http://arxiv.org/abs/1603.06331
We modify upon the algorithm we proposed before in \citep{aghamousa_timedelay_1} on time delay estimation of the strong lens systems incorporating weighted cross correlation, defining two tuning parameters in the analysis and optimizing the method (deriving its parameters) by trading off between the bias and variance using many Monte Carlo simulations. We apply our proposed method on the light curves of the lensed quasar SDSS J1001+5027 since this system has been well studied by other groups to compare our results with their findings. In this work we propose two estimators namely “mean” and “mirror” estimators and we show that these two estimators should result to consistent and accurate estimations. Our mirror estimator results to $-117.1^{+1.6}_{-1.0}$ days time delay for this system and our mean estimator results to $-119.8^{+1.8}_{-0.8}$ days using Monte Carlo simulations to estimate the uncertainties. These two estimations are very much consistent with results of the other groups, however, the small discrepancy between these estimations hints towards some possible minor systematics in the data or inaccurate estimation of the uncertainties of the data epochs. Using simulations provided by \citep{COSMOGRAIL_XIV}, where they account for some possible systematics in the simulations, we got larger error bars on the estimated time delays while the mean estimated values remained unchanged.
A. Aghamousa and A. Shafieloo
Tue, 22 Mar 16
60/72
Comments: N/A