A chain rule in \(L^1(\text{div};\Omega)\) and its applications to lower semicontinuity (Q1424341)

Let \(\Omega\) be a bounded open set, and let \(L^1(\text{div};\Omega)=\{u\in L^1(\Omega;{\mathbb R}^N) : \text{div }u\in L^1(\Omega)\}\), where div is the distributional divergence operator. The paper provides various results, under weak regularity conditions, related to the validity of the chain rule, and to its applications to the lower semicontinuity of integral functionals. In particular, if \(B\colon \Omega\times{\mathbb R}^d\to{\mathbb R}^N\) is Borel, \(u\in W^{1,1}(\Omega;{\mathbb R}^d)\), and \(v(x)=B(x,u(x))\) for \(x\in\Omega\), sufficient conditions are proposed providing that \(v\in L^1(\text{div};\Omega)\), and the validity of the formula \[ \text{div\,}v(x)=\text{div}_xB(x,u(x))+\text{ tr}(\nabla_uB(x,u(x))\nabla u(x)) \] for \({\mathcal L}^N\)-a.e. \(x\in\Omega\). This result is then applied to the study of the \(L^1(\Omega)\)-lower semicontinuity of the integral functional \[ F\colon u\in W^{1,1}(\Omega)\mapsto\int_\Omega f(x,u(x),\nabla u(x))dx, \] where \(f\) is a not necessarily coercive integrand, convex in the last set of variables. Two types of results are achieved under the assumption \[ \nabla_zf(\cdot,u,z)\in L^1_{\text{loc}}(\text{div};\Omega)\text{ with }\text{div}_x\nabla_zf\in L^1_{\text{loc}}(\Omega\times{\mathbb R}\times{\mathbb R}^N). \] In the first one, continuity of \(f\) in the \(u\) variable is assumed. In the second one, the same condition is assumed in the \(x\) variable. A survey on the story of the lower semicontinuity problem for the above functional is presented, starting from the pioneering results by J. Serrin, and several corollaries and applications of the results of the paper are furnished.