Hans Werner Van Wyk | Searcharxiv

arXiv2020

A uniqueness determination of the fractional exponents in a three-parameter fractional diffusion

In this article, we consider the space-time Fractional (nonlocal) diffusion equation $$\partial_t^βu(t,x)={\mathtt{L}_D^{α_1,α_2}} u(t,x), \ \ t\geq 0, \ x\in D, $$ where $\partial_t^β$ is the Caputo fractional derivative of order $β\in (0,1)$ and the differential operator ${\mathtt{L}_D^{α_1,α_2}}$ is the generator of a Lévy process, sum of two symmetric independent $α_1-$stable and $α_2-$stable processes and ${D}$ is the open unit interval in $\mathbb{R}$. We consider a nonlocal inverse problem and show that the fractional exponents $β$ and $α_i, \ i=1,2$ are determined uniquely by the data $u(t, 0) = g(t),\ 0 < t < T.$ The uniqueness result is a theoretical background for determining experimentally the order of many anomalous diffusion phenomena, which are important in many fields, including physics and environmental engineering. We also discuss the numerical approximation of the inverse problem as a nonlinear least-squares problem and explore parameter sensitivity through numerical experiments.

math.AP↗

arXiv2019

Simultaneous inversion for the fractional exponents in the space-time fractional diffusion equation $\partial_t^βu= -\big(-Δ\big)^{α/2}u-\big(-Δ\big)^{γ/2}u$

In this article, we consider the space-time fractional (nonlocal) equation characterizing the so-called "double-scale" anomalous diffusion $$\partial_t^βu(t, x) = -(-Δ)^{α/2}u(t,x) - (-Δ)^{γ/2}u(t,x) \ \ t> 0, \ -1<x<1, $$ where $\partial_t^β$ is the Caputo fractional derivative of order $β\in (0,1)$ and $0<α\leq γ<2.$ We consider a nonlocal inverse problem and show that the fractional exponents $β$, $α$ and $γ$ are determined uniquely by the data $u(t, 0) = g(t), \ 0 < t \leq T.$ The existence of the solution for the inverse problem is proved using the quasi-solution method which is based on minimizing an error functional between the output data and the additional data. In this context, an input-output mapping is defined and its continuity is established. The uniqueness of the solution for the inverse problem is proved by means of eigenfunction expansion of the solution of the forward problem and some basic properties of fractional Laplacian. A numerical method based on discretization of the minimization problem, namely the steepest descent method and a least squares approach, is proposed for the solution of the inverse problem. The numerical method determines the fractional exponents simultaneously. Finally, numerical examples with noise-free and noisy data illustrate applicability and high accuracy of the proposed method.

math.AP↗