Benchmarking machine learnt interatomic potentials for analysis of neutron spectroscopy data

We seek to benchmark the abilities of recent advances in machine learnt (ML) potentials, through comparison to inelastic neutron spectroscopy (INS) measurements. INS provides unparalleled sensitivity to the local atomic environment by probing vibrations. Our chosen system provides a unique challenge to ML methods as the interactions are strongly hydrogen bonding/dispersion related (something ab initio methods typically struggle with). We hope to also make use of new high pressure capabilities on TOSCA to examine a hypothesised phase change at high pressure.