Using archival SDSS multi-epoch imaging data ( Stripe 82 ) , we have searched for flares due to the tidal disruption of stars by super-massive black holes in non-active galaxies .
Two candidate tidal disruption events ( TDEs ) are identified , using a pipeline with high rejection efficiency for supernovae ( SNe ) and Active Galactic Nucleus ( AGN ) flares , and minimal selection bias .
The flares have optical black-body temperatures 2 \times 10 ^ { 4 } K and their cooling rates are very low ; their observed peak luminosities are M _ { g } = -18.3 and -20.4 ( \nu L _ { \nu } = 5 \times 10 ^ { 42 } , 4 \times 10 ^ { 43 } { erg } { s } ^ { -1 } , in the rest-frame ) , qualitatively consistent with expectations for tidal disruption flares .
The properties of the TDE candidates are examined using i ) SDSS imaging to compare them to other flares observed in the search , ii ) UV emission measured by GALEX and iii ) spectra of the hosts and of one of the flares .
Our pipeline excludes optically identifiable AGN hosts , and our variability monitoring over 9 years provides strong evidence that these are not flares in hidden AGNs : 1 ) the luminosity increases in the candidate TDE flares are estimated to be greater by factors 4 and 15 than in any of the variable AGNs monitored , and 2 ) the TDE candidates ’ hosts are much quieter in the seasons not including the primary flare than are hosts of AGN flares ( less activity on average in each of the non-primary-flare seasons than in 95 % of AGNs , with a cumulative probability estimated to be \lesssim 10 ^ { -5 } for a flaring AGN to have all other seasons as quiet as for the TDE candidates ) .
The spectra and color evolution of the flares are unlike any SN observed to date , and their strong late-time UV emission is particularly distinctive .
These features , along with the high resolution with which they are placed at the nucleus , argue against these being first cases of a previously-unobserved class of SNe or more extreme examples of known SN types .
Taken together , the observed properties are difficult to reconcile with a SN or AGN-flare explanation , although an entirely new process specific to the inner few-hundred parsecs of non-active galaxies can not be excluded .
Based on our observed rate , we infer that hundreds or thousands of TDEs will be present in current and next-generation optical synoptic surveys .
Using the approach outlined here , a TDE candidate sample with O ( 1 ) purity can be selected using geometric resolution and host and flare color alone , demonstrating that a campaign to create a large sample of tidal disruption events , with immediate and detailed multi-wavelength follow-up , is feasible .
A by-product of this work is quantification of the power-spectrum of extreme flares in AGNs .