[Back]

J. Backhoff:
"Risk-sensitive optimal transport as a limit of particle systems";
Talk: Workshop on stochastic analysis applied in economics, finance and insurance, Universidad de Chile (invited); 2018-03-19 - 2018-03-23.

Optimal transport, a linear programming problem, can be obtained as a limit of entropic optimization problems. The latter, also known as Schrödinger problems, have a solid interpretation in terms of non-interacting particle systems. In this way Optimal transport itself can be given a probabilistic interpretation in terms of random particles in the zero-noise limit. In this talk we introduce a family of (non-linear) risk-sensitive optimal transport problems, and prove that such problems can be interpreted via a zero-noise limit of a system of random particles with random weights. Risk-sensitive optimal transport problems serve to find worst-case dependence structures between random variables, and are therefore relevant in actuarial science and finance. This is work in progress with G. Conforti and M. Beiglböck.

http://eventos.cmm.uchile.cl/stochasticfinance2018/abstracts/