Gravitationally lensed quasars: light curves, observational constraints, modeling and the Hubble constant.

The central topic of this thesis is gravitational lensing, a phenomenon that occurs when
light rays from a background source pass near a massive object located on the line of
sight and are deflected. It is one of the most wonderful observational fact in favour
of the General Theory of Relativity (Einstein, 1916). This phenomenon constitutes a
powerful tool to probe different areas in astrophysics, including cosmology, which is our
main interest. In particular we study gravitationally lensed quasars. Refsdal (1964) was
the first to state that time delays between different lensed images of the same object,
if this one is intrisically variable, can lead to the measurement of the Hubble constant
H0, which is related the actual expansion rate of the Universe. Up to now, only a few
lensed quasars have led to H0 and the precision on it has never reached the one obtained
with other methods as the ones based on the Cosmic Distance Ladder. That is why
some scientists from around the globe decided to unite their force to measure H0 from
about thirty lensed quasars. To reach that goal, these objects are being monitored with
some mid-sized ground-based telescopes located in both hemispheres. This thesis is
realised in the framework of this collaboration called COSMOGRAIL for COSmological
MOnitoring of GRAvItational Lenses.
This work focuses on image processing and on several steps mandatory to obtain a
measurement of H0 from lensed quasars: the acquisiton of the light curves from which it
is possible to extract the time delays and the acquisition of the observational constraints
necessary to model the gravitational potential responsible for the observed configura-
tion. The central technique of this work is the image deconvolution with the MCS
algorithm (Magain, Courbin & Sohy, 1998). The main principle of this algorithm is
the non-violation of the sampling theorem in trying to obtain a better resolution in the
deconvolved frame instead of an infinite one. The final resolution in the deconvolved
frame is chosen by the user and as it is known, every image is decomposed in a contri-
bution from the point sources and another one from all the extended structures such as
arcs, rings and galaxies.
To obtain good light curves from data coming from several telescopes, good reduction
procedures are needed. That is why Vuissoz (2008) developed a semi-automated reduc-
tion pipeline including deconvolution with the MCS algorithm. In the framework of the
i
ii Abstract
present thesis, we adapt it to one of the telescopes used by the collaboration whose data
were never used before, i.e. the Mercator telescope. We also bring some modifications
to this pipeline, e.g. concerning the estimation of the error on the magnitudes of the
light curves. We apply this revised version of the reduction pipeline to HE 0435-1223,
a quadruply imaged quasar with already measured time delays (Kochanek et al., 2006).
Another object, the quad WFI J2026-4536, is then investigated: we obtain light curves
for each of the four lensed images.
Thanks to the CASTLES project (Cfa-Arizona Space Telescope LEns Survey1),
many lensed quasars have been observed with the camera 2 of NICMOS (Near Infrared
Camera and Multi-Object Spectrometer ) on board the Hubble Space Telescope. With
these high resolution images, we can obtain very accurate constraints on the geometry
of the lensed systems. But most of the time no star is available in the field of view
to obtain a good Point Spread Function (PSF). That is why we develop an iterative
strategy combined with the MCS algorithm: we call it ISMCS. This technique allows to
use the lensed images themselves to improve the PSFs step by step while simultaneously
deconvolving the frame to obtain better estimations of the extended structures in the
image.
We first test this strategy on a quadruply imaged quasar, the Cloverleaf gravitational
lens (H1413+117), and obtain relative positions precise to 1 milliarcsecond (mas). We
then apply ISMCS to the quadruply imaged quasar WFI J2033-4723 in order to con-
tribute to the estimate of the Hubble constant, as this object was monitored by our team.
We then study a sample of seven lensed systems currently monitored by COSMOGRAIL
and for which time delays have never been obtained. Here again, we obtain positional
constraints with an accuracy of around 1 to 2 mas thanks to the application of ISMCS.
We then model these systems with simple mass profiles for the main lens galaxy and
obtain an estimation of the values of the time delays. Finally we apply ISMCS to a
sample of eleven lensed quasars which already have measured time delays. When the
delays have been remeasured by our team, in four cases until now, we also model the
potential of the lens with simple mass profiles to estimate H0.
1http://www.cfa.harvard.edu/castles