Skip to content
Preprint

Recovering the reaction coefficient in a time-fractional diffusion equation with variable coefficients in any spatial dimension

Oct 2026 · 0 citations · 27 references
Mathematics

Abstract

We explore the inverse problem of determining the reaction coefficient $q(x',t)$ in the time-fractional diffusion equation ${}^CD_t u-Lu+q(x',t)u=f$ in $\mathbb{R}^n\times(0,T]$, $x=(x',x_n)$, $0<\alpha<1$, from the trace of the solution on the hyperplane $\{x_n=0\}$. Here ${}^CD_t$ is the Gerasimov--Caputo derivative and $L$ is a uniformly elliptic second-order operator with variable, $x_n$-independent coefficients and without zero-order term. For $L=\Delta$ this problem was studied in dimensions $n=1,2,3$ by methods based on pointwise bounds of the kernels, which are not available for $n\ge4$. Our approach relies on the Eidelman--Kochubei fundamental solution: although its singularity on the diagonal depends on the dimension, we show that it satisfies integral bounds of the same form for all $n\ge1$. Using these bounds, we prove existence, uniqueness and stability of regular solutions of the direct problem with a H\"{o}lder continuous reaction coefficient. Without assuming additional smoothness of the solution, we show that the inverse problem is equivalent to a closed system of integral equations for $q$ and the derivatives $\partial_{x_n}u$, $\partial^2_{x_n}u$. For the inverse problem we establish a conditional Lipschitz stability estimate on the whole time interval and global uniqueness in every dimension, and prove local existence of a solution. In the case $a_{in}\equiv0$, $b_n\equiv0$ we obtain a stability estimate with explicit constants, which reduces to the known estimates for the Laplacian. An explicit two-dimensional example with a mixed coefficient and a normal drift shows that the hypotheses of the main theorems are compatible and that the corresponding terms in the trace condition cannot be omitted.

View source

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.