1. Studying Quasar Spectra with Machine Learning in Sloan Digital Sky Survey
- Author
-
Monadi, Reza
- Subjects
Astrophysics ,Absorption lines ,Galaxy Evolution ,Intergalactic Medium ,Machine Learning ,Quasars - Abstract
In this thesis, we designed an algorithm to provide robust selection criteria in the parameterspace of measured properties of quasars. Our method combines the prior knowledge of an expertobserver with what unsupervised machine learning understands about the underlying structures inthe data to get a data-driven boundary in the multi-dimensional parameter space of quasar physicalproperties. We did that by quantifying the dissimilarity of our target group to the majority of thequasars in our data set. Our versatile method can select a cluster of similar data points that arelocated in statistically significant lower-density regions of the parameter space. We could find morequasars in the class of extremely red quasars and show our new sample has even more exotic outflowbehavior. Our final selection produces three times more quasars with visually verified CIV broadabsorption line feature, which is the signature of outflow, than the previous extremely red quasarsample. Our method is very useful in selecting the most important follow-up targets for observingred quasars.In the second project, we could assemble the largest CIV absorption line catalogue todate. By providing a probability for the existence of absorption systems in a quasar spectrum thatviis a by-product of our Bayesian model selection and Gaussian Processes methods, we removed theneed for visual inspection which is essential in dealing with the upcoming surveys with millions ofspectra. After carefully validating our method by comparing a subset of the spectra inspected inthe largest visually inspected CIV catalog to what our method predicts, we could find 113,775 CIVabsorption systems with at least 95% confidence among 185,425 selected quasar spectra from SDSSDR12. We obtain a posterior distribution for column density, velocity dispersion, and absorptionredshift for each investigated spectrum which can be used to get the maximum a posteriori value andthe credible interval. Our method is specifically useful when we want to obtain information fromlow signal-to-noise ratio data.
- Published
- 2023