A sampling theorem for transforms with discontinuous kernels

Nonlocal and multipoint boundary value problems for ordinary differential equations (34B10) Eigenfunctions, eigenfunction expansions, completeness of eigenfunctions of ordinary differential operators (34L10) General integral transforms (44A05) Scattering theory of linear operators (47A40) Sampling theory in information and communication theory (94A20)

Review text

It is known that some boundary-value problems, such as Sturm-Liouville problems, self-adjoint boundary-value problems with simple eigenvalues, and Dirac systems, give rise to sampling theorems that are akin to the Whittaker-Shannon-Kotel'nikov sampling theorem for bandlimited functions. The bridge between boundary-value problems and sampling theorems is Kramer's sampling theorem which was introduced by Kramer in 1959. In this paper the authors give an example of another type of boundary-value problems that give rise to a sampling theorem. The boundary-value problem is given by the differential equation \(\ell(y)= \lambda y\), together with the boundary conditions \(U_\nu(y)= 0\) and the compatibility conditions \(V_\nu(y)= 0\), \(1\leq\nu\leq n\), where the differential operator \(\ell(\cdot)\) is defined by \[ \ell(y):= \begin{cases} \ell_1(y),\quad &\text{for }-1\leq x< 0,\\ \ell_2(y),\quad &\text{for }0< x\leq 1,\end{cases} \] \[ \ell_i(y)= y^{(n)}+ \sum^n_{j=1} p_{ij}(x) y^{(n-j)}(x),\quad i= 1,2, \] with the coefficients \(p_{ij}(\cdot)\), \(1\leq j\leq n\) satisfying the smoothness conditions, \[ p_{1j}(\cdot)\in C^{n-j}([-1, 0]);\quad p_{2j}(\cdot)\in C^{n-j}([0, 1]). \] The linear forms generating the boundary and compatibility conditions \(U_\nu(y)= 0\) and \(V_{\nu}(y)= 0\), \(1\leq\nu\leq n\), are defined by \[ U_\nu(y)= U_{\nu,-1}(y_1)+ U_{\nu, 1}(y_2), \] \[ U_{\nu,-1}(y_1)= \alpha_\nu y^{(k_\nu)}_1(- 1)+ \alpha_{\nu, k_\nu- 1} y^{(k_\nu- 1)}_1(- 1)+\cdots, \] \[ U_{\nu, 1}(y_2)= \beta_\nu y^{(k_\nu)}_2(1)+ \beta_{\nu,k_\nu- 1} y^{(k_\nu- 1)}_2(1)+\cdots; \] \[ V_\nu(y)= V_{\nu,0^-}(y_1)+ V_{\nu,0^+}(y_2), \] \[ V_{\nu,0^-}(y_1)= \gamma_\nu y^{(l_\nu)}_1(0^-)+ \gamma_{\nu, l_\nu-1} Y^{(l_\nu- 1)}_2(0^-)+\cdots, \] \[ V_{\nu,0^+}(y_2)= \delta_\nu y^{(l_\nu)}(0^+)+ \delta_{\nu, l_\nu-1} y^{(l_\nu- 1)}_2(0^+)+ \cdots. \] The numbers \(k_\nu\) and \(l_\nu\), are the orders of \(U_\nu\) and \(V_\nu\), and they satisfy \[ n- 1\geq k_1\geq k_2\geq\cdots\geq k_n,\quad k_\nu> k_{\nu+1}, \] \[ n- 1\geq l_1\geq l_2\geq\cdots\geq l_n,\quad l_\nu> l_{\nu+1}. \] Moreover, \(\alpha_\nu\), \(\beta_\nu\), \(\gamma_\nu\) and \(\delta_\nu\) are assumed to satisfy \[ |\alpha_\nu|+ |\beta_\nu|> 0,\quad |\gamma_\nu|+ |\delta_\nu|> 0,\quad 1\leq \nu\leq n. \] The sampling theorem takes the form: If \(f\in L^2[-1,1]\) and \[ F(\lambda)= \int^1_{-1}\overline f(x)\Phi(x, \lambda)\,dx. \] Then \(F(\lambda)\) is an entire function of order \(1/n\) that admits the sampling representation \[ F(\lambda)= \sum^\infty_{k= 1} F(\lambda_k){\Delta(\lambda)\over (\lambda- \lambda_k)\Delta'(\lambda_k)}, \] where \(\Phi(x,\lambda)\) is a function that generates the eigenfunctions of the problem, \(\{\lambda_k\}\) are the eigenvalues, and \(\Delta(\lambda)\) is the characteristic determinant of the problem.

Recommendations

Cites work

Cited in

(22)

This page was built for publication: A sampling theorem for transforms with discontinuous kernels

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4664039)