Determination of diffusion kernel on fractals (Q4533674)

Diffusion on fractals exhibits many anomalies due to the geometrical constraints imposed by the complex structure of the diffusion process. More important analytical calculations show that on a large class of fractal structures the probability \(P(r,t)\) has asymptotically a non-Gaussian shape of the form \[ P(r,t)\sim t^{d_f/d_w} \exp[-\text{const} (r/R)^u], \] where \(r/R\gg 1\) and \(t\to\infty\), \(u={d_w\over{d_w-1}}\), \(d_w\) is the anomalous diffusion exponent of the diffusion set, and \(d_f\) is the fractal dimension. If a diffusion fractal and a probability vector \(p=(p_1,\dots,p_k)\) are given then the diffusion kernel \(K(t)\) is approximately determined. However, in general, the probability vector \(p=(p_1,\dots,p_k)\) is not known. How is the corresponding diffusion kernel determined? The main result of the paper is the following. If the diffusion fractal is a Cantor type set, whether the generating mappings of the diffusion set are linear or nonlinear, the concrete expressions for the diffusion kernel are obtained. Moreover, the fractional diffusion exponent \(\gamma\) of the diffusion kernel must satisfy \(0<\gamma<1\).