Inequivalent embeddings of the Koras-Russell cubic 3-fold (Q621296)

The Koras-Russell cubic 3-fold is the hypersurface \(X\) of the complex affine space \({\mathbb A}^4\) defined by the equation \(x+x^2y+z^2+t^3=0\). It is well known that \(X\) is an affine contractible smooth 3-fold that is not algebraically isomorphic to an affine 3-space. The main result of this paper is to show that there exists another hypersurface \(Y\) of \({\mathbb A}^4\) that is isomorphic to \(X\) but such that there exists no automorphism of the ambient 4-space that restricts to an isomorphism between \(X\) and \(Y\). In other words, the two hypersurfaces are inequivalent. In~order to prove this, the authors give a description of the automorphism group of \(X\). They show that all algebraic automorphisms of \(X\) extend to automorphisms of \({\mathbb A}^4\). This result implies the surprising corollary: any algebraic automorphism of \(X\) fixes the point \((0,0,0,0)\in X\). More precisely, the hypersurface \(Y\) is defined by the equation \(x^2y+(1+x)(z^2+x+t^3)=0\), and the isomorphism between \(X\) and \(Y\) is induced by the endomorphism \((x,y,z,t) \mapsto(z,(1+x)y,z,t)\) of \({\mathbb A}^4\). It turns out that the hypersurfaces \(X\) and \(Y\) in \({\mathbb A}^4\) are holomorphically equivalent and stably equivalent. In particular, there is no topological obstruction to equivalence of \(X\) and \(Y\). The methods for studying the question of equivalent embeddings used in the paper are similar to the methods developed in [\textit{A.~Dubouloz} and \textit{P.-M.~Poloni}, J. Algebra 321, No. 7, 1797--1812 (2009; Zbl 1168.14042)] and [\textit{L.~Moser-Jauslin} and \textit{P.-M.~Poloni}, Ann. Inst. Fourier 56, No. 5, 1567--1581 (2006; Zbl 1120.14056)]. Recall that a Danielewski hypersurface is a subvariety in \({\mathbb A}^3\) defined by an equation of the form \(x^ny=q(x,z)\), where \(n\) is a positive integer and \(q(x,z)\) is a polynomial such that \(q(0,z)\) has degree at least 2. The authors note that the Koras-Russell cubic 3-fold may be considered as a 1-parameter family of Danielewski hypersurfaces and then adapt the methods of the papers cited above to a higher-dimensional variety.