| Supernovae (SNe) have played an important role in the recent dramatic development of observational cosmology. They possess homogeneous observable properties, and thus approximate "standard candles", allowing them to be precise, luminosity distance indicators. Over the last decade, the observed sample of supernovae (SNe) has increased by more than an order of magnitude. Further advances will no longer be limited by statistical errors, but rather by the control of systematic uncertainties, associated with source diversity and evolution. Over 500 SNe Ia have been discovered by the Sloan Digital Sky Survey (SDSS)---II SN Survey during the three fall seasons from 2005 to 2007. We combine spectroscopic and photometric data to explore reducing these systematic errors. One challenge is to remove accurately the host galaxy light from the observed spectra. We have developed an effective host-subtraction tool using a composite color-constrained PCA+template-fitting program. We have applied this technique to more than 700 spectra from SNe with redshifts up to 0.4 obtained from the SDSS-II SN Survey and more than 200 spectra from low-z SNe at redshifts less than 0.01 obtained from the Center for Astrophysics (CfA) SN archive. We have also developed an automatic method to quantify spectral features of SNe Ia and applied it to the CfA and SDSS-II samples. By comparing the time series of the spectral features between these two samples, we have found no sign of cosmological evolution. We have, however, found evidence for luminosity-dependent differences in the Mg II 4300, Si II 4000, Si II 5800, and Si II 6150 lines. This should be useful for cosmological studies. |
