On a system of linear Volterra integral equations with variable integration limits (Q6114641)

The author studies a system of two-dimensional linear integral equations with variable lower and upper limits of integration. This is associated with the problem of identifying a nonlinear dynamical system represented as a finite segment (polynomial) of the Volterra integro-power series. The main integral equation under consideration has the form \[ y(t)= \int^t_0 K_1(t,\lambda_1) x(\lambda_1)d\lambda_1+\int^t_0 \int^t_0 K_2(\lambda_1,\lambda_2) x(t-\lambda_1)x(t-\lambda_2)d\lambda_1d\lambda_2, \] with \( t\in[0,T]\) and input signals (piecewise linear functions) \[ x^{\alpha_i}_\nu(t)=\begin{cases} 0, &t\le 0,\\ \frac{\alpha_it}{\nu}, &0<t\le\nu,\\ \alpha_i,&\nu\le t, \end{cases} \] where \(i=1,2\). Note that the \(y^{\alpha_i}\)'s are interpreted as the responses of the dynamical system to input signals \(x^{\alpha_i}_\nu\), \(i=1,2\). Substituting the input signals into the integral equation above one gets the following linear system of Volterra integral equations: \[ \alpha_1 \int^\nu_0 K_1(t,\lambda_1)\frac{\lambda_1}{\nu} d\lambda_1+\alpha_1 \int^t_\nu K_1(t,\lambda_1)d\lambda_1 \] \[ +\, \alpha^2_1 \int^t_{t-\nu} \int^t_{t-\nu} K_2(\lambda_1,\lambda_2)\frac{t-\lambda_1}{\nu} \frac{t-\lambda_2}{\nu} d\lambda_1d\lambda_2 +\alpha^2_1 \int^{t-\nu}_0 \int^{t-\nu}_0 K_2(\lambda_1,\lambda_2)d\lambda_1 d\lambda_2\] \[+\, 2\alpha^2_1 \int^{t-\nu}_0 d\lambda_1 \int^t_{t-\nu} K_2(\lambda_1,\lambda_2)\frac{t-\lambda_2}{\nu} d\lambda_2 =y^{\alpha_1}(t,\nu), \] \[ \alpha_2 \int^\nu_0 K_1(t,\lambda_1)\frac{\lambda_1}{\nu} d\lambda_1+\alpha_2 \int^t_\nu K_1(t,\lambda_1)d\lambda_1 \] \[+\, \alpha^2_2 \int^t_{t-\nu} \int^t_{t-\nu} K_2(\lambda_1,\lambda_2)\frac{t-\lambda_1}{\nu}\frac{t-\lambda_2}{\nu} d\lambda_1 d\lambda_2 +\alpha^2_2 \int^{t-\nu}_0 \int^{t-\nu}_0 K_2(\lambda_1,\lambda_2) d\lambda_1 d\lambda_2 \] \[+\, 2\alpha^2_2 \int^{t-\nu}_0 d\lambda_1 \int^t_{t-\nu} K_2(\lambda_1,\lambda_2) \frac{t-\lambda_2}{\nu} d\lambda_2 = y^{\alpha_2}(t,\nu). \] The main problem is to find the Volterra kernels \(K_1(t,\lambda_1)\), \(K_2(\lambda_1,\lambda_2)\) if \(x^{\alpha_i}_\nu(t)\) and \(y^{\alpha_i}(t,\nu)\) are given. The necessary and sufficient conditions that ensure the existence and uniqueness of \(K_1(t,\lambda_1)\), \(K_2(\lambda_1,\lambda_2)\) are given. The formulae for determining \(K_1(t,\lambda_1)\) and \(K_2(\lambda_1,\lambda_2)\) are found and some properties such as their regularity are discussed. The results are finally illustrated by considering the Volterra equation \[ y_{et}(t)= \int^t_0 x(s)ds+ \left(\int^t_0 x(s)ds\right)^2+\left(\int^t_0 x(s)ds\right)^3.\]