Approximation of Smooth Maps by Real Algebraic Morphisms

From MaRDI portal





A main problem in real algebraic geometry is to determine those \(C^\infty\) maps between two nonsingular algebraic varieties \(X\) and \(Y\) occurring in the closure, with respect to the \(C^\infty\) compact-open topology, of the subset \(R(X,Y)\) of regular mappings from \(X\) to \(Y\). To state properly the theorem proved in the reviewed paper, let \(G_{p,q}(\mathbb{F})\) be the Grassmannian of all \(q\)-dimensional \(\mathbb{F}\)-vector subspaces of \(\mathbb{F}^p\) where \(\mathbb{F}\) stands for \(\mathbb{R}\), \(\mathbb{C}\) or the quaternions \(\mathbb{H}\), and let \(d\) be the dimension of \(\mathbb{F}\) as \(\mathbb{R}\)-vector space. Let \(X\) be a nonsingular affine real algebraic variety such that \(\dim(X)\leq d [p+1-\min(q,p-q)]\) (of course, \(p>q\geq 1)\). Then, for a \(C^\infty\) map \(f:X\to G_{p,q}(\mathbb{F})\), the following conditions are equivalent: (i) \(f\) belongs to the closure of \(R(X,G_{p,p} (\mathbb{F}))\) in \(C^\infty(X,G_{p,q}(\mathbb{F}))\); (ii) \(f\) is homotopic to a regular map from \(X\) into \(G_{p,q}(\mathbb{F})\); (iii) the pull-back \(f^* (\gamma_{p,q}(\mathbb{F}))\) of the universal \(\mathbb{F}\)-vector bundle \(\gamma_{p,q} (\mathbb{F})\) on \(G_{p,q}(\mathbb{F})\) admits an algebraic structure, i.e. it is topologically isomorphic to a strongly algebraic \(\mathbb{F}\)-vector bundle. The same theorem, for compact \(X\) and without restrictions on \(\dim X\) is theorem 13.3.1 of the book by \textit{J. Bochnak}, \textit{M. Coste} and \textit{M.-F. Roy}, ``Géométrie algébrique réelle (1987; Zbl 0633.14016); see also the English translation (1998; Zbl 0912.14023). However, the case \(F=\mathbb{R}\) and \(X\) is compact was firstly proved by \textit{N. V. Ivanov} [Russ. Math. Surv. 37, 1-59 (1982); translation from Usp. Mat. Nauk 37, No. 1(223), 3-52 (1982; Zbl 0571.14013)]. The equivalence between (i) and (ii) was proved by the first author [\textit{W. Kucharz}, Boll. Unione Mat. Ital. (7), A8, No. 3, 345-352 (1994; Zbl 0829.14006)] while (ii) obviously implies (iii). So, the goal of the paper is to prove: (iii) implies (i) in the non compact case. Here the constrain on \(\dim X\), which is obviously satisfied if \(\dim X\leq 2d\), is used in the proof, although it is unknown if it is necessary. In particular, \(R(X,G_{p,q} (\mathbb{R}))\) is dense in \(C^\infty(X,G_{p,q} (\mathbb{R}))\) if \(p>q\geq 1\) and \(X\) is a real nonsingular curve. -- As it is classical, the theorem yields approximation results for \(C^\infty\) maps into the spheres \(S^1\), \(S^2\) and \(S^4\) because \(S^k\), for \(k=1,2\) and 4 is biregularly isomorphic to the projective line. The paper is very clearly written, in a very direct style. To substitute the compactness hypothesis the authors use Hironaka's theorem on resolution of points of indeterminancy.











This page was built for publication: Approximation of Smooth Maps by Real Algebraic Morphisms

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4384054)