Twisted quasar light curves: implications for continuum reverberation mapping of accretion disks

TL;DR

High-quality quasar light curves reveal that continuum reverberation mapping is affected by distortions in multi-band data, necessitating improved methods for accurate accretion disk size measurements.

Contribution

This paper demonstrates that multi-band quasar light curves are convolved with wavelength-dependent transfer functions, affecting delay measurements and highlighting the need for methods that account for this distortion.

Findings

Current methods underestimate disk sizes by ~30%.

Proper modeling yields unbiased measurements.

Measurement errors can be below 5% with LSST-like data.

Abstract

With the advent of high-cadence and multi-band photometric monitoring facilities, continuum reverberation mapping is becoming of increasing importance to measure the physical size of quasar accretion disks. The method is based on the measurement of the time it takes for a signal to propagate from the center to the outer parts of the central engine, assuming the continuum light curve at a given wavelength has a time shift of the order of a few days with respect to light curves obtained at shorter wavelengths. We show that with high-quality light curves, this assumption is not valid anymore and that light curves at different wavelengths are not only shifted in time but also distorted: in the context of the lamp-post model and thin-disk geometry, the multi-band light curves are in fact convolved by a transfer function whose size increase with wavelength. We illustrate the effect with simulated light curves in the LSST ugrizy bands and examine the impact on the delay measurements when using three different methods, namely JAVELIN, CREAM, and PyCS. We find that current accretion disk sizes estimated from JAVELIN and PyCS are underestimated by $\sim30\%$ and that unbiased measurement are only obtained with methods that properly take the skewed transfer functions into account, as the CREAM code does. With the LSST-like light curves, we expect to achieve measurement errors below $5\%$ with typical 2-day photometric cadence.