PS1 photometry of 2863 ICRF3 quasars

A dedicated method based on the Functional Principal Component Analysis is developed for these sparse and irregularly sampled data. With certain regularization and normalization constraints, it allows us to obtain uniform and compatible estimates of the variability amplitudes and average magnitudes between the passbands and objects.

We find that the starting assumption of affinity of the light curves for a given object at different wavelengths is violated for several percent of the sample. The distributions of rms variability amplitudes are strongly skewed toward small values, peaking at ~0.1mag with tails stretching to 2mag. Statistically, the lowest variability is found for the r band and the largest for the reddest y band.