'.