Multivariate Study of the Star Formation Rate in Galaxies: Bimodality Revisited.

Subjective classification of galaxies can mislead us in the quest of theorigin regarding formation and evolution of galaxies. Multivariate analyses arethe best tools used for such kind of purpose to better understand thedifferences between various objects, in an objective manner. In the presentstudy an objective classification of 362~923 galaxies of the Value Added GalaxyCatalogue (VAGC) is carried out with the help of three methods of multivariateanalysis. First, independent component analysis (ICA) is used to determine aset of derived independent variables that are linear combinations of variousobserved parameters (viz. ionized lines, Lick indices, photometric andmorphological parameters, star formation rates etc.) of the galaxies.Subsequently, K-means cluster analysis (CA) is applied on the independentcomponents to find the optimum number of homogeneous groups. Finally, astepwise multiple regression is carried out on each group to predict and studythe star formation rate as a function of other independent observables. Theproperties of the ten groups thus uncovered, are used to explain theirformation and evolution mechanisms. It is suggested that three groups are youngand metal poor, belonging to the blue sequence, three others are old and metalrich (red sequence). The remaining four groups of intermediate ages cannot beclassified in this bimodal sequence: two belong to a pronounced mixture ofearly and late type galaxies whereas the other two mostly contain old earlytype galaxies. The above result is indicative of a continuous evolutionaryscenario of galaxies instead of two discrete modes, blue and red, so farsuggested by previous authors. Some of our groups occupy the transition regionwith different quenching mechanisms. This establishes the elegance of amultivariate analysis giving rise to a sophisticated refinement over subjectiveinference.