radioGREAT - a radio GRavitational lEnsing Accuracy Testing Challenge
The radioGREAT challenge seeks to test the ability of algorithms to measure shearing of galaxy sources in radio interferometer data to the levels of precision and accuracy necessary for weak lensing cosmology. Below are very(!) short primers on key topics.
A more detailed introduction to weak gravitational lensing cosmology can be found in the GREAT3 handbook.
As light from distant galaxies traverses the Universe it is gravitationally lensed by all matter, both dark and baryonic, distorting the shapes of the galaxy images with a shearing transformation. The presence of large concentrations of matter along the line of sight between us and populations of galaxies causes a slight alignment in galaxy shapes which is coherent across a patch of sky. By measuring the distribution of such alignments, we can track how structures in matter have grown over cosmic time and learn about the physical theories, such as Dark Energy models, which determine how structure growth proceeds.
We typically parameterise this information in terms of the ellipticity of a source, related to the major and minor axes of an ellipse and it's orientation angle. Typically, galaxies have an intrinsic ellipticity of ~0.3 whilst gravitational lensing causes a shear of ~0.01. Performing useful cosmology with weak lensing requires being able to measure this shear over a large number of source galaxies with minimal bias.
Radio Interferometer Data
A more detailed introduction to radio interferometry can be found in these slides (courtesy of Oleg Smirnov).
In order to get the resolution necessary to view distant galaxies, radio telescopes operate as interferometers -- signals from antennas large (10s of kilometres) apart are cross correlated giving a measurement of the emission coming from the sky at a particular spatial scale and a particular orientation. Arrays of 10s to 100s of telescopes can take data simultaneously, using the Earth's rotation to sample a variety of scales and orientations.
This creates visibility or UV-plane data, analogous to a representation of the sky in a poorly sampled Fourier space. This data is under-constraining: because of the unsampled scales and orientations (due to the finite number of antennas) an infinite number of real-space skies are consistent with the UV-plane data. Assumptions must be made about this missing data when "imaging": deconvolving the effect of the interferometer and reproducing a map of the sky.
For radio weak lensing, we must measure the shear over 1000s to 10,000s of faint galaxies in a single field of view, with systematic uncertainties less than 1% for detection and less than 0.01% for useful cosmology with future surveys.
History of GREAT Challenges
GRavitational lEnsing Accuracy Testing (GREAT) and Shear TEsting Programme (STEP) challanges have been run for ~10 years by the optical aand Near-IR weak lensing community. The challenges create large and high quality dimulated data sets with which competitors to blindly try to infer an included cosmic shear from populations of galaxy images. The precision necessary for weak lensing cosmology typically means this needs to be done over a large number (~104 to ~106) objects with biases in the inferred galaxy ellipticity parameters much less than 1%.
GREAT08: handbook, results.
GREAT10: handbook, results.
GREAT3: handbook, results.
These challenges have been curcial in ensuring shear measurement methods are of the necessary quality for current and future observational surveys to be dominated by statistical uncertainties, rather than systematics. The radioGREAT challenge aims to begin this process for data from radio interferometers.