Dual formulation of second order target problems (Q1948690)

In their paper the authors reformulate the second-order target problem they introduced in [\textit{H. M. Soner} and \textit{N. Touzi}, SIAM J. Control Optim. 48, No. 4, 2344--2365 (2009; Zbl 1229.91324)]. Their reformulation concerns two aspects: First, the target constraint is reinforced by requiring that it is satisfied under various mutually singular measures and not only on the support of the governing Brownian motion, and, second, the choice of a more natural set of admissible controls here avoids technical aspects. This reformulations allows the authors to establish a dual formulation of the target problem as supremum of the solutions of standard backward stochastic differential equations. Although strongly inspired by the works of \textit{L. Denis} and \textit{C. Martini} [Ann. Appl. Probab. 16, No. 2, 827--852 (2006; Zbl 1142.91034)] on quasi-sure stochastic analysis and by \textit{S. Peng} on \(G\)-Brownian motion [Abel Symposia 2, 541--567 (2007; Zbl 1131.60057); Stochastic Processes Appl. 118, No. 12, 2223--2253 (2008; Zbl 1158.60023)], the authors provide an alternative approach based on general results on the aggregation of random variables, see [\textit{M. H. Soner} et al., Electron. J. Probab. 16, 1844--1879 (2011; Zbl 1245.60062)]. The authors' purely probabilistic approach doesn't require a restriction to the Markovian framework.