ON THE MONGE–AMPÈRE EQUATION IN INFINITE DIMENSIONS (Q5714720)

The earliest version of the optimal transportation problem was formulated by Monge in 1781. In modern language, two mass distributions in \(n\)-dimensional space are given by two nonnegative Borel functions \(f(x), g(x)\) with equal integrals. A transport or transportation plan \(T\) is a Borel map satisfying the local balance of mass condition \[ \int_{T^{-1}(e)} f(x) \,dx = \int_e g(x) \,dx \quad (e = \text{Borel set}) \, , \] and the problem is to minimize the cost of transportation, \[ \min \bigg\{ \int | T(x) - x| ^2 f(x) \,dx : T \;\text{transport} \bigg\} \] (the function \(| x - y| ^2\) is only one of many possible cost functions). When the transportation plan \(T\) is a gradient of a convex function \(V,\) it can be shown that \(V\) satisfies a partial differential equation known as the Monge-Ampère equation. Monge's problem has undergone various generalizations, where the \(f, g\) become measures \(\gamma, \mu\) in a more general space \(X.\) Weak solutions (called plannings) of the transportation problem were defined by Kantorovich in 1942, which led to a gradual revival of interest in Monge's problem; a large part of the existing results came into being in the last fifteen years. In the authors' formulation \(X\) is a locally convex space and \(\gamma\) a centered Radon Gaussian measure in \(X,\) with \(\mu = g \cdot \gamma.\) The main result is that, under suitable conditions on \(g,\) an infinite dimensional generalization of the Monge-Ampère equation holds.