On the topology and index of minimal surfaces. II (Q6158175)

Since they were introduced in the work of \textit{J.-L. Lagrange} [Œuvres. Tome 1. Publiées par les soins de J.-A. Serret. Avec une notice sur la vie et les ouvrages de J.-L. Lagrange par J.-B. J. Delambre. Nachdruck der Ausgabe Paris. Georg Olms Verlag, Hildesheim-New York, 1973. LI+733 pp. (1867)] in \(1760\), minimal surfaces in Euclidean spaces have always occupied a special place in the heart of geometers. A minimal surface \(M^2\subset \mathbb{R}^3\) is the image of an immersion \(X:\Sigma\rightarrow \mathbb{R}^3\) from an orientable surface (the article under review has results on non-orientable surfaces, but we will stick to orientable surfaces here) for which the mean curvature \(H\) vanishes identically. Equivalently, \(M^2\) is minimal if any small enough compact domain \(D\subset M^2\) minimises the area amongst all surfaces with the same boundary \(\partial D\). An important class of minimal surfaces are surfaces of \emph{finite total curvature}, i.e., that satisfy \begin{align*} \frac{1}{2}\int_{\Sigma}|A|^2d\mathrm{vol}_{\Sigma}=\int_{\Sigma}-K\,d\mathrm{vol}_{\Sigma}<\infty, \end{align*} where \(A\) is the second fundamental form and \(K\) is the Gauss curvature. Equivalently, if \(N:\Sigma\rightarrow S^2\) is the unit normal associated to the minimal immersion (also known as the Gauss map), the condition is equivalent to the fineteness of its Dirichlet energy: \begin{align*} \frac{1}{2}\int_{\Sigma}|dN|_{\Sigma}^2d\mathrm{vol}_{\Sigma}<\infty. \end{align*} According to a celebrated theorem of \textit{A. Huber} [Comment. Math. Helv. 32, 13--72 (1957; Zbl 0080.15001)], provided that this condition on the Gauss map is verified, the surface \(\Sigma\) is conformally equivalent to a closed Riemann surface \(\bar{\Sigma}\) with finitely many punctures \(p_1,\cdots,p_n\) that are called the ends of the minimal surface. Furthermore, the Gauss map \(N\) extends continuously on \(\bar{\Sigma}\), which yields a smooth harmonic map \(\bar{N}:\bar{\Sigma}\rightarrow S^2\), that we identify with \(N\). Using the Weierstrass representation of \textit{R. Osserman} [Ann. Math. (2) 80, 340--364 (1964; Zbl 0134.38502)], one deduces that \begin{align*} \int_{\Sigma}-K\,d\mathrm{vol}_{\Sigma}=4\pi\,\mathrm{deg}(N). \end{align*} Furthermore, according to Theorem \(3.2\) of the aforementioned article of R. Osserman, if \(d_i\geq 1\) is the integer multiplicity -- or branching order -- of the end \(p_i\) (\(1\leq i\leq n\)), and \(\bar{\Sigma}\) has genus \(g\), we have \begin{align*} \mathrm{deg}(N)=g-1+\frac{1}{2}\sum_{i=1}(d_i+1). \end{align*} The formula is also named the ``Jorge-Meeks formula'', see [\textit{L. P. Jorge} and \textit{W. H. Meeks III}, Topology 22, 203--221 (1983; Zbl 0517.53008)]. The minimal surfaces of finite total curvature are important from the variational viewpoint for inverted minimal surfaces with finite total curvature are Willmore surfaces, i.e., critical points of the functional \begin{align*} W(\Sigma)=\int_{\Sigma}H^2d\mathrm{vol}_{\Sigma}, \end{align*} which not only constitute a geometrically meaningful extension of minimal surfaces but have applications to elasticity theory, general relatively, cellular biology, etc. The article under review fills an important gap in the literature about classical minimal surfaces. Relating the topology of a minimal surfaces with its Morse index (the number of negative eigenvalues of \(-\Delta_{\Sigma}+2\,K\) on \(\Sigma\); notice that the index must be defined by a limiting procedure because minimal surfaces are never compact due to the maximum principle) is an idea that in this context goes back to the work of \textit{D. Fischer-Colbrie} [Invent. Math. 82, 121--132 (1985; Zbl 0573.53038)]: a complete minimal surface in \(\mathbb{R}^3\) has finite Morse index if and only if it has finite total curvature. Quickly after, bounds of the form \begin{align*} \mathrm{Ind}(\Sigma)\leq \lambda_0\int_{\Sigma}-K\,d\mathrm{vol}_{\Sigma} \end{align*} (where \(\lambda_0>0\) is a universal constant) were found by \textit{J. Tysk} [Pac. J. Math. 128, No. 2, 361--366 (1987; Zbl 0594.58018)] using a heat kernel argument and later refined by \textit{N. Ejiri} and \textit{M. Micallef} [Adv. Calc. Var. 1, No. 3, 223--239 (2008; Zbl 1163.58006)] who managed to quantitatively relate the index of a minimal surface as a critical point of the area and its index as a critical point of the Dirichlet energy. Lower bounds of the same form proved to be more elusive. First, \textit{A. Ros} [J. Differ. Geom. 74, No. 1, 69--92 (2006; Zbl 1110.53009)] showed the lower bound \begin{align*} \mathrm{Ind}(\Sigma)\geq \frac{2\,g}{3} \end{align*} by relating the index of a minimal surface with the existence of special harmonic \(1\)-forms. By building up on this idea, the authors generalised Ros' inequality to obtain the following result [J. Differ. Geom. 104, No. 3, 399--418 (2016; Zbl 1357.53016)]: \begin{align*} \mathrm{Ind}(\Sigma)\geq \frac{2}{3}(g+n)-1. \end{align*} Finally, in the present article, the authors show that \begin{align*} \mathrm{ind}(\Sigma)\geq \frac{2}{3}\left(g-1+\sum_{i=1}^{n}(d_i+1)\right)-1. \end{align*} In particular, they deduce the following inequality \begin{align*} \mathrm{Ind}(\Sigma)\geq \frac{1}{6\pi}\int_{\Sigma}-K\,d\mathrm{vol}_{\Sigma}+\frac{1}{3}. \end{align*} Combining this new result with previously known upper bounds, the authors deduce that there are no complete minimal surfaces of index \(2\). They prove similar results for non-orientable minimal surfaces.