[Previous] | [Session 50] | [Next]
R. Cid Fernandes (JHU & UFSC-Brazil), L. Sodre (IAG-USP-Brazil)
A generalized Poissonian formulation of quasar variability is developed and used as a mathematical tool to extract relevant parameters such as the energy, rate and lifetimes of the flares through the analysis of observed light curves. It is shown that in this very general framework the well established anti-correlation between variability amplitude and \lambda can only be understood as an effect of an underlying spectral component which remains stable on long time-scales, and is redder than the variable component. The formalism is applied to the B and R light curves of 42 PG quasars collected by the Wyse Observatory group (Giveon et\ al.\ 1999). Variability indices for these data are obtained with a Structure Function analysis. The mean number of living flares is constrained to be in the range between {\cal N} ~5 and 100, while their rates are found to be of order \nu ~1--100 yr-1. Monochromatic flare energies E\lambda ~1046-48 erg\,Å-1 and life-times \tau ~0.5 to 3 yr are derived. Lower limits of typically 25% are established for the contribution of a non-variable component in the R band. The diversity in these properties among quasars invalidates simple versions of the Poissonian model in which flare energies, lifetimes and the background contribution are treated as universal invariants. Light Curve simulations confirm the applicability of the method.
The significant correlation between EW(H\beta) and the long term variability amplitude is interpreted in a scenario where only the variable component participates in the ionization of the line emitting gas. This is consistent with the observed trends of the asymptotic variability amplitude with \lambda, EW(HeII) and the X-ray to optical spectral index. The parameter estimates derived under the framework of Poissonian models are applicable to several scenarios for the nature of quasar variability, and can help guiding, testing and discriminating between detailed physical models.
This work was supported by a NSF-Gemini fellowship, as well as the Brazilian agencies CNPq, CAPES, PRONEX and FAPESP.
The author(s) of this abstract have provided an email address for comments about the abstract: cid@pha.jhu.edu