Reverberation Mapping of the Accretion Discs in the Quasars 3C 273 and 1H 2106-099

Abstract

The main aim of this thesis is to perform the first accretion disc reverberation mapping analysis on the AGN 3C 273 and 1H 2106-099 using the modern reverberation mapping algorithms Javelin, PyceCREAM and PyROA. This is with the intention of obtaining useful physical insights into these AGN and to compare the performance of the algorithms. Through spectral, photometric and reverberation mapping measurements we find evidence to suggest the accretion disc spectrum in 3C 273 follows a power law with a slightly shallower exponent β ∼ 1 than expected from the approximated thin disc model (β = 4/3 ). However the difference does not seem very significant and good agreement was found with the more physically meaningful unapproximated thin disc model simulated with a boundary condition at the radius of innermost stable circular orbit. We therefore conclude 3C 273 likely conforms to the thin disc model and displays the ’accretion disc size problem’ with a scale ∼ 2 − 3 larger than expected. For 1H 2106-099, we found an unexpected discontinuity in the PyROA and Javelin lag estimates which is reflected in the spectrum and which we cannot identify with any specific contamination. Investigating the possibility that the discontinuity is anomalous, we obtain corrected Javelin and PyROA lag estimates in near perfect agreement with the thin disc model without an up-scaled accretion disc. Our PyceCREAM RM results also indicate that the discontinuity is anomalous but differ from PyROA and Javelin in suggesting the accretion disc in 1H 2106-099 is up-scaled by a factor ∼ 2. A secondary aim of the thesis was to investigate how the uncertainties on lag estimates returned by the algorithms depends on the length of the observing period and cadence of the light-curve data. Analysis on two dust reverberation mapping campaigns returned results which suggest that the length of the light-curve relative to the expected lag has a more significant effect on the size of the uncertainties than the cadence relative to the lag. We then estimate the optimum light-curve length to be ∼ 10× the expected lag and the optimum cadence to be ∼ 6× smaller than the expected lag which had associated lag uncertainties of ∼ 11% and ∼ 10% respectively.