arXiv:astro-ph/0211188 v2 25 Nov 2002 Astron. Nachr./AN 324 (2003) 1/2, 000–000 Statistical Identification of the XMM-Newton Sources Using the XID-DB HOURI ZIAEEPOUR and SIMON ROSEN Mullard Space Science Laboratory (MSSL), Holmbury St.Mary, Dorking RH5 6NT, Surrey, UK The XID Database is the main repository of data for the ”XMM-Newton Serendipitous Survey” (XID-project) and permits easy selection and correlation of the XMM X-ray and UV (from Optical Monitor (OM)) data with Optical/IR obser- vations. One of the first data-mining applications of the XID database is the statistical classification of X-ray sources. Up- to now attempts at statistical classification of XMM sources have been mostly concentrated on the relation between hard- ness ratios, fluxes and count rates (Della Ceca et al. 2002). Plots of hardness ratio and flux in the XID band (2 - 4.5keV ) of identified XMM sources however show that there is a large overlap between the parameter space of different classes. From ROSAT data it was already observed (Carrera et al. 2001, Page et al. 2000) that in color-flux and color-color dia- grams, most of identified AGNs and unidentified hard sources are segregated from other classes and both occupy a relatively small part of the parameter space. The correlation between hard X-ray and B-r has been interpreted as being due to intrin- sic absorption of the X-ray and optical radiation in the source (Francis et al. 1993). We have produced plots of 2-parameter combinations of X-ray flux, hardness ratios and optical mag- nitudes Hα - r, z - r, i - r, b - r, g - r and U - r for spectroscopically identified XMM sources. A preliminary in- vestigation of HR2 and HR3 (defined in Osborne 2000), versus U - r and g - r magnitudes 1 indicates the potential usefulness of these parameters for separating different classes of objects. As mentioned above this is the result of the intrin- sic absorption of X-ray which are re-emitted in longer wave lengths. Nonetheless, a more extensive set of verified identifi- cations within the database will be needed, specially for non- AGN objects, to underpin future analyses. As a first test of this method we have plotted detected sources in a randomly selected XMM field without any XID identification over the identified objects (see Fig.1). Amongst unidentified sources we have found 5 classified objects from the Simbad Catalog. Regarding the place of these objects and other objects of the same class on the plot, it seems that classes are insufficiently distinct. Better statistics and greater exploration of parameter 1 Here we have plotted both colours together to increase the statis- tics. Most objects in the database have only one of them. This only increases the scatter in U/g - r direction because for most objects in our sample the difference between two magnitudes is around 0.7. space are clearly required. More specifically, qualitative com- parison of a number of plots shows that segregation of objects seems more significant in the multiple parameter spaces. The study will be extended accordingly. Evidently using more so- phisticated statistical methods like principal component anal- ysis are useful but we postpone the use of these methods until we have better statistics. Moreover, the database will increas- ingly contain UV data from the OM, yielding a larger param- eter space for exploitation. References Carrera, F.J., Mittaz, J.P.D., Page, M.J., in Proceedings of “X-ray astronomy ’999:Stellar End-points, AGN and the Diffus Back- ground”, eds. G. Malaguti, G. Palumbo & N. White. Della Ceca, R., et al. astro-ph/0202150. Francis P.J., et al., 1993, AJ, 106, 417. Osborne, J., ed., 2000, SSC-LUX-SP-0004. Page, M.J., Mittaz, J.P.D., Carrera, F.J., astro-ph/0103055.