Constructing a solution of a two-point boundary value problem for a nonlinear matrix Lyapunov equation (Q2188064)

In this paper, the authors study the following boundary value problem \[ dX/dt=A\left( t\right) X+XB\left( t\right) +F\left( t,X\right),\text{ }X\in \mathbb{R}^{n\times n}, \] with \(MX\left( 0\right) +NX\left( \omega \right) =0,\) where \(A,B\in C\left( I,\mathbb{R}^{n\times n}\right) ,\) \(F\in C\left(D_{\rho },\mathbb{R}^{n\times n}\right) ,\) \(I\in \left[ 0,\omega \right] ,\) \(D_{\rho }=\left\{ \left( t,X\right) :t\in I,\left\| X\right\| <\rho \right\} ,\) \(\omega >0,\) \(0<\rho \leq \infty ,\) the function \(F\left( t,X\right) \) (locally) satisfies the Lipschitz condition with respect to \(X\) in the domain \(D_{\rho },\) and \(M\) and \(N\) are real \(n\times n\) matrices. Let \[ R=N^{-1}\left( M+N+N\int_{0}^{\omega }A\left( \sigma \right) d\sigma \right)\text{, }S=-\int_{0}^{\omega }B\left( \sigma \right) d\sigma \text{, }\Phi X=RX-XS, \] \[ m=\left\| N^{-1}\left( M+N\right) \right\| ,\gamma =\left\| \Phi ^{-1}\right\| ,\alpha =\max_{t\in I}\left\| A\left( t\right) \right\| ,\beta =\max_{t\in I}\left\| B\left( t\right) \right\| ,h=\max_{t\in I}\left\|F\left( t,0\right) \right\| \] and \[ q=\gamma \omega \left( \left( \alpha +\beta +L\right) \left( m+\frac{\left(\alpha +\beta \right) \omega }{2}\right) +L\right) ,\text{ }p=\gamma \omega h\left( m+\frac{\left( \alpha +\beta \right) \omega }{2}+1\right), \] where \(L\) is the Lipschitz constant of the function \(F\left( t,X\right) \) in the domain \(D_{\rho _{1}}\) with \(\rho_1<\rho\). The main result in this paper is that if \(\det N\neq 0\), \(q<1\), \(\frac{p}{q-1}\leq\rho _{1}\) and the matrices \(R\) and \(S\) do not have common characteristic numbers, then there exists a unique solution \(X\left( t\right) \) of the studied problem in the domain \(D_{\rho _{1}}.\) Moreover, this solution can be represented as the limit of a sequence of matrix functions uniformly convergent on the interval \(I\). These matrix functions are determined by a recursion integral relation with an implicit computational scheme with the estimate \(\left\| X\right\| _{C}\leq \frac{p}{q-1}\).