On the Harnack inequality for a class of hypoelliptic evolution equations (Q4813853)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2091247
Language Label Description Also known as
default for all languages
No label defined
    English
    On the Harnack inequality for a class of hypoelliptic evolution equations
    scientific article; zbMATH DE number 2091247

      Statements

      On the Harnack inequality for a class of hypoelliptic evolution equations (English)
      0 references
      0 references
      0 references
      13 August 2004
      0 references
      variational argument
      0 references
      fundamental solution
      0 references
      optimal constant
      0 references
      The authors consider solutions of the degenerate evolution equation \[ Lu=\text{div}(ADu)+\langle x, BDu\rangle-\partial_t u=0 \] where \(u=u(x,t)\), \((x,t)\in\mathbb{R}^n\times\mathbb{R}.\) Here \(A=(a_{ij})_{n\times n}\) and \(B=(b_{ij})_{n\times n}\) are real constant matrices described as follows. For \(r\leq n\) and for each \(k=0,1,2,\dots ,r-1\), let \(m_k\geq m_{k+1}\geq 1\) be such that \(\sum_{k=0}^r m_k=n\). They assume that \(A\) is symmetric and \(a_{ij}=a_{ji}=0\), \(\forall i\geq m_0+1\) and \(\forall\;1\leq j\leq i.\) Also \(A_0=(a_{ij})_{m_0\times m_0}\) is a strictly positive definite matrix. The matrix \(B\) satisfies (a) (i) \(b_{ij}=0\), \(i=1,\dots ,n\), \(j=1,\dots ,m_0\); for each \(k=1,2,\dots ,r\), (ii) \(b_{ij}=0\), \(i=1+m_0+\cdots +m_{k-1},\dots ,n\), \(j=1,\dots ,m_0+\cdots +m_k,\) and (iii) \(b_{ij}=0\), \(i=1,\dots ,m_0+\cdots+ m_{k-1}\), \( j=1+m_0+\cdots +m_k,\dots ,n.\) The operator \(L\) satisfies Hormander's rank condition and is hypoelliptic. Such operators arise in a wide class of stochastic processes and kinetic models. The authors in this work provide a short and a direct proof of the Harnack inequality for positive solutions of \(Lu=0\), i.e., \(u(x_2,t_2)\geq cu(x_1,t_1)\), \((x_1,t_1)\), \((x_2,t_2)\in\mathbb{R}^n\times\mathbb{R}\), \(t_1<t_2\), where \(c=c(x_1,x_2,t_1,t_2)>0.\) This was previously proven by Kupcov, Garofalo-Lanconelli and Lanconelli and Polidoro. In this work, a proof is worked by adapting a variational argument originally due to Li and Yau. The argument utilizes a gradient estimate \(-Yu+Qu/(2t)\geq \langle ADu,Du\rangle/u,\) where \(Y=\langle x,BDu\rangle-\partial _t,\) and \(Q=m_0+3m_1+\dots +(2r+1)m_r.\) This is shown by employing a representation formula for positive solutions that involves the fundamental solution. The authors also find the optimal constant \(c\) in the Harnack inequality which includes the sharp versions known for some interesting cases such as the heat equation. This is also extended to the case of a Lie algebra of step two and three, considered previously by other authors. The work also makes use of \(L\)-admissible paths to prove various results.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references