Casey Law edited new.tex  almost 10 years ago

Commit id: 560f03febc91d39151f59faab767003aeb80cf73

deletions | additions      

       

Why are detected SNR so high relative to nominal limits?  VLA imaging is most robust and will produce good upper limits.  Our observations are limited by the thermal noise in the images. Because each pixel in the image is independent, compact configurations produce images with fewer pixels and lower tail probabilities. This will allow us improve sensitivityslightly  by reducing our threshold from $8\sigma$ to $7.6\sigma$. Observations have been conducted in the hybrid CnB and B configurations, which have maximum baselines ranging from 3 to 11 km and \emph{uv} grids extending from 512--2048 on a side. The processing time and memory footprint are dominated by the FFT stage of imaging. In the more compact of these configurations, the processing pipeline using 14 nodes at the AOC can search two hours of data in 8 hours, equivalent to roughly 660 images per second per node or roughly $10^4$ images per second.