|
|
Master Colloquium |
Sviatoslav Khukhlaev
| SCHEDULED |
MPIfR
MOJAVE (Monitoring Of Jets in Active galactic nuclei with VLBA
Experiments) was organized as a long-term monitoring program to
repeatedly image a large sample of bright, radio-loud AGN at
milliarcsecond resolution, and these observations continue today. Until
now, all MOJAVE images have been reconstructed using the traditional
imaging algorithm CLEAN, which defines the final image resolution by the
restoring beam and does not provide any uncertainty estimations. In
turn, modern Bayesian algorithms, for example RESOLVE, are capable of
retrieving finer and more continuous structures, especially in bright
regions of the source. They also naturally provide uncertainty estimates
for the reconstructions, which are invaluable for any statistical
analysis. In this thesis, I use RESOLVE in an automatic regime on all
12285 MOJAVE VLBI observations and compare the performance with CLEAN. I
show that RESOLVE is generally able to reconstruct images with high
accuracy and uncover small-scale details. I fit core components in the
image domain and highlight the consistency between the estimated
brightness temperatures and those from the analogous visibility-domain
approach. These findings indicate the potential of using modern methods
alongside CLEAN to improve the image-domain analysis.