Are complete intersections complete intersections? (Q1952139)

The main aim of this paper is to clarify the concept of complete intersection in local algebra. Two different definitions of complete intersection are in fact currently given. In this paper, the classical definition is referred to as the ``absolute'' definition, which says that the local ring \((R;\mathfrak{M})\) is isomorphic to the quotient of a regular local ring by an ideal generated by a regular sequence. The more usual definition, instead introduced by Scheja and popularized by Grothendieck, is mentioned here as the ``formal'' definition; it requires the same property for the completion \(\hat R\) of \(R\) with respect to the \(\mathfrak{M}\) topology. The authors show that the two definitions are equivalent if \((R,\mathfrak{M})\) is a one dimensional integral domain of arbitrary codimension; moreover, they produce an example of a dimension three integral domain which is ``formally'' but not ``absolutely'' complete intersection. They first prove a fundamental theorem, which, in short, says that if \(R\) contains a field of characteristic zero and and \( \pi: T\to \hat R\) is a surjective morphism defined on a complete regular local ring with embedding dimension equal to the embedding dimension of \(R\), then a presentation of \(R\) as an absolute complete intersection exists iff it can be obtained as a restriction of \(\pi\) to a convenient regular local subring \(S\). On this result are based their following arguments. In section one the authors prove that such an \(S\) exists provided \(R\) is a one dimensional integral domain. They summarize their procedure as follows. ``The method will be to inductively define subrings \(R_i\) of \(T/K^i\), where \(\hat R = T/K\), in such a way that the natural maps \(T/K^{i+1} \to T/K^i\) restrict to surjections \(R_{i+1}\to R_i\) and that \(T/K^i\) is naturally isomorphic to the completion of \(R_i\). In order to achieve this we must carefully lift elements from \(R_i\) to \(T/K^{i+1}\) to form \(R_{i+1}\), and this relies heavily on the fact that \(R\) is a one dimensional integral domain. We thus obtain an inverse system of rings \(\{R_i\}\) and the inverse limit \(S= \lim R_i\) we show has the required properties.'' Section 2 is devoted to the construction of a three dimensional local unique factorization domain, whose completion is \(T = [x,y,z,w]/ (x^2+y^2)\), which is not a quotient of a regular ring by an ideal generated by a regular sequence. Here is a description of the procedure, as given by the authors. ``We build an example with the property that if \(R\) can be lifted to a regular local ring \(S\) contained in \(R[[x,y,z,w]]\), than \(S\) must contain an element \(\Theta = f(1+z\omega + z^2h +xa +yb)\) where \(f= x^2 + y^2\) and \(\omega \in R\) is known, but we have no information about \(h,a,b\). We also equip \(R\) with a large collection of prime ideals whose extension to \(T\) contains \((x,y)T\) and the construction of each of these prime ideals guarantees the existence of an element in the lifting which is congruent to \(f\) modulo that particular prime ideal \(Q\). This new element and \(\Theta\) are unit multiples of each other and so their quotient will be in \(S\). So \(1+z\omega+z^2h \in R/Q\) and it follows that \(z(\omega+zh)\in R/Q\). We will construct \(R\) so that \(z\in R\) and so \(\omega +zh\) is in the quotient field of \(R/Q\). We can't control \(h\) but we do know that, for some fixed value of \(h\), \(\omega+zh\) must be in the quotient field of \(R/Q\) for every one of the primes we construct. So, for every possible value of \(h\), we construct a prime ideal \(Q_h\) such that \(\omega +zh\) is trascendental over \(R/Q_h\), and so certainly not in the quotient field. This contradiction rules out the possibility of a lifting.''