Abstract:
The Giant Metrewave Radio Telescope (GMRT)[1] currently is the most
sensitive low frequency radio telescope in operation. However, there are many
problems associated with low frequency radio observations(chap 2). Low fre-
quency radio observation is seriously a ect by man-made radio frequency
interference and dynamics of the ionosphere. This requires long observation
time with high sampling rate in order to obtain good sensitivity. Therefore
the data size becomes very large, even as big as 50-200 GB. The huge size of
les and recursive way of calibration requires a lot of computing resources.
Visual inspection, which is the traditional way of identifying bad data, is to-
tally impractical with these large data sizes.We have implemented a hybrid
scheme (Fig.2.7) wherein we carry out most of the analysis on a decimated
data set obtained by averaging over several time intervals and channels (sec-
tion 2.1.1). The key to this scheme is the ability to transfer the results of
data processing from the decimated le to the original le(section 2.1.2).
With this we save computation time without losing the sensitivity a orded
by the full data. We have also implemented algorithm to detect, quantify and
correct deviations in residual data if possible else
ag them (section 2.1.3).
Altogether, we have developed a software tool(Fig.2.7) which handles large
data set e ciently and remove many kind of corrupted data by analysing
residual UV data le, which improves sensitivity.
We planned to look at two problems which require high sensitivity imag-
ing at low radio frequencies: Feedback from Active Galactic nuclei(AGN)into
the cluster medium (section 3.1) and O -pulse emission of Pulsars(section
3.2).
