Homing in on new computational algorithms will allow for better testing of probability distribution, ideal for large sets of variables and data in emerging research.
The EU-funded Distribution testing (1) project is examining the most efficient ways of understanding probability distribution. It is studying the complexity of samples with respect to distributions over a large area or domain. Analysing sample distributions and random variables in such cases has traditionally been complex and challenging. Thus, the main objective of the project is to develop different mathematical and computer-generated algorithms that can study distributions and probabilities in better ways. The project team is scrutinising different kinds of distributions to develop these novel algorithms. Numerous tests are being conducted and the detailed observations documented. The Distribution testing project is also probing previously unstudied properties, as well as the relationship between computational complexity and sample complexity, to reach its aims. These new algorithms will shed light on emerging applications in data mining and natural sciences. They will facilitate this area of statistics and support new sets of data and variables in research.
Coordinator: Tel Aviv University, Israel
(1) ‘Algorithms for testing properties of distributions’