Obstructions to special Lagrangian desingularizations and the Lagrangian prescribed boundary problem (Q860163)

This is a well written description of the obstructions to the asymptotic Lagrangian smoothing of conical singularities of a special Lagrangian submanifold. Many illustrative examples are included. Following Harvey and Lawson one says that a closed \(p\)-form of a Riemannian manifold is a \textit{calibration} if \(\varphi| _{V} \leq vol| _{V}\) for every oriented \(p\)-plane \(V\) in TM. A \(p\)-dimensional oriented submanifold \(L\) is a \textit{calibrated~submanifold} if \(\varphi = vol\) on \(T_{x}L\) for every \(x \in L\). If \(\Omega\) is the complex \(n\)-form on \(\mathbb{C}^{n}\) given by \(\Omega = dz_{1} \wedge \dots \wedge dz_{n}\), then Re \(\Omega\) is a calibration on \(\mathbb{C}^{n}\) whose calibrated submanifolds are called \textit{special~Lagrangian} (or SL). Calibrated submanifolds are known to be volume minimizing in their homology class. If one replaces \(\mathbb{C}^{n}\) by a Calabi-Yau manifold \(M\) and \(\Omega\) by the natural parallel holomorphic \((n,0)\) form on \(M\), then one obtains a natural extension of special Lagrangian geometry. The SL manifolds may have singularities, and understanding these singularities is an important problem. D. Joyce has developed an approach to the study of compact SL manifolds in (almost) Calabi-Yau manifolds, and in particular he has studied examples of this type that have a finite number of isolated conical singularities modeled on SL cones \(C\) in \(\mathbb{C}^{n}\). The purpose of the present article is to study the asymptotic smoothing of such cones; that is, the existence of nonsingular SL submanifolds \(L\) that asymptotically approach a given SL cone at infinity. This is called the \textit{ACSL~ smoothing~ problem}. Three specific questions in this study, all stated in the introduction of the article, are given below at the end of this review. If \(\Sigma\) is a compact, oriented, imbedded submanifold of the unit sphere \(S^{2n-1}(1) \subset \mathbb{C}^{n}\), then \(\Sigma\) defines a cone \(C(\Sigma) = \{tx : x \in \Sigma, t \geq 0 \} \subset \mathbb{C}^{n}\). A cone \(C \subset \mathbb{C}^{n}\) is called \textit{regular} if \(C = C(\Sigma)\) for some submanifold \(\Sigma \subset S^{2n-1}(1)\) as above. The unit sphere \(S^{2n-1}(1)\) has a natural contact 1-form \(\gamma\), and an (\(n-1\))-dimensional submanifold \(\Sigma \subset S^{2n-1}(1)\) is said to be \textit{Legendrian} if \(\gamma \equiv 0\) on the tangent spaces of \(\Sigma\). It is known that a real, \(n\)-dimensional regular cone \(C(\Sigma)\) is Lagrangian iff \(\Sigma\) is a Legendrian submanifold of \(S^{2n-1}(1)\). The cones studied in this article are the regular Lagrangian cones \(C(\Sigma)\) that are special Langrangian. An \textit{asymptotically~ conical~ special~ Lagrangian~ \(n\)-fold} (or ACSL \(n\)-fold) is a nonsingular SL \(n\)-fold \(L \subset \mathbb{C}^{n}\) that tends asymptotically at infinity to a given regular SL-cone \(C(\Sigma)\). Such an \(n\)-fold \(L\), if it exists, is called an \textit{ACSL~ smoothing} of \(C(\Sigma)\). There are topological obstructions to the ACSL smoothing of a regular SL cone \(C(\Sigma)\). For example, if \(\Sigma\) does not bound a smooth submanifold, then an ACSL smoothing \(n\)-fold \(L\) does not exist. The authors give an elementary example of an SL-cone \(C(\Sigma)\) of this type. There are other obstructions to ACSL smoothings that arise for SL-cones \(C(\Sigma)\) where \(\Sigma\) is a Legendrian submanifold of a unit sphere that is defined by a Lie group action possessing a moment map. The authors list three such examples \(C(\Sigma)\), all of an elementary nature, for which the ACSL smoothing of \(C(\Sigma)\) has no solution. In one of the three examples the manifold \(\Sigma\) is diffeomorphic to SU\((n)\), which is null cobordant, and hence for this infinite family of examples indexed by n the topological obstructions to ACSL smoothing mentioned above are all zero. For null cobordant submanifolds \(\Sigma \subset S^{2n-1}(1)\), that is, those that bound a compact submanifold L of \( S^{2n-1}(1)\), it is natural to ask if there exist Lagrangian immersions \(f : L \rightarrow \mathbb{R}^{2n}\) for a suitably chosen ``filling'' \(L\) of \(\Sigma\). An immersion \(f : L \rightarrow \mathbb{R}^{2n}\) is \textit{Lagrangian} if \(f^{*} \omega = 0\), where \(\omega = \sum_{k=1}^{n} dx_{k} \wedge dy_{k}\) and \(\{z_{k} = x_{k} + i y_{k}, 1 \leq k \leq n \}\) are the natural coordinate functions for \(\mathbb{C}^{n} \approx \mathbb{R}^{2n}\). The authors answer this question using a version of the h-principle of Gromov and Lees. This question is a special case of a more general problem the authors consider called the \textit{prescribed~boundary~problem} for Lagrangian immersions. We omit a precise definition of this problem. In general there are also obstructions for an affirmative answer to the prescribed boundary problem. Sections 3 and 4 are devoted to a careful discussion of this problem, and section 5 discusses the problem in \(\mathbb{R}^{4}, \mathbb{R}^{8}\) and \(\mathbb{R}^{10}\). In section 6 the authors use the results of section 5 to obtain answers in low dimensions to the following questions posed in the introduction to the article: {\parindent7mm \begin{itemize}\item[(1)] Let \(C\) be a SL-cone in \(\mathbb{C}^{n}\). Does there exist a smooth, complete, oriented, immersed Lagrangian submanifold \(Y\) in \(\mathbb{C}^{n}\) that coincides with \(C\) outside the unit sphere \(S^{2n-1}(1)\)? What can we say about the topology of \(Y\)? Can we find such a \(Y\) with zero Maslov class? \item[(2)] Let \(X\) be an SL manifold with isolated conical singularities in an (almost) Calabi-Yau manifold. Does there exist a Maslov-zero Lagrangian submanifold \(Y\) which coincides with \(X\) away from a neighborhood of its singularities? \item[(3)] Let \(C(\Sigma\)) be a SL-cone in \(\mathbb{C}^{n}\), and let \(f : \Sigma\times [1,\infty] \rightarrow \mathbb{C}^{n}\) be a Maslov-zero Lagrangian immersion which is asymptotic to \(C\) at infinity. What Maslov-zero Lagrangian ``fillings'' of the conical end \(f\) exist. \end{itemize}}