In this paper we have examined the dependance of amplitude on redshift and luminosity for large samples of quasars selected on the basis of ultra-violet excess and variability. The quasars span a redshift range 0 < z < 3.5 and a luminosity range -20 > MB > -28. There is evidence for a correlation of amplitude with luminosity and/or redshift for the sample as a whole, but when it is binned in redshift the correlation with luminosity becomes significant. This result could be strengthened by the possible existence of a population of large amplitude low luminosity objects in the sample. No convincing evidence is found for a correlation between amplitude and redshift, either for the sample as a whole or when it is binned in luminosity.
Various models of quasar variability are examined with respect to the observed correlations. It is concluded that any straightforward interpretation of an accretion disk model is incompatible with the data. The Christmas Tree model has some merits, especially in the regime of low luminosity quasars and Seyfert galaxies, but for luminous quasars the rate of supernovae required is too large to be compatible with the observed variability. The microlensing model can be used to explain all the data, although it does require that the Einstein radius of the microlensing bodies is comparable in size to the quasar nucleus.
Acknowledgements
I thank Andy Lawrence and Omar Almaini for making some excellent suggestions for improvements to the paper.
Copyright The European Southern Observatory (ESO)