Numerical identification of the fractional derivatives in the two-dimensional fractional cable equation (Q2631052)

The two-dimensional fractional cable equation \[ \begin{aligned} \frac{\partial u}{\partial t} = {}_0 D_t^{1-\gamma_1} \Big(K_1 \frac{\partial^2 u}{\partial x^2} + K_2 \frac{\partial^2 u}{\partial y^2} \Big) - \mu^2 {}_0 D_t^{1-\gamma_2} u + f(x,y,t),\\ (x,y) \in \Omega, \;0 \leq t \leq T,\end{aligned} \] is considered, with initial and boundary conditions of the form \( u (x,y,0) = \phi(x,y) \) for \( (x,y) \in \Omega \), and \( u (x,y,t) = \psi(x,y,t) \) for \( (x,y) \in \partial \Omega\), \(0 \leq t \leq T\), respectively, where \( 0<\gamma_1, \gamma_2 <1\), \(\Omega = [0, L ] \times [0,L]\), and \( K_1, K_2 \) and \( \mu \) are positive constants. In addition, \( \phi, \psi \) and \( f \) are sufficiently smooth functions, and \[ _0 D_t^{1-\gamma} u(x,y,t) = \frac{1}{\Gamma(\gamma)} \frac{\partial }{\partial t} \int_0^t \frac{u(x,y,\xi)}{(t-\xi)^{1-\gamma}} d\xi, \quad 0< \gamma< 1, \] denotes the Riemann-Liouville time fractional derivative of order \( 1 - \gamma \). The aim of this paper is to propose an efficient numerical method for solving a related inverse problem, i.e., determining the orders \( \gamma_1, \gamma_2 \) of the fractional derivatives. For the corresponding direct problem, a compact finite difference scheme is introduced which is of first order in time and fourth order in space. Its stability is also established using the Fourier method. The solution of the inverse problem is obtained by the Levenberg-Marquardt method for solving the least squares problem \([{\mathbf{Y}} - {\mathbf{U}}({\mathbf{P}})]^T [ {\mathbf{Y}} - {\mathbf{U}}({\mathbf{P}})] \to \min \), where \( \mathbf{Y} = [y_1, \dots, y_n]^T \) is the vector of measured data, \( \mathbf{U}(\mathbf{P}) = [U_1(\mathbf{P}), \dots, U_n(\mathbf{P})]^T \) contains the estimated values obtained from the direct problem, and the vector \( \mathbf{P} = [\gamma_1,\gamma_2]^T \) contains the unknown parameters. Finally, results of some numerical experiments are presented.