The ‘curse of dimensionality’ resolved
The effects of climate change and trade on national incomes and optimal surveillance measures for the early detection of pests and diseases in large dimensional economic modelling.
ACEX 250 (Thursday 25th May 5-7pm)
Abstract: Solving large scale optimisation problems over space and time quickly generates a computational impasse, termed the ‘curse of dimensionality’. This severely limits the practical use of economic models, especially for trade, natural resource management and biosecurity. In this presentation, we detail two innovative approaches for solving (otherwise unsolvable) large scale systems through: (a) the use of matrix reordering in CGE modelling, with parallel processing techniques, and (b) a ‘sample average approximation’ method, combined with a pruning device along a probabilistic network-tree. The methods are illustrated by three key examples: (1) The effects of climate change on national incomes and agricultural production; (2) The effects of changes in trade agreements on global and national incomes; and (3) Active surveillance measures for foot-and-mouth disease and the optimal trapping grid for the early detection of fruit fly invasions in Australia.
Biography: Tom Kompas is a Professor of Environmental Economics and Biosecurity in the School of Biosciences and the School of Ecosystem and Forest Sciences at the University of Melbourne. He is also one of four Chief Investigators in the Centre of Excellence for Biosecurity Risk Analysis (CEBRA), Director of the Centre for Environmental and Economic Research (CEER) at the University of Melbourne, and the Foundation Director of the Australian Centre for Biosecurity and Environmental Economics at the Australian National University. Tom is the former Editor-in-Chief of the Australian Journal of Agricultural and Resource Economics and a member of the Eminent Scientists Group in the Commonwealth Department of Agriculture. He is a Fellow of the Academy of Social Sciences Australia.