Chemulator: Fast, accurate thermochemistry for dynamical models through emulation

Chemical modelling serves two purposes in dynamical models: accounting for the effect of microphysics on dynamics and providing observable signatures. Ideally, former must be done as part hydrodynamic simulation but this comes with a prohibitive computational cost which leads to many simplifications being used practice. To produce statistical emulator that replicates full chemical model capable solving temperature abundances gas through time. This should suffer only minor loss accuracy over including solver would have fraction cost. The gas-grain code UCLCHEM was updated include heating cooling processes large dataset outputs from possible starting conditions produced. A neural network then trained map directly inputs outputs. Chemulator an overall mean squared error (MSE) 0.0002 single time step 1000 yr is shown stable iterations MSE 0.003 log scaled after one 0.006 steps. found approximately 50,000 times faster than dependent it emulates can introduce significant some models.