Poster abstract details

Bayesian model selection applied to gravitational lens time-delays
Irène Balmès, Pier-Stefano Corasaniti


In this case study we use Bayesian model selection techniques to extract a subsample of double gravitational lenses with measured time-delays. The self-consistent assignment of model probabilities which characterizes the Bayesian approach allows us to decide whether time-delays in lens systems are influenced by quadrupole structures or whether they can be ignored in the lens modeling. It appears that three lenses out of the twelve in our initial sample, B1600+434, SBS 1520+530 and SDSS J1650+4251, are in the latter situation. The value of $H_0$ that we obtain after marginalizing over all model parameters including angular position errors is $76^{+20}_{-42}$ km s$^{-1}$ Mpc$^{-1}$. Future cosmic surveys will provide larger time-delays lens samples. Using this technique it will be possible to select homogeneous subsample that can be used for cosmological analyses.