Project name: Tracksearch: An analysis algorithm for long-lived unmodeled bursts

Scientific goals: This analysis pipeline will search for unmodeled bursts which are long-lived. A long-lived burst is a signal which exists for tens of cycles or more. This number of cycles could typically correspond to time scales greater than seconds and quite possibly up to minutes. Signals that tracksearch is designed to identify will come from quasi-periodic GW signals. One possible example of a long-lived type of burst signal are the theoretical GWs emitted by the gamma ray burst models by Putten.

Data set(s): white noise, simulated LIGO type noise

Analysis Approach: This analysis uses a time frequency representation(TFR) of the data. Tracksearch "looks" at these representation relying on an image processing algorithm developed by Carston Steger. Long-lived GW signals will create curvi-linear(CL) features in the TFR. These CL features are then further refined in single instrument event candidates. A future multi-IFO approach with tracksearch involves combining event candidates using a simple coincidence check.

Relationship to other projects: Tracksearch is an different than the current burst projects focused on long duration signals. The project most similar to tracksearch is "GWs associated with long quasi-periodic signals (triggered)".

People involved: Cristina Torres (UTB), Warren Anderson (UWM)

Milestones: *Major pipeline functions have been tested in simulated LIGO noise. (Standalone non-pipeline test) *A completed condor based pipeline test has been run end to end with white noise and arbitrary waveform injections. This has been run on a non-LSC computing cluster.

Documentation Links:

Presentations and paper drafts: tracksearch report

Projected timeline: March 07: Get tracksearch pipeline running on a LSC cluster.
April 07: Run tracksearch with "playground" S5 data to perform thorough pipeline debugging
July 07: Begin a "first attempt" search with a significant set of S5 data.