The accurate prediction of molecular emission spectra for environments out of thermal equilibrium (TE) requires relevant collisional rate coefficients due to the dominant gas constituents. Such regions, in so-called non-local TE (NLTE), typically consist of low density gas and/or are exposed to intense irradiation, e.g. photodissociation regions (PDRs). While rotational collisional data are generally available, the same cannot be said for diatomic rovibrational (RV) transitions, particularly due to H2 impact, or for polyatomic RV transitions. Here we review progress on the computation of RV collisional data using accurate theoretical methods, decoupling approximations, scaling methods, machine learning approaches, and high-accuracy potential energy surfaces. We highlight the failure of some scaling approaches and similarity arguments. Results for H2, HD, HCl, CO, SiO, CS, and SO are presented and demonstrated in NLTE models of star-forming regions and PDRs. Prospects for a robust collisional database to model space- and ground-based observations are discussed.
This work was partially supported by NASA grants NNX16AF09G and 80NSSC20K0360.