Models
The continuum light curve variability of the 6 objects studied here appears to contradict the predictions of a standard black body discs reprocessing model. The results clearly show that in all 6 objects, light curve continuum lags are larger than expected for a disc reprocessing model.
This anomalous behaviour extends to the X-ray-to-optical variability. According to most disc reprocessing models, X-rays drive the optical variability. X-ray light curves were available for Zw229, NGC 6814, NGC 4151 and NGC 5548 (see Chapter 5. In the case of NGC 6814, I find that the X-rays correlate well with the UV and optical variability but show a high lag of
3.8. Conclusions and Future Work: Implications for Standard Disc Models 1 day. The same is true for NGC 4151, where many of the strong X-ray light curve features propagate through to the UV light curve, but with a lag of around 3 days (Edelson et al., 2017). This is much larger than expected by lamppost-reprocessing models. One interpretation of this is that the X-rays do drive the UV and optical variability, but are first reprocessed by an outer corona into far UV emissions. The far UV then drives the disc variability (Gardner & Done, 2016). Additionally, Buisson et al. (2017) study the variability of 21 AGN, but only detect significant correlations between the X-ray and UV in 9 targets.
CREAM
does not need to fit an assumed driving light curve as a proxy for the real thing: the driving light curve can instead be inferred from the accretion disc response. There are a num-ber of limitations to the
CREAM
modelling. First of all it includes a parameterized temperatureradius law. It could be advantageous to free the temperature profile in some way to depart from the black-body accretion disc parameterization of Equation 2.3 (Frank et al., 2002). The problem here is that high cadence, long duration, high signal-to-noise observations are al-
ready required to constrain the MM˙ and inclination parameters in the current model. If we
free this model further and expand the parameter space, it is not clear that the light curves would be of high enough quality to constrain a more flexible temperature profile to any useful precision. Modelling techniques such as maximum entropy (Horne, 1994)and linearized in- version (Done et al., 1995) already exist that constrain high dimensional parameter spaces by imposing a simplicity constraint: namely that the light curves should be as smooth as possible.
It might be useful to include such a constraint in
CREAM
and then introduce a more flexibletemperature profile. This would be a long term project that could form the basis of future post doctoral work.
The story continues in Chapters 4 and 5 where I will conduct more detailed studies of the X-ray, UV and optical variability of Zw229 and NGC 5548. In Chapter 4, I will investigate the implications of the lamp post model on the high-frequency power spectrum of AGN light curves. I will compare the features expected in high frequency power of a lamp post model with observations in Edelson et al. (2014) and discuss any further implications for standard
disc reverberation models. In Chapter 5 I will apply
CREAM
to light curve observations of NGC5548, and free the slope of the temperature radius expression to be fitted as part of
CREAM
’sMCMC fit. I will discuss the physical implications of such a change and investigate whether this is able to fit the continuum light curve lags better than traditional lamppost models. I will go further in Chapter 7 and use a more general (but less physically rigorous) approach to fit
Chapter 3. AGN Variability in 6 Seyfert 1 Galaxies.
the NGC 5548 and investigate how this improves the fit. In Chapter 8 I will consider a torn accretion disc geometry as a possible explanation for the continuum lags in NGC 5548, and investigate possible implications for the broad line region. Chapter 9 will end the thesis by
modifying
CREAM
to fit BLR light curves and measure continuum to line lags in a large sample4
Reverberating Disc Models: Implications for the
High-Frequency Variability of Continuum Light
Curves.
The following chapter investigates the effects of
CREAM
’s lamp-post-disc-reverberation modelon the high frequency variability of continuum light curves.
MacLeod et al. (2010) find that AGN continuum light curves behave as damped random walks: their power spectra exhibit a flat slope with frequency that steepens to a -2 log slope above a break frequency of several hundred days. Despite the success of the damped random walk (DRW) in modelling AGN variability in many reverberation mapping experiments (via the
JAVELIN
code Zu et al. 2011), the DRW is observationally constrained by continuum lightcurves with only around two-month cadence (MacLeod et al., 2010). The high-frequency (≈
Chapter 4. Reverberating Disc Models: Implications for the High-Frequency Variability of Continuum Light Curves.
4.0.1
Kepler’s Quest
The Kepler satellite graced the skies in 2009 and now sits in an Earth-trailing 371 day or- bit. Kepler’s primary goal was to seek out exoplanets as they pass between the line of sight from Kepler to their host star, causing dips in the star’s brightness. To date, Kepler has moni-
tored around 105 stars and been responsible for the discovery of more than 1000 exoplanets
(Lissauer et al., 2014).
Less well known are Kepler’s contributions to the study of AGN variability (Pei et al., 2014; Barth et al., 2011). The minute dips in star brightness caused by planet transits require every ounce of the milli-mag accuracy at Kepler’s disposal. In addition, the rapidity of planet transits close to their parent star demand high cadence observations. To that end, Kepler boasts spectacularly high (30 minute) cadence. Due to the nature of Kepler’s orbit, regular observations are also possible without the problem of occultation of the target by the Earth. All of these properties make Kepler a fundamentally useful instrument for any studies involving either the light curves or power spectral densities of AGN and thus handy to constrain any property that might depend on behaviour in the time or frequency domain.
The Kepler satellite has recently undertaken high-cadence (≈30 minute) observations of
a local Seyfert, Zw229-15 (Edelson et al. 2014, also see
CREAM
fits in Chapter 3). The authorsdiscover a break in the high-frequency power spectrum with a 5-day time scale. The light curve and fitted-power spectrum (Edelson et al., 2014) are reproduced in Figure 4.1. This break, at 0.2 cycles per day, is at much higher frequencies than predicted by the damped random walk model and its origin is unknown. Physical process that operate on the viscous, thermal and dynamical time scales (Chapter 1 are ruled-out as they are simply too slow to account for a the observed 5-day break (Edelson et al., 2014).