2 posts tagged with "bayesian inference"

Authors: Holdship, Jonathan; Viti, Serena; Codella, Claudio; Rawlings, Jonathan ; Jimenez-Serra, Izaskun; Ayalew, Yenabeb ; Curtis, Justin ; Habib, Annur ; Lawrence, Jamel ; Warsame, Sumaya ; Horn, Sarah
Arxiv: https://arxiv.org/abs/1904.11360
DOI: https://doi.org/10.3847/1538-4357/ab1f8f

As part of the ORBYTS programme, a group of secondary school students in London studied the emission of COMs in protostellar outflows. They were provided with data from IRAM 30 m Observations toward eight protostellar outflow sources that were taken in the 96-176 GHz range. The students detected transitions of CH3OH and CH3CHO were detected in seven of them. The integrated emissions of the transitions of each species that fell into the observed frequency range were measured and fit using RADEX and LTE models. Column densities and gas properties inferred from this fitting are presented. The ratio of the A and E-type isomers of CH3OH indicates that the methanol observed in these outflows was formed on the grain surface. Both species demonstrate a reduction of terminal velocity in their line profiles in faster outflows, indicating destruction in the post-shock gas phase. This destruction, and a near constant ratio of the CH3OH and CH3CHO column densities, imply it is most likely that CH3CHO also forms on the grain surface.

Authors: Holdship, J.; Jeffrey, N.; Makrymallis, A.; Viti, S.; Yates, J.
Arxiv: https://arxiv.org/abs/1809.06245
DOI: https://doi.org/10.3847/1538-4357/aae1fa

Grain surface chemistry and its treatment in gas-grain chemical models is an area of large uncertainty. While laboratory experiments are making progress, there is still much that is unknown about grain surface chemistry. Further, the results and parameters produced by experiments are often not easily translated to the rate equation approach most commonly used in astrochemical modeling. It is possible that statistical methods can reduce the uncertainty in grain surface chemical networks. In this work, a simple model of grain surface chemistry in a molecular cloud is developed and a Bayesian inference of the reactions rates is performed through Markov Chain Monte Carlo sampling. Using observational data of the solid state abundances of major chemical species in molecular clouds, the posterior distributions for the rates of seven reactions producing CO, CO2, CH3OH, and H2O are calculated in a form that is suitable for rate equation models. This represents a vital first step in the development of a method to infer reaction rates from observations of chemical abundances in astrophysical environments.