Abstract
I will present some recent results on recovering matchings planted in collections of points in Euclidean space, a model of motion tracking and related applications more realistic than previously studied models with i.i.d. weights substituted for distances. I will especially emphasize different phenomena in the behavior of maximum likelihood estimation that arise in different scalings of the ambient dimension. Time permitting, I will also propose some open questions to the workshop concerning the algorithmic challenges of tracking moving particles over many time steps. This talk is based on joint work with Jonathan Niles-Weed (NYU).