On the mean curvature flow of grain boundaries (Q1687857)

A family of \(n\)-dimensional surfaces \(\{\Gamma(t)\}_{t\geq 0}\) in \(\mathbb R^{n+1}\) is called the mean curvature flow (abreviated by MCF) if the velocity is equal to its mean curvature at each point and time. The thorough study of this structure began in 1978 with an ample work by \textit{K. A. Brakke} [The motion of a surface by its mean curvature. Tokyo: University of Tokyo Press (1978; Zbl 0386.53047)]. The MCF structure appears interesting and useful in fields such as differential geometry, general relativity, image processing and material science. This explains the large number of authors and articles that refer to this structure, among them the second author of this article. He is the author or co-author of several articles on MCF [Math. Ann. 364, No. 3--4, 857--935 (2016; Zbl 1351.53083); Hiroshima Math. J. 33, No. 3, 323--341 (2003; Zbl 1059.35061); Adv. Calc. Var. 7, No. 1, 91--138 (2014; Zbl 1283.53064); Arch. Ration. Mech. Anal. 221, No. 3, 1161--1222 (2016; Zbl 1347.35048)] Given a smooth surface \(\Gamma_0\), one can find a smoothly moving MCF starting from \(\Gamma_0\) until some singularities such as vanishing or pinching occur. The authors call this type of MCF structure a Brakke flow since Brakke first proved a global-in-time existence of such structure. However some researchers raised objections to Brakke's theorem, namely that there is no guarantee that the MCF he obtained is always nontrivial. For this reason, rigorous global-in-time existence of MCF of grain boundaries has been considered completely open among the specialists. In order to answer this problem, in the present paper the authors rephrase and modify the approximation scheme so that they always obtain a nontrivial MCF even if \(\Gamma_0\) is singular. The authors say that they prove ``for the first time a rigorous global-in-time existence theorem of the MCF of grain boundaries which was not known before even for the 1-dimensional case''. The main existence theorem proved in this paper is stated by the authors in Introduction as follows: Theorem 1.1. Let \(n\) be a natural number and suppose that \(\Gamma_0 \subset \mathbb R^{n+1}\) is a closed countably \(n\)-rectifiable set whose complement \(\mathbb R^{n+1}\backslash \Gamma_0\) is not connected. Assume that the \(n\)-dimensional Hausdorff measure of \(\Gamma_0\) is finite or grows at most exponentially near infinity. Let \(E_{0,1},\dots ,E_{0,N}\subset \mathbb R^{n+1}\) be mutually disjoint non-empty open sets with \(N\geq 2\) such that \(\mathbb R^{n+1}\backslash \Gamma_0=\cup_{i=1}^NE_0\). Then, for each \(i=1,\dots ,N\), there exists a family of open sets \(\{E_i(t)\}_{t\geq 0}\) with \(E_{i}(0)=E_{0,i}\) such that \(E_1(t),\dots ,E_N(t)\) are mutually disjoint for each \(t\geq 0\) and \(\Gamma(t):=\mathbb R^{n+1}\backslash \cup_{i=1}^NE_i(t)\) is a MCF with \(\Gamma(0)=\Gamma_0\), in the sense that \(\Gamma(t)\) coincides with the space-time support of a Brakke flow starting from \(\Gamma_0\). Each \(E_i(t)\) moves continuously in time with respect to the Lebesgue measure. The proof of Theorem 1.1, reformulated more technically in Theorem 3.5, occupies much of this voluminous article. In addition, the authors make numerous modifications, simplifications and new proofs of some results from Brakke's work [loc. cit.]. These additional original contributions refer to the smoothing of varifold, the integrality theorem, Brakke's inequality (used in the proof of Theorem 3.5/Theorem 1.1) etc. A rectifiability theorem in the sense of \textit{W. K. Allard} [Ann. Math. (2) 95, 417--491 (1972; Zbl 0252.49028)] is also proved. The last paragraph of the paper includes some additional comments on tangent flow, a short-time regularity and ``other settings''.