Li-Yau type estimates for a nonlinear parabolic equation on complete manifolds (Q982607)

The author is interested in the study of the following non-linear second order PDE \[ u_t=\Delta u-\nabla\phi\cdot \nabla\phi-a\, u\, \log u-q\, u\tag{\(\clubsuit\)} \] on \(M\times[0,\tau]\), where \(M\) is a complete, noncompact, Riemannian manifold with \(m\)-dimensional Bakry-Émery Ricci curvature (see the paper for a definition) bounded below. Furthermore, \(a\in{\mathbb R}\), \(\phi\) is a smooth function on \(M\), \(q\) a numerical function on \(M\times[0,\tau]\), \(C^2\) on \(M\) and \(C^1\) on \([0,\tau]\). The author refers to some previous results by \textit{L. Ma} [J. Funct. Anal. 241, No.~1, 374--382 (2006; Zbl 1112.58023)] and \textit{Y. Y. Yang} [Proc. Am. Math. Soc. 136, No.~11, 4095--4102 (2008; Zbl 1151.58013)]. The main result is a particular estimate of \textit{P. Li} and \textit{S.-T. Yau} [Acta Math. 156, 154--201 (1986; Zbl 0611.58045)] type for the gradient of \(f\equiv \log u\), where \(u\) is a smooth positive solution on \(M\times[0,\tau]\), that includes many well-known gradient results. The author emphasizes also that the method used to obtain his estimate originated from some previous works by S. T. Yau (1975), S. Y. Cheng and S. T. Yau (1975) and it is directly related to a work by P. Li and S. T. Yau [loc. cit.]. In particular, his estimate improves the result by Y. Y. Yang [loc. cit.]. This subject is related to the so-called gradient Ricci solitons, i.e., complete Riemannian manifolds \((M,g)\), that differ from Einstein ones by the Hessian of some smooth potential function \(f:M\to{\mathbb R}\): \(\text{Ric}(g)-\lambda\, g=-\mathrm{Hess}(f)\). If \(\lambda>0\) (resp., \(\lambda=0\), resp., \(\lambda<0\)) a gradient Ricci soliton is called shrinking (resp., steady, resp., expanding). Such manifolds arise in connection with Ricci flow and their classification is currently under investigation (see, e.g., works by R. Hamilton (1998) and C. Böhm and B. Wilking (2008)). The author states that his local gradient estimate may be useful in understanding Ricci solitons. The paper after a detailed introduction, where the main results are summarized and related to some previous works on the same subject, splits into three more sections: 2. A basic lemma; 3. Proof of Theorem 1.1 (containing the main result); 4. Proof of Corollary 1.3. Reviewer's remark: It may be useful to emphasize that the algebraic topological methods introduced by the reviewer of this paper to characterize PDEs, and to study the Ricci flow equation, can be usefully applied also to equation (\(\clubsuit\)) [see for example the reviewer, ``Extended crystal PDE's'', \url{arXiv:0811.3693v13}]. Furthermore, it is interesting to recast equation \(\text{Ric}(g)-\lambda\, g+\mathrm{Hess}(f)=0\) defining a gradient Ricci soliton in the geometric theory of PDEs, too. In fact, it identifies an analytic submanifold \(E_2\subset J{\mathcal D}^2(W)\), \(F_{ij}\equiv u_{ij}-h_{ij}:J{\mathcal D}^2(W)\to {\mathbb R}\), \(1\leq i,j\leq n\), with \(h_{ij}\equiv-\lambda g_{ij}+ \text{Ric}(g)_{ij}\), over the trivial vector, analytic, fiber bundle \(\pi:W\equiv M\times{\mathbb R}\to M\), \((x^i,u)\mapsto (x^i)\). \(E_2\) is an affine involutive, formally integrable and completely integrable PDE of second order. So, in suitable neighborhoods of any point \(q\in E_2\), there exist analytic solutions, i.e., analytic functions \(f:U\subset M\to{\mathbb R}\), solutions of \(E_2\). Since the symbols of \((E_2)_{+r}\), for \(r\geq 0\), are zero, it follows that such solutions are stable in finite times. (Here, \((E_2)_{+r}\) is the \(r\)-prolongation of \(E_2\) and the above referred times are ones of the characteristic vector fields for the solutions considered.) This means that locally any Riemannian manifold can be considered as a gradient Ricci soliton. Different is the existence of global solutions, where the algebraic topological characterization of \(E_2\) is necessary in order to be able to talk about. There the \((n-1)\)-dimensional integral bordism group, \(\Omega^{E_2}_{n-1}\), for smooth solutions and integral bordism group, \(\Omega^{E_2}_{n-1,w}\), for weak solutions, bording smooth \((n-1)\)-dimensional smooth Cauchy manifolds of \(E_2\), are related by the following short exact sequence: \(0\to K^{E_2}_{n-1,w}\to \Omega^{E_2}_{n-1}\to \Omega^{E_2}_{n-1,w}\to 0\) with \(K^{E_2}_{n-1,w}\equiv\{[N]\in \Omega^{E_2}_{n-1}\, |\, N=\partial V\, \text{for some weak solution }V\)