The presentation on this webpage largely follows the logic of the review talk jointly presented by Beke and Kornai at the May 2004 PI meeting at Monona Terrace, Madison WI, but it still gets new material added as it becomes available.
While the original proposal is more theory-oriented than the practical work presented here, it is really the same problem, same agenda, approached from two ends: Beke works on geometrically smooth distortions, Kornai works on a discrete approximation of a continuous problem. They both share the belief that the optimal discretized solution is near a global minimum, and therefore one can use the global minimum as guidance to find the nearest lattice point.
The empirical data used in this work comes from The US Census Bureau and is in the public domain. Since this is a large and complex dataset (over 2 GB compressed) and it is gradually replaced by newer sets, for full reproducability a mirror is provided here. Note that you need to have your own GIS setup to visualize these files -- we had good luck with Quantum GIS
The software is split in two parts: Display contains the utilities used for visualizing dixels, and Analyze has the utilities employed in analyzing the Tiger map data.