A pure power product version of the Hilbert Nullstellensatz (Q1590921)

Let \(k\) be a field and \(R=k[T_0,\dots,T_n]\). Then the radical version of Hilbert's Nullstellensatz states that, for any homogeneous ideal \(I=(f_1,\dots,f_m)\), there exists a minimal natural number \(e\) such that Rad\((I)^e\subset I\). The problem is to give an effective bound on \(e\). In the paper under review, the author proves the following prime version of the effective Hilbert's Nullstellensatz (theorem 1): Assume that the polynomials \(f_i\) are indexed so that their degrees satisfy \(d_2\geq \cdots \geq d_m \geq d_1\), and let \(\mathfrak m\) be the maximal (homogeneous) ideal \((T_0,\dots,T_n)\), which is also called the \textit{irrelevant} prime ideal. Furthermore, suppose that \(d_{\mu-h+1}\geq 3\) where \(h>1\) denotes the height of \(I\) and \(\mu=\min(m,n)\). Then, there are relevant prime ideals \(\mathfrak{p}_1,\dots,\mathfrak{p}_r\), containing \(I\), positive integers \(e_1,\dots,e_r\) and a natural number \(e_0\) such that \(\mathfrak{m}^{e_0}\mathfrak{p}^{e_1}_1\cdots \mathfrak{p}^{e_r}_r \subset I,\) where \(e_0+\sum e_i\deg(\mathfrak{p}_i)\leq d_1\cdots d_\mu,\) when card\((k)\geq d_1\cdots d_\mu\). An earlier version of author's result and its proof, was used by \textit{S. Ji, J. Kollár} and \textit{B. Shiffman} [Trans. Am. Math. Soc. 329, 813-818 (1992; Zbl 0762.14001)]. As a corollary the author recovers the bound given by \textit{J. Kollár} [J. Am. Math. Soc. 1, 963-975 (1988; Zbl 0682.14001)], namely \(e\leq d_1\cdots d_\mu\). In the present paper, the main result appears as a consequence of a ``pure power version'' of the Nullstellensatz (theorem 2), where ``prime ideal'' is replaced by pure ideal (i.e. ideal such that all its associated prime ideals have the same height). The proof of this ``pure power version'' is based on the cohomological technique of Kollár (op. cit.), and relies on an inductive construction of ideals annihilating certain cohomology groups (local cohomology) and by reducing to the case where the generators \(f_1,\dots,f_m\) form a sequence that is as regular as possible. The condition \(d_{\mu-h+1}\geq 3\) appears as a technical assumption. Notice that the first two sections of the paper provide a nice short survey on the subject.