Martingale characterization of \(G\)-Brownian motion (Q1001847)

In 2006, motivated by risk measures, \textit{S. Peng} [\(G\)-expectation, \(G\)-Brownian motion and related calculus to Ito's type, \url{http://abelsymposium.no/symp2005/preprints/peng.pdf} (2006)] introduced a new concept of nonlinear expectation, the so-called \(G\)-expectation, generated by a nonlinear heat equation with infinitesimal generator \(G\). Associated with, he investigated \(G\)-normally distributed random vectors and the \(G\)-Brownian motion, a natural generalization of the concept of Brownian motion to the framework of non linear expectation spaces. Going farther in his studies he developed for the \(G\)-Brownian motion an Itô type calculus and studied stochastic differential equations under \(g\)-expectation. The objective of the authors of the present paper is to extend the Lévy martingale characterisation for classical Brownian motions to the framework of a \(G\)-Brownian motion \(B\). However, the framework of the nonlinear \(G\)-expectation space and the fact that under a \(G\)-expectation the quadratic variation process \(\langle B\rangle\) is not a deterministic function impose in addition to Lévy's conditions that \(B\) is a (symmetric) martingale and \(B_t^2-t\) is a martingale three further conditions. As an application of their Lévy type characterisation of the \(G\)-Brownian motion the authors present a method for constructing a \(G\)-Brownian motion using a Markov chain. Last not least they obtain an integral representation theorem for some special symmetric martingales with respect to the \(G\)-Brownian motion.