A monotone scheme for \(\mathrm{G}\)-equations with application to the explicit convergence rate of robust central limit theorem (Q6540466)

In the setting of a sublinear expectation space with sublinear expectation \(\hat{\mathbb{E}}\), the authors consider fully non-linear PDEs of the form\N\[\N\partial_t u-\hat{\mathbb{E}}\left[\langle D_xu,Y\rangle+\frac{1}{2}\langle D_x^2 uX,X\rangle\right]=0\N\]\Ndefined on the domain \((0,T]\times\mathbb{R}^d\) for some \(T\geq1\), and with initial condition \(u|_{t=0}=\phi\) for some \(\phi:\mathbb{R}^d\to\mathbb{R}\) bounded from below and \(\beta\)-Hölder continuous for some \(\beta\in(0,1]\). Such PDEs are used to characterise the G-distributions. In the present paper the authors propose, and establish convergence of, a monotone numerical scheme to approximate the viscosity solution \(u\) of this PDE, together with an explicit convergence rate, under appropriate moment assumptions on \(X\) and \(Y\). Two applications are also given. The first is to the robust central limit theorem for random vectors on this sublinear expectation space, for which the authors establish a Berry-Esseen-type inequality with an explicit constant. The second application is to a numerical approximation scheme for the Black-Scholes-Barenblatt equation arising in option pricing with volatility uncertainty.