On a method for construction of successive approximations for investigation of multipoint boundary-value problems (Q4940543)

This paper deals with the boundary value problem \[ \dot{x}=f(t,x), \quad t\in [0,T], \qquad \sum_{i=0}^{N}A_ix(t_i)-d=0 \quad (t_0=0,\;t_N=T), \] with \(f:[0,T]\times (D\subset{\mathbb{R}}^n)\mapsto {\mathbb{R}}^n\) and \(A_i\) are constant \(n\times n\)-matrices. It is supposed that the numbers \(k_0,\ldots ,k_N\) exist such that \(H=\sum_{i=0}^{N}k_iA_i\) is nondegenerate. The problem is reduced to the system of equations \(x=Q_{\xi }x\), \(\Delta _\xi (x)=0\), with \((Q_\xi x)(t)=\xi +\int_{T/2}^{t}f(s,x(s)) ds+l(t)H^{-1}\Delta _\xi (x)\), \(\Delta _\xi (x)=d-\sum_{i=0}^{N}A_i(\xi +\int_{T/2}^{t_{i}}f(s,x(s)) ds)\), and \(l(t)\) is an interpolation polynomial taking the values \(k_i\) at points \(t_i,\;i=1,\ldots ,N\), \(\xi \in D\). The authors obtain sufficient conditions for the convergence of the successive approximations \(x_m(t,\xi)=Q_\xi x_{m-1}(t,\xi)\) and for the solvability of the determining equation \(\Delta _{\xi }(x_\infty (t,\xi))=0\).