Representation theorems for generators of backward stochastic differential equations (Q1764117)

The author of this short note studies the representation of the generator of a backward stochastic differential equation (BSDE, for short) with the help of its solution.\ To be more precise, given a \(d\)-Brownian motion \(B\) defined on some probability space \((\Omega,\mathcal{F},P)\) endowed with the filtration \((\mathcal{F}_{t})\) generated by \(B\), let \(g:\Omega\times[0,T]\times R\times R^{d}\rightarrow R\) be a progressively measurable process such that \(g(\omega,t,.,.)\) is Lipschitz, uniformly with respect to \((\omega,t)\), and \(E[\int_{0}^{T}| g(t,0,0)| ^{2}dt]<+\infty\). \textit{P. Briand, F. Coquet, Y. Hu, J. Mémin} and \textit{S. Peng} [Electron. Commun. Probab. 5, 101--117 (2000; Zbl 0966.60054)] have shown that, under the additional assumption of continuity of \(t\rightarrow g(t,y,z)\) and a more restrictive integrability assumption, the driver \(g\) of the BSDE can be represented as \[ g(t,y,z)=L^{2}-\lim_{\varepsilon\downarrow0}\varepsilon ^{-1}(Y_{t}(t+\varepsilon,y+z(B_{t+\varepsilon}-B_{t}))-y), \quad (t,y,z)\in [0,T)\times R\times R^{d}, \] where \((Y_{.}(\tau,\xi),Z_{.}(\tau,\xi))\) denotes the unique solution of the BSDE \[ dY_{t}=-g(t,Y_{t},Z_{t} )dt+Z_{t}dB_{t},\;t\in[0,\tau], \;Y_{\tau}=\xi\in L^{2}(\mathcal{F}_{\tau }). \] The author of the present note renounces on these supplementary conditions and studies equivalent conditions to this representation property of \(g\).\ He shows in particular that this representation property is equivalent to the fact that \(g\) is a Lebesgue generator, i.e., \[ g(t,y,z)=L^{2} -\lim_{\varepsilon\downarrow0}E\left[\varepsilon^{-1}\int_{t}^{t+\varepsilon }g(s,y,z)ds\Bigl| \mathcal{F}_{t}\right], \] for all \((t,y,z).\) This property is extended to \(L^{p}\)-convergence, for \(1\leq p\leq2\).