Understanding galaxies using principal component analysis
Cool science result I heard about today -- what does it mean??
A group of astronomers performed a principal component analysis on a dataset composed of six properties of galaxies including mass, luminosity, baryon density, age, etc., and found that all of the "power" in the distribution is in a single principal component.
http://dx.doi.org/10.1038/nature07366
It's like they have determined the "eigenproperties" of galaxies and there is only 1 or 2 "eigenvectors".
Of course the problem is that principal components don't tell you the *physical meaning* of the eigenvectors -- i.e., whether it's the ratio of baryon density to mass. So interpreting the results is dicey.
Also cool: Andrew West (SSP '94) is an author on the paper.
A group of astronomers performed a principal component analysis on a dataset composed of six properties of galaxies including mass, luminosity, baryon density, age, etc., and found that all of the "power" in the distribution is in a single principal component.
http://dx.doi.org/10.1038/nature07366
It's like they have determined the "eigenproperties" of galaxies and there is only 1 or 2 "eigenvectors".
Of course the problem is that principal components don't tell you the *physical meaning* of the eigenvectors -- i.e., whether it's the ratio of baryon density to mass. So interpreting the results is dicey.
Also cool: Andrew West (SSP '94) is an author on the paper.
