Simulations of single and two-component galaxy decompositions for spectroscopically selected galaxies from the Sloan Digital Sky Survey

Abstract: We present the results of fitting simulations of an unbiased sample of SDSS galaxies utilizing the fitting routine GALFIT and analysis pipeline PyMorph. These simulations are used to test the two-dimensional decompositions of SDSS galaxies. The simulations show that single S{\'e}rsic models of SDSS data are recovered with $\sigma_{\mathrm{mag}} \approx 0.025$ mag and $\sigma_{\mathrm{radius}} \approx 5\%$. The global values (half-light radius and magnitude) are equally well constrained when a two-component model is used. Sub-components of two-component models present more scatter. SDSS resolution is the primary source of error in the recovery of models. We use a simple statistical correction of the biases in fitted parameters, providing an example using the S{\'e}rsic index. Fitting a two-component S{\'e}rsic + Exponential model to a single S{\'e}rsic galaxy results in a noisier, but unbiased, recovery of the input parameters ($\sigma_{\mathrm{total mag}} \approx 0.075$ mag and $\sigma_{\mathrm{radius}} \approx 10\%$); fitting a single S{\'e}rsic profile to a two-component system results in biases of total magnitude and halflight radius of $\approx 0.05-0.10$ mag and 5\%-10\% in radius. Using an F-test to select the best fit model from among the single- and two-component models is sufficient to remove this bias. We recommend fitting a two-component model to all galaxies when attempting to measure global parameters such as total magnitude and halflight radius.