Bad extensions of analytic algebra (Q1117984)

It is a well-known fact that a noetherian k-algebra B is of finite type if B/\(\sqrt{0}\) is. We show that for analytic k-algebras the analogous property does not hold. Indeed, for \(A_ 1=k<x>\) there exists an infinitesimal extension \(0\to A_ 1\to B_ 1\to A_ 1\to 0\) which does not split; therefore \(B_ 1\) cannot be analytic. It follows that to each analytic algebra A of positive dimension there exists a k-algebra B such that \(B_{red}\cong A\) and B is not analytic. In order to obtain this result we determine the group \(Exalcom_ k(A_ 1,A_ 1).\) Globally the situation is similar. Consider a ringed space X, a sheaf \(I\subset {\mathcal O}_ X\) of ideals with \(I^ 2=0\), and \(I/I^ 2\) of finite type as \({\mathcal O}_ X/I\)-sheaf. If \(X_ 0=(X,{\mathcal O}_ X/I)\) is an algebraic k-scheme then X is too as it is well known. If \(X_ 0\) is an analytic space then under further conditions it follows [cf. \textit{J. Bingener}, Invent. Math. 64, 25-67 (1981; Zbl 0509.32004)] that X is an analytic space. We show, however, that in general X is not analytic, even if \(X_ 0\cong {\mathbb{C}}^ 1\).