English

Causal diffusion and its backwards diffusion problem

Analysis of PDEs 2013-08-05 v2 Mathematical Physics math.MP Numerical Analysis

Abstract

This article starts over the backwards diffusion problem by replacing the \emph{noncausal} diffusion equation, the direct problem, by the \emph{causal} diffusion model developed in \cite{Kow11} for the case of constant diffusion speed. For this purpose we derive an analytic representation of the Green function of causal diffusion in the wave vector-time space for arbitrary (wave vector) dimension NN. We prove that the respective backwards diffusion problem is ill-posed, but not exponentially ill-posed, if the data acquisition time is larger than a characteristic time period τ\tau (2τ2\,\tau) for space dimension N3N\geq 3 (N=2). In contrast to the noncausal case, the inverse problem is well-posed for N=1. Moreover, we perform a theoretical and numerical comparison between causal and noncausal diffusion in the \emph{space-time domain} and the \emph{wave vector-time domain}. The paper is concluded with numerical simulations of the backwards diffusion problem via the Landweber method.

Keywords

Cite

@article{arxiv.1107.3952,
  title  = {Causal diffusion and its backwards diffusion problem},
  author = {Richard Kowar},
  journal= {arXiv preprint arXiv:1107.3952},
  year   = {2013}
}

Comments

In the replacement I have rewritten the abstract and the introduction. Moreover, I have added Remark 1 and simplified a little bit the proof of Theorem 4. The reference 25 is updated, since the paper is now published

R2 v1 2026-06-21T18:39:21.390Z