scientific article; zbMATH DE number 1355370

From MaRDI portal
Publication:4269474





Let \((\Omega,\Sigma,\mu)\) be a probability space and \(T(x), x\in \mathbb{R}^n\) be a dynamical system acting on \(\Omega\). In a bounded Lipschitz domain \(Q\subset \mathbb{R}^n\) the authors consider the problem \[ -\text{div}_x(A(x,T(\varepsilon^{-1}x)\omega)\nabla_xu_\varepsilon)=f ,\quad x\in Q,\quad u_\varepsilon=u_0,\quad x\in \partial Q\tag{1} \] with \(f\in L^p(Q)\), \(u_0\in W^{1-{1/p},p}( \partial Q)\) and \(\varepsilon>0\). They develop the stochastic vector calculus, introduce the notions of stochastic gradient \(\nabla_\omega f\) and stochastic divergence \(\text{div}_{\omega,p}v\) and apply them to homogenization problems. Assuming that (1) has a unique solution for each \(\omega\in \Omega\) satisfying some additional assumptions they prove that there exists \((u, \xi)\) such that \(\{(u_\varepsilon,\nabla_xu_\varepsilon)\}\) converges in the mean value to \((u, \xi)\) as \(\varepsilon\to 0\). In addition \((u, \xi)\) is proved to be a unique solution to the problem \[ \text{div}_{\omega,p}(A(x,\cdot)[\xi(x,\cdot)+\nabla_xu(x,\cdot)])=0,\;\text{for a.e.} x\in Q, \] \[ -\text{div}_x \{\tilde E_p(A(\xi+\nabla_x u))(\cdot,\omega)\},\;x\in Q ,\text{ for } \mu\text{-a.e. }\omega\in \Omega, \] \[ u(\cdot,\omega)=u_0, \text{ for } \mu\text{-a.e. }\omega\in\Omega. \] Here \(\tilde E_pu=E_p[u(x, \cdot)](\omega)\) for \(dx\times d\mu\)-a.e. \((x,\omega)\in Q\times \Omega\). Similar results are proved for various elliptic boundary value problems with \(L^p\)-data for both scalar equations and systems of equations in domains which are either unperforated or randomly perforated.




Cited in
(24)








This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4269474)