Population analysis of faint galaxies with mixture modeling

We investigate the use of spectral mixture modeling to decompose the integrated spectra of several galaxies. This technique uses a linear combination of template spectra to model over 93% of the variance in Kennicutt's sample of galaxy spectra. One result of our model is a stellar population parameter for each galaxy, which correlates with both star formation rate and the population's mass-to-light ratio. Armed with the template spectra, we use mixture modeling to develop a photometric system that is independent of redshift, in order to more precisely measure the color evolution of galaxies. We show how to use the photometric system to model the composition of high-redshift galaxies in terms of the template spectra, and hence estimate their star-formation history. This photometric technique avoids being biased by the nucleus, unlike techniques which rely on slit spectra.