Principal component analysis of synthetic galaxy spectra

We analyse synthetic galaxy spectra from the evolutionary models of Bruzual & Charlot and Fioc & Rocca-Volmerange using the method of principal component analysis (PCA). We explore synthetic spectra with different ages, star formation histories and metallicities, and identify the principal components (PCs) of variance in the spectra resulting from these different model parameters. The PCA provides a more objective and informative alternative to diagnostics by individual spectral lines. We discuss how the PCs can be used to estimate the input model parameters, and explore the impact of dust and noise in this inverse problem. We also discuss how changing the sampling of the ages and other model parameters affects the resulting PCs. Our first two synthetic PCs agree with a similar analysis on observed spectra obtained by Kennicutt and the 2dF redshift survey. We conclude that with a good enough signal-to-noise ratio (S/N>> 10) it is possible to derive age, star formation history and metallicity from observed galaxy spectra using PCA.