Bright pulses from distant objects probe the universe and are used to test General Relativity and reveal spatial variations in the extremely dilute gas between stars and galaxies. A recently discovered class of erratic and very bright neutron stars adds diversity to the list of cosmic probes, but their sporadic bursts are more difficult to detect than the regular beats of standard pulsars.
I will explain some basic principles of radio telescope arrays and recent efforts to use them for detecting millisecond impulses. Although single-dish telescopes pioneered the study of transient bursts, interferometric arrays can localize sources more accurately, survey more quickly, and suppress interference more effectively. The catch is running large arrays at millisecond time steps and generating massive data streams to represent the interference fringes between every pair of antennas in a large array.
I will describe a new technique that collapses burst detection on an array to the computational scale required of a single-dish telescope yet maintains the ability to follow up with imaging and precise source localization. The technique, based on phase closure constraints, is computationally efficient enough for real-time, all-time burst detection. A statistical extension to the technique recovers coherent fringe information across frequency channels to substantially open the field of view and improve survey speeds.