How to approximate solutions of variational inequalities with monotone mappings in a Banach space if the data are known approximately (Q2387995)

Let \(X\) be a uniformly convex and uniformly smooth Banach space with the dual space \(X^*\). Denote by \(J: X \to X^*\) the normalized dual mapping. Suppose that \(A: X \to X^*\) is a monotone hemicontinuous bounded operator and \(\Omega\) a closed convex subset of \(D(A)\). The paper is concerned with an ill-posed variational inequality \(\langle Ax-f, x-y \rangle \leq 0\) \(\forall y \in \Omega\) \((x \in \Omega)\). Suppose that instead of \(f\), \(A\) and \(\Omega\), their approximations \(f(t)\), \(A(t)\) and \(\Omega(t)\) are available. The author studies the approximating regularized problem \(\langle dJu(t)/dt, u(t)-y\rangle+ \langle Ju(t)-Jy,du(t)/dt\rangle+ \gamma(t) \langle A(t) u(t)+\alpha(t)Ju(t)-f(t),u(t)-y \rangle \leq 0\) \(\forall y \in \Omega(t)\), \(u(t) \in \Omega(t)\), \(t \geq t_0\); \(u(t_0)=u_0 \in \Omega(t_0)\), where \(\alpha(t), \gamma(t)>0\). Under the assumption that the regularized inequality has a unique bounded solution \(u=u(t)\) of class \(C^1[t_0,\infty)\), the convergence of \(u(t)\) \((t \to \infty)\) to the normal solution of the original inequality is established.