“…Such emulation approaches in predicting the outcomes of physical systems have a long history, including applications in physics-based animation (Grzeszczuk et al, 1998), complex multi-physics simulators (Lu et al, 2021;Bianchi et al, 2016), climate models (Beucler et al, 2019;Krasnopolsky et al, 2005;Castruccio et al, 2014;Kashinath et al, 2021), and emulating fluid flow through dolomite using a neural network (Li et al, 2022). In an emulator approach, the underlying physical system is approximated by a statistical model (the emulator) which can be evaluated more quickly than a conventional forward model.…”