Very slow stabilization for a nonlinear Fokker-Planck equation (Q272735)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Very slow stabilization for a nonlinear Fokker-Planck equation
scientific article

    Statements

    Very slow stabilization for a nonlinear Fokker-Planck equation (English)
    0 references
    0 references
    0 references
    20 April 2016
    0 references
    The authors start by considering the Cauchy problem for the fast diffusion equation \[ u_{\tau} = \nabla \cdot(u^{m-1} \nabla u),\quad u(y,0) = u_0(y)\tag{1} \] where \(y \in \mathbb{R}^n\) and \(\tau \in (0,T)\). It is assumed that \(m<1\) and \(u_0\) is continuous and bounded. It is known that if \(m< m_c = (n-2)/n\) all solutions of (1) with initial data in some suitable function space vanish in finite time. The authors study the rate of extinction of such solutions for \(m \leq m_* = (n-4)/(n-2)\), \(n>2\). It is noted that for \(m<m_c\) equation (1) has self-similar solutions called generalized Barenblatt solutions which are denoted \(U_{D,T}(y, \tau)\) for a parameter \(D \geq 0\). These functions have a decay rate near extinction of the form \(\| u(\cdot,\tau)\|_{\infty} = O(T-\tau)^{\gamma}\) where \(T\) is the extinction time and \(\gamma\) is a parameter which differs depending on whether \(D> 0\) or \(D = 0\). The analysis in the paper proceeds by making a change of variables in (1) to write the equation in the form of a nonlinear Fokker-Planck equation. For \(R(\tau) = (T-\tau)^{-\beta}\), \(\mu = 2/(1-m)\), \(\beta = 1/(n(m_c-m))\) let \[ v(x,t) = R(\tau)^nu(y,\tau),\quad t = \frac{1}{\mu}\log \Big(\frac{R(\tau)}{R(0)} \Big),\quad x = \sqrt{\frac{\beta}{\mu}} \frac{y}{R(\tau)}. \] Then equation (1) is transformed as \[ v_t = \nabla \cdot (v^{m-1}\nabla v) + \mu \nabla \cdot (xv),\quad x \in \mathbb{R}^n,\quad t>0,\tag{2} \] and the Barenblatt solutions \(U_{D,T}(y,\tau)\) become stationary solutions of (2) given by \[ V_D(x) = \Big (D + |x|^2 \Big )^{-\mu/2}. \] The goal of the paper is to show that there are initial functions such that the rate of convergence of \(v(x,t)\) to \(V_D(x)\) is arbitrarily slow as \(t \rightarrow \infty\). Previously rates of convergence obtained for (2) have been exponential or algebraic. An example of slow convergence is \(1/\log(\dots (\log t))\). A formal definition of slow decay is given. Then several theorems are proved on the slow decay of the solutions of (2) to \(V_D(x)\). Two theorems are proved for the convergence to regular Barenblatt profiles for which \(D>0\). One is for \(m<m*\), the other for \(m = m*\). Two more theorems are proved, for \(m<m*\) and \(m = m*\) respectively, on the convergence to the singular Barenblatt solution for which \(D = 0\). It is shown that the rates of convergence for the subcritical case, \(m< m*\), are different from the convergence rates for the critical case, \(m = m*\). As a consequence of results obtained on slow convergence to the singular steady state, \(V_0(x)\), the authors find a new class of extinction rates for equation (1).
    0 references
    0 references
    0 references
    0 references
    0 references
    fast diffusion equation
    0 references
    nonlinear Fokker-Planck equation
    0 references
    rate of convergence to steady state
    0 references
    0 references