The behaviour of molecular clouds in the interstellar medium is governed, in part, by the collisions of H\(_2\) molecules. These provide ways of cooling interstellar shock fronts which can set the stage for star formation. The first mechanism has to do with collisional dissociation of molecular hydrogen. Since the probability of recombination of the molecule is negligible due to the low density regime, the dissociation energy is permanently removed from the shock. The second mechanism is due to the quadrupole emission from a collisionally excited molecule of a photon that has little chance of being reabsorbed. Depending on the density, this quadrupole emission can compete with the collisional processes. Because the gas is not in equilibrium, state-specific information is required for energy transfer and dissociation. With 348 (v,j) states for each molecule, this presents a computational challenge.

Current work is focused on dissociation. With calculations of over 800 combinations of initial states, preparations are being made to apply machine learning techniques to evaluate the over 60000 state-specific rate coefficients needed.