Automatic objective classification of Supernovae

In this Thesis I present the tools for an automatic, objective classification of Supernova (SN) spectra I have developed within the Padova SN group, as well as several scientific applications of these tools. The classification of SNe is one of the fundamental issues of our study. The final aim of the classification is the ordering vast arrays of SNe in groups on the basis of their intrinsic physical properties. Our capacity to classify SNe is limited by our understanding of their physics, hence, the current classification scheme is still confused and unsatisfactory. Among many classes and subclasses only the most experienced specialists are able to orientate themselves. Thus, the classifications of not obvious events might often be subjective or biased. In a situation like this a way out and also a conservative approach to SN classification is the comparison of SN spectra with those of other well studied SNe. Such a comparison, performed in an automatic and quantitative manner, is able to sort out the objects with similar properties (i.e. to classify them) in an objective way. Therefore, the first, necessary ingredient for an objective classification is an archive of SN spectra, i.e. a large number of SN spectra of different types, to be used as templates spanning a wide range of properties. In our case, the templates come from Asiago SN spectra archive which is presented in Chapter 2, along with the current amount of data, the SNe included in it and their main properties (types, redshifts, phases). The archive contains SNe of all types (including peculiar ones) with various properties and having an extended temporal coverage. We stress that this is an essential issue, since number and variety of SN spectra in the archive have a direct impact on the ability of the comparison software to identify a given SN. The large number of spectra and the continuous additions to the archive required automatic procedures for the data management. I spent the first months of my PhD developing the software for organisation, update and enrichment of the archive data. The core ingredient of the objective classification is the software for SN spectra comparison. Chapter 3 presents the PAdova SN Spectra comPARison TOOl (PASSPARTOO), a collection of procedures developed by the author to this aim. Among these programs, the Generic cLAssification TOol (GELATO) is the most used. The algorithm, optimised for classification of SNe, is described in detail. In particular, the creation of the templates with an almost lossless technique of SN spectra processing is presented. Further, the methods for finding the best match template and for a quantitative evaluation of "goodness-of-fits" are described. Not only I show the GELATO's ability to classify SNe, but also its potential to estimate the ages of SN spectra, i.e. the time elapsed from the maximum light (or epoch of the explosion). The chapter is concluded by an overview of the other spectra comparison procedures available, by a critical discussion on the novelty of our approach and by the perspective of future improvements. Currently GELATO is routinely used by the members of Padova SN group for classification and ageing of newly discovered objects, as well as for comparative studies of SN spectra. In Chapter 4 I then show how the application of the above mentioned components into a single complete classification tool has been applied to our SN studies. I describe, in particular (Section 4.1), how our tools were used for an analysis of SN spectra obtained by the European Supernova Collaboration, never studied in detail and published before. The observational data, obtained by 7 different telescope-instruments combinations, were reduced and analysed. The important information from the spectra, like the expansion velocities, reddening and possible peculiarities of the corresponding SNe were extracted. Finally, using GELATO we verified the original classification of the objects and estimated their ages. GELATO showed to be a valuable tool for this kind of "a posteriori" analysis. The results of this study are being published in A&A (Harutyunyan et al., 2008, submitted). In the following section of the chapter (Section 4.2) I present the on-going SN classification program at Telescopio Nazionale Galileo (TNG) leaded by me. The program, initially approved for two periods, was aimed to the early-time observations of SNe and to select targets for further observational campaigns. The Target of Opportunity (ToO) observations of newly discovered SNe were followed by the rapid and efficient reduction of the data and, then, the reduced spectra were classified using GELATO. This latter showed to be a fast and reliable instrument in such "real-time" application. The program gave more than ten SN classification, which were published in the telegrams of IAU Central Bureau of Astronomical Telegrams (CBAT). The same analysis of newly discovered SNe is now an integral part of the long-term project on SNe at TNG. The most important achievement of the program were the prompt classifications of two extraordinary objects, SNe 2006gy and 2006jc, which were correctly classified as peculiar SNe, while other groups failed. The further observations and detailed studies of the data obtained suggested new new classes of SN progenitors and novel explosion mechanisms. Not only our classification program was crucial for the identification of these objects, but also the early-time spectra of these SNe obtained during the program were very important inputs for their study. Section 4.2 provides a more detailed discussion on these objects, highlighting the most important results obtained. Section 4.3 is devoted to SN 2002ic, an object whose nature has been largely debated. The observational properties of the object were interpreted by an explosion of a type-Ia SN with a dense H-rich circum-stellar matter (CSM) around it. Our work, based on a PASSPARTOO analysis, suggested a new classification as type Ic. During the study, the objective classification of SN 2002ic showing its similarity to the type-Ic SNe was one of the important clues, which suggested the true nature of this explosion. Chapter 5 summarises the Thesis and discusses the important issues for the future work.