Mappings of bounded mean distortion and cohomology (Q987362)

To state the main results, the author gives the following definitions: We say that a nonconstant continuous mapping \(f: \mathbb{R}^{n} \to N\) is a mapping of finite distortion if \(f\) belongs to the Sobolev space \(W_{loc}^{1, n}(\mathbb{R}^{n}; N)\), and there exists a measurable function \(K: \mathbb{R}^{n} \to [1, \infty)\) such that \[ |Df(x)|^{n} \leq K(x)J_{f}(x) \] for almost every \(x \in \mathbb{R}^{n}\). We define the outer distortion function \(K_{f}\) of \(f\) to be the function \(K_{f}(x)=|Df(x)|^{n} / J_{f}(x)\) whenever \(J_{f}(x)>0\), and \(K_{f}(x)=1\) otherwise. We say that a mapping of finite distortion \(f\) has \((K-)\) bounded \(p\)-mean distortion, \(p \geq 1\), if there exist constants \(K \geq 1\) and \(r_{0}>0\) such that \[ \displaystyle{\left( \frac{1}{|B^{n}(r)|} \int_{B^{n}(r)} K^{p}_{f}(x)dx \right)^{1/p} \leq K} \] for every \(r \geq r_{0}\). Here \(B^{n}(r)\) is the open ball of radius \(r\) centered at the origin in \(\mathbb{R}^{n}\), and \(|B^{n}(r)|\) is the Lebesgue \(n\)-measure of \(B^{n}(r)\). We also say that \(f\) has finite lower order \(\lambda\) if \[ \displaystyle{\lambda=\liminf\limits_{r \to \infty} \frac{\log A_{f}(r)}{\log \, r}< \infty }. \] Here and in what follows \(A_{f}\) is the averaged counting function \[ \displaystyle{A_{f}(r)=\int_{B^{n}(r)} J_{f}(x)dx}. \] Theorem 1. For every \(n \geq 2\), \(\lambda \geq 0\), and \(K \geq 1\), there exist constants \(p=p(n)>n-1\) and \(C=C(n, \lambda, K)>0\) with the following property. Let \(N\) be a closed, connected, and oriented Riemannian n-manifold, and let \(f: \mathbb{R}^{n} \to N\) be a mapping of K-bounded p-mean distortion having finite lower order \(\lambda\). Then dim \(H^{*}(N) \leq C\). The proof of the main theorem relies on two ingredients of possible independent interest. In section \(2\), the author gives a very simple proof for an extension of a special case of the Mattila-Rickman equidistribution theorem. For mappings of bounded mean distortion, the result reads as follows. In the statement of the theorem, the logarithmic measure \(m_{\log}(E)\) of a set \(E \subset (0, \infty)\) is \[ \displaystyle{m_{\log}(E)=\int_{E} \frac{dr}{r}}. \] Theorem 2. Let \(N\) be a closed, connected, and oriented Riemannian n-manifold, \(n \geq 2\), \(u \in L^{q}(N\)), \(q>n\), and suppose that \(f: \mathbb{R}^{n} \to N\) is a mapping of bounded \((n-1)\)-mean distortion. Then for every \(\varepsilon >0\), there exist \(r_{0} \geq 1\) and a set \(F \subset [1, \infty)\) such that \(m_{\log}\big([r/2, r] \setminus F\big)< \varepsilon\) for every \(r \geq r_{0}\), and \[ \displaystyle{\lim_{r \to \infty, r \in F} \frac{1}{A_{f}(r)} \int_{B^{n}(r)} (u \circ f)(x)J_{f}(x)dx= \frac{1}{|N|} \int_{N} u(y)dy, \,\,\,\,\,\,\,\,\, (1.1)} \] where the integral on the right is with respect to the Riemannian measure of \(N\), and \(|N|\) is the volume of \(N\). In section \(3\), the author considers Caccioppoli-type potential estimates for pull-backs of forms under mappings of finite distortion. Instead of focusing on the solutions of degenerate \(A\)-harmonic equations arising in the pull-back, the author considers pairs of closed forms \((\xi, \zeta)\) satisfying a nonnegativity condition \(\star (\xi \wedge \zeta) \geq 0\), where \(\star\) denotes the Hodge star duality operator. Such pairs arise naturally in linear and non-linear Hodge theory; the pairs \((\xi, \star \xi)\) and \((\xi, \star |\xi|^{p-2} \xi)\) are nonnegative if \(\xi\) is a harmonic or a p-harmonic form, respectively. These pairs are special cases of Cartan forms as considered by Hajlasz, Iwaniec, Maly, and Onninen. Using the equidistribution result and a Caccioppoli-type estimate, the author finishes the proof of Theorem \(1\) in section \(4\). The arguments closely follow the proof of Bonk and Heinonen. The main difference is in the replacement of conformal exponents by exponents within a range determined by a Sobolev-embedding theorem for differential forms.