2005/05/03 by Ata Kabán, Ata Kaban, Kaban, Ata +5
Chemistry · Computer Science · Physics and Astronomy · #Astrophysics (astro-ph) #Bayesian Modeling and Causal Inference #Blind Source Separation Techniques #FOS: Physical sciences #Spectroscopy and Chemometric Analyses #astro-ph
paper · pdf · doi:10.48550/arxiv.astro-ph/0505059
12 Pages, 7 figures; accepted in SIAM 2005 International Conference on Data Mining, Newport Beach, CA, April 2005
arxiv created 2005/05/03 · openalex publication_date 2005/05/03 · arxiv updated 2009/12/01 · openalex created_date 2025/10/24 · openalex updated_date 2026/07/28
Elliptical galaxies are believed to consist of a single population of old\nstars formed together at an early epoch in the Universe, yet recent analyses of\ngalaxy spectra seem to indicate the presence of significant younger populations\nof stars in them. The detailed physical modelling of such populations is\ncomputationally expensive, inhibiting the detailed analysis of the several\nmillion galaxy spectra becoming available over the next few years. Here we\npresent a data mining application aimed at decomposing the spectra of\nelliptical galaxies into several coeval stellar populations, without the use of\ndetailed physical models. This is achieved by performing a linear independent\nbasis transformation that essentially decouples the initial problem of joint\nprocessing of a set of correlated spectral measurements into that of the\nindependent processing of a small set of prototypical spectra. Two methods are\ninvestigated: (1) A fast projection approach is derived by exploiting the\ncorrelation structure of neighboring wavelength bins within the spectral data.\n(2) A factorisation method that takes advantage of the positivity of the\nspectra is also investigated. The preliminary results show that typical\nfeatures observed in stellar population spectra of different evolutionary\nhistories can be convincingly disentangled by these methods, despite the\nabsence of input physics. The success of this basis transformation analysis in\nrecovering physically interpretable representations indicates that this\ntechnique is a potentially powerful tool for astronomical data mining.\n