BS in Physics
Vardha N. Bennert
Prior to this study, Dr. Vardha N. Bennert and collaborators selected a sample of ~100 local active galaxies to study the relationships between black hole mass and host galaxy properties. The broad Hβ width is necessary to determine black hole mass. This value is determined using a spectral decomposition code was scripted in IDL by Dr. Daeseong Park. The script fit spectral features and collected data for properties such as width of emission lines and continuum contribution percentages. The results were logged for further analysis.
To probe the accuracy of the fitting process, artificial spectra were created and fitted to determine if the output width values for Hβ reasonably match the width values of the input components contained in the pseudo spectra. To test fitting accuracy with spectra containing variable host galaxy continuum contributions, the artificial spectra were given such contributions.
For this study, the extracted components from the real active galaxy spectra were used to create artificial spectra containing relevant characteristics typically found in the sample objects' spectra. 105 artificial spectra were created with their components cataloged. The pseudo-spectra contained features whose physical quantities spanned ranges typically found within the sample of real, collected spectra, as well as added random noise, in order to appropriately represent the original sample set. The varied components were Hβ width and continuum contributions from a power law continuum, host galaxy starlight, and FeII emissions. The artificial spectra were then ran through the same IDL decomposition code that originally fit the real objects. The quantitative results of the applied fitting process were then compared with the properties of the known components that composed the artificial spectra.
The comparison of input versus output through the IDL code produces a fitting percent error that was used for analysis of accuracy of fit features. The Hβ emission line width is the focus feature for contrast between the variable pseudo spectra. The errors were plotted against the variable components in order to determine a correlation between feature properties and fitting error, if any exist.
The results of my research suggest that the IDL fitting code recovers the input parameters to a typical uncertainty of ~(5±5)% Moreover, it is independent of Hβ full width at half maximum, power law continuum contribution percentage, host galaxy continuum contribution percentage, or FeII continuum contribution percentage. The creation of more artificial spectra is necessary to create a better representation of active galaxy spectra that may include extreme features not included in this study.