Regularized extragradient method of finding a solution to an optimal control problem with inaccurately specified input data (Q2317217): Difference between revisions

The authors consider a minimization problem \[ J(u) = g^0(x(t_1;u)) \longrightarrow \inf \tag{1} \] subject to the Cauchy problem \[ \dot{x} = D(t)x(t) + B(t)u(t) + f(t), \quad t_0\le t\le t_1, \quad x(t_0) = x_0, \tag{2} \] under the inequality conditions \[ g^1(x(t_1;u)) \le 0, \ldots, g^m(x(t_1;u)) \le 0, \ldots, \quad u=u(t)\in U_0. \tag{3} \] In the case that the input data \(D(t)\), \(B(t)\), \(f(t,g^0(x),\ldots,g^m(x)\), and \(x_0\) are exactly known, the problem (1)--(3) is solved by the extragradient method [\textit{A. S. Antipin} et al., Comput. Math. Math. Phys. 57, No. 1, 64--70 (2017; Zbl 1373.49034); translation from Zh. Vychisl. Mat. Mat. Fiz. 57, No. 1, 49--54 (2017)]. However, in practice the input data contain errors. In this paper, the extragradient method is extended to the case with inaccurately specified input data. A stable method is developed for solving the problem (1)--(3) using iterative regularization. The convergence proof of the proposed method is given and a regularizing operator is constructed for approximating the uncertain input data.