Inverse coefficient problem for one-dimensional evolution equation vanishing initial condition

Oleg Y,Imanuvilov,Masahiro Yamamoto
2024-09-30
Abstract:We consider an inverse problem of determining a coefficient $p(x)$ of an evolution equation $\sigma\ppp_tu = a(x)\ppp_x^2u - p(x)u$ for $0<x<\ell$ and $0<t<T$, where $\sigma \in \C \setminus \{0\}$, $\ell>0$ and $T>0$ are arbitrarily given. Our main result is the uniqueness: by assuming that the zeros of initial value $b(x):= u(0,x)$ on $[0, \ell]$ is a finite set and each zero is of order one at most, if two solutions have the same Cauchy data at $x=0$ over $(0,T)$ and the same initial value $b(x)$, then the coefficient $p(x)$ is uniquely determined on $[0,\ell]$.
Analysis of PDEs
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the uniqueness problem of determining the spatial coefficient \(p(x)\) in the equation given the Cauchy data and initial values of a one - dimensional evolution equation. Specifically, the research object is the one - dimensional linear evolution equation: \[ \sigma\frac{\partial u}{\partial t}(t, x)=a(x)\frac{\partial^{2}u}{\partial x^{2}}(t, x)-p(x)u(t, x),\quad0 < t < T,\quad0 < x < \ell \] where the initial condition is: \[ u(0, x)=b(x),\quad0 < x < \ell \] The inverse problem considered in the paper is: **Given \(u(t, 0)\) and \(\frac{\partial u}{\partial x}(t, 0)\) for \(0 < t < T\), determine the value of the coefficient \(p(x)\) on \(0\leqslant x\leqslant\ell\).** The main result is the uniqueness theorem. That is, under certain assumptions, if two solutions have the same Cauchy data at \(x = 0\) and the same initial value \(b(x)\), then the coefficient \(p(x)\) is uniquely determined on the interval \([0,\ell]\). These assumptions include that the set of zeros of the initial value \(b(x)\) is a finite set and each zero is at most a zero of first order. In addition, the paper also discusses the uniqueness problem of the coefficient \(p(x)\) when \(b(x)\) is not identically zero in \((0,\ell)\), and solves the open problem when \(b(x) = 0\).