Principal Component Analysis Revealing AGN Spectral Variability Michael Parker with Andy Fabian, Giorgio Matt, Erin Kara, Andrea Marinucci, Dom Walton, Karri Koljonen, Will Alston, Laura Brenneman and Guido Risaliti June 16, 214 M. L. Parker et al. Principal Component Analysis June 16, 214 1 / 18
A simple example y(x) (Arbitrary units) 3 2 1 1 1 1 Original Functions y(x).75.5.25.25.5 Noisy Data y(x) (Abitrary units).1.5.1.1 Reconstructed Functions 1 1 2 3 4 5 x/π.75 1 2 3 4 5 6 x/π.1 1 2 3 4 5 x/π PCA isolates and returns the different components of a signal, removing some of the noise. When we apply this to AGN spectra, it retrieves the different variable spectral components, in a model-independent way, and we can match these to predictions from simulations. M. L. Parker et al. Principal Component Analysis June 16, 214 2 / 18
Applying PCA to spectra 1 1.5 1 F(E).1.5.1.5.5 1 2 5 1 1.5 1 2 5 1 To apply this to spectra, we divide the dataset into 1 ks spectra, then calculate normalised residuals. These residual spectra are then fed into the code. This removes the effects of the effective area of the detector, and prevents bias from higher flux at low energies. M. L. Parker et al. Principal Component Analysis June 16, 214 3 / 18
Predictions We can generate unique predictions for the PCs returned from different spectral models, by simulating a set of fake spectra and allowing the model components to vary. These simulated components can then be compared with the components returned from real data, to identify the cause of spectral variability in a particular source. M. L. Parker et al. Principal Component Analysis June 16, 214 4 / 18
Sample We have applied this method to a sample of 3 bright, variable AGN from the XMM-Newton archive. The method is highly dependent on the total number of counts, so we need at least one complete orbit, preferably more. There are now many sources in the archive with this much time, however! We plan to extend this to other instruments, as well as looking at binaries. It is interesting to note that there is no reason data from multiple observatories could not be combined, if the instrumental response is properly accounted for. M. L. Parker et al. Principal Component Analysis June 16, 214 5 / 18
PCA of a powerlaw: 3C 273.2.2.1.2.1.1.2.5 1 2 5 1.1.2.2.5 1 2 5 1 Left Fig shows the PCs returned from PCA of a simulation of a powerlaw, varying in normalisation and photon index. Right Fig shows the same thing, but for real data from 3C 273. M. L. Parker et al. Principal Component Analysis June 16, 214 6 / 18
PCA of absorption I: NGC 4395.15.1.5.2.1.2.1.2.1 93.7 % 3.59 % NGC 4395.5 1 2 5 1.5 1 2 5 Left Fig shows the PCs returned from PCA of a simulation of a partially-covered powerlaw, varying in covering fraction and continuum flux. Right Fig shows the same thing, but for real data from NGC 4395. M. L. Parker et al. Principal Component Analysis June 16, 214 7 / 18
PCA of absorption II: NGC 1365.2.15.1.5.2.1.1.2 kev 2 (Photons cm -2 s -1 kev -1 ) 5 2 1 5 2 1.5 1 2 5 1 Normalized Flux.25.2.15.1.5.2.1.1.2.5 1 2 5 1 Left Fig shows a simulation of varying column density, with a constant BB component at low energies. Right Fig shows the PCs from NGC 1365, where the low energy variability is damped out by diffuse gas around the AGN. M. L. Parker et al. Principal Component Analysis June 16, 214 8 / 18
PCA of absorption III: Where are the warm absorbers?.1 log(ξ)=2.2 log(ξ)=2.5.1.2 log(ξ)=3.2.1 log(ξ)=3.5.3.2.1.5 1 2 5 1 Many of the AGN in our sample show unambiguous evidence of warm absorption, but we don t see any robust signatures of ionized absorption variability in our sample. Two explanations for this: 1) PCA is optimised for broad-band spectral variability, so narrow features get lost; 2) warm absorbers are generally less rapidly variable than either the intrinsic source spectrum or the absorption caused by BLR clouds etc. M. L. Parker et al. Principal Component Analysis June 16, 214 9 / 18
PCA of reflection I.2.15.1.5.2.1.1.2.2.2.5 1 2 5 1.15.1.2.2.2.2.5 1 2 5 1 Left Fig shows the PCs returned from PCA of a simulation of a powerlaw, varying in normalisation and photon index AND a blurred reflection component, which is less variable than the powerlaw. Right Fig (from Parker et al. 214) shows the same thing, but for real data from MCG 6-3-15... M. L. Parker et al. Principal Component Analysis June 16, 214 1 / 18
PCA of reflection II.2.15.1.5.2.1.1.2.2.2.1.2.2.4.2 1H 77-495.2.5 1 2 5 1.2...and 1H 77-495....5 1 2 5 1 M. L. Parker et al. Principal Component Analysis June 16, 214 11 / 18
PCA of reflection III.2.15 Mrk 766.1.5.2.1.1.2.2.2.5 1 2 5 1.1.2.2.2.2...and Mrk 766....5 1 2 5 1 M. L. Parker et al. Principal Component Analysis June 16, 214 12 / 18
PCA of reflection IV.2.15.1.5.2.1.1.2.2.2.1.2.2.2 NGC 3516.2.5 1 2 5 1.2...and NGC 3516....5 1 2 5 1 M. L. Parker et al. Principal Component Analysis June 16, 214 13 / 18
PCA of reflection V.2.15.1.5.2.1.1.2.2.2.1.2.2.2 NGC 451.2.5 1 2 5 1.2...and NGC 451!.5 1 2 5 1 M. L. Parker et al. Principal Component Analysis June 16, 214 14 / 18
PCA of reflection All five of these sources are dominated by the same variability mechanism. The method shows, in a completely model independent way, that there has to be a spectral component responsible for both the soft excess and broad iron line in these sources. In all these sources (and several others) there is a strong pivoting term, which can be well modelled with changes in the photon index. In all cases, the soft excess and iron line is less variable than the continuum. M. L. Parker et al. Principal Component Analysis June 16, 214 15 / 18
Conclusions PCA is a powerful tool for examining AGN variability. It returns completely unbiased, model-independent spectral components, and can be used to examine and quantify their variability. An analysis of a large sample of bright, variable AGN has revealed a large number of different variability patterns. These patterns can be matched to the predictions from simulations to unambiguously determine the nature of the variability in each source. M. L. Parker et al. Principal Component Analysis June 16, 214 16 / 18
Mrk 335 Normalized Flux Normalized Flux.4.3.2.1.4.2.2.4.6 Data Simulation Simulation 1 2 5 1 E (kev) Gallo et al., in prep. M. L. Parker et al. Principal Component Analysis June 16, 214 17 / 18
IRAS 13349.2.2 64.2 %.2.2 19.4 %.2.2 NGC 3227 PDS 456.1.1.2.1.4 75.6 % 82.2 %.2.2.2.2 12.8 % 9.6 %.2.2 2.59 % RE 134 1ES 128.2.4.2.4 92.5 % 66.2 %.2.2.2.2 2.94 % 16.1 %.2 PG 1116.2.2 73. %.2.2 19.4 %.2.2 1.91 % 3C 273.1.2 68.2 %.2.2 3.3 %.2.55 % Ark 12.1.2 98.6 %.2.2.62 %.2 Mrk 59.1 98.2 %.2 1.28 %.2.11 %