Fat lines in \(\mathbb P^3\): powers versus symbolic powers (Q2437950)

In this paper the authors study the symbolic and regular powers of ideals \(I\) for a family of special configurations of lines in \(\mathbb{P}^3\). Let \(R = k[x_0,\dots,x_N] = k[\mathbb{P}^N]\) be a polynomial ring over an algebraically closed field of characteristic zero and let \(I\) be an ideal of \(R\). The \(m-\)th symbolic power of \(I\), denoted \(I^{(m)}\), is computed as follows. If \(I^m=\bigcap_i Q_i\) is the homogeneous primary decomposition of \(I^m\), then \(I^{(m)}=\bigcap_j Q_j'\), where the intersection is over those primary components \(Q_i'\) which have \(\sqrt{Q_i'}\) contained in an associated prime ideal of \(I\). From the definition, we always have \(I^m \subseteq I^{(m)}\), but we do not always have \(I^{(m)} \subseteq I^m\). For non-trivial ideals we never have \(I^{(m)} \subseteq I^r\) when \(m<r\), but, by the results of Ein-Lazarsfeld-Smith [\textit{L. Ein} et al., Invent. Math. 144, No. 2, 241--252 (2001; Zbl 1076.13501)] and Hochster-Huneke [\textit{M. Hochster} and \textit{C. Huneke}, Invent. Math. 147, No. 2, 349--369 (2002; Zbl 1061.13005)], we always have \(I^{(m)} \subseteq I^r\) when \(m\geq rN\); in fact, \(I^{(me)}\subseteq I^m\) where \(e\) is the big height of \(I\) (i.e., the height of the associated prime ideal of \(I\) of biggest height). Consequently, for each fixed \(r\), it is of interest to find the smallest \(m\geq r\) with \(I^{(m)} \subseteq I^r\). Apart from some sporadic examples, most of the cases for which this smallest \(m\) are known either are ideals of complete intersections (i.e., ideals I generated by a regular sequence, in which case \(I^{(m)} = I^m\) for all \(m\geq 1\)) or are ideals defining zero-dimensional schemes. Thus the construction of families of ideals \(I\) of positive dimensional schemes which are not complete intersections but for which we can determine the least \(m\) for each \(r\) such that \(I^{(m)} \subseteq I^r\) is of particular interest. The authors focus on ideals with extremal behavior, in the sense that they satisfy \(I^{(m)} = I^m\) for all \(m \geq 1\). The key idea behind their constructions is to build finite configurations of lines in \(\mathbb{P}^3\) so that the associated ideals have a bigraded structure with respect to which the ideals define finite sets of points in \(\mathbb{P}^1\times \mathbb{P}^1\). Previous works, of the authors, on points in \(\mathbb{P}^1\times \mathbb{P}^1\) provides them with tools which they can exploit in their study of ideals of lines in \(\mathbb{P}^3\). More in detail, let \(R = k[\mathbb{P}^3] = k[x_0, x_1, y_0, y_1]\) and consider the two skew lines \(L_1\) and \(L_2\) in \(\mathbb{P}^3\) defined by \(I(L_1)=(x_0,x_1)\) and \(I(L_2)=(y_0,y_1)\). If \(B=[0:0:b_0:b_1]\) is a point on \(L_1\) and \(A=[a_0:a_1:0:0]\) is a point on \(L_2\), then the line \(L\) through \(A\) and \(B\) has ideal \(I(L) = (a_1x_0 - a_0x_1,b_1y_0 - b_0 y_1)\). We can regard \(R\) as being \(k[\mathbb{P}^1\times \mathbb{P}^1]\), and thus endowed with an \(\mathbb{N}^2\)-graded structure by setting \(\deg(x_i) = (1,0)\) and \(\deg(y_i) = (0, 1)\). With respect to this grading, \(I(L)\) is a bigraded ideal which defines the point \((P,Q)\in \mathbb{P}^1\times \mathbb{P}^1\) where \(P =[a_0 :a_1]\) and \(Q =[b_0 :b_1]\) and thus \(I(L)=I((P,Q))\). The configurations studied in this paper are finite unions of such lines, i.e., each line \(L\) in the configuration meets both the lines \(L_1\) and \(L_2\). This allows the authors to reinterpret the union of lines in \(\mathbb{P}^3\) as a finite set of points \(X\) in \(\mathbb{P}^1\times \mathbb{P}^1\). Conversely, the ideal of every finite set of points in \(\mathbb{P}^1\times \mathbb{P}^1\) defines a finite union of lines in \(\mathbb{P}^3\) where every line intersects \(L_1\) and \(L_2\). In addition, the authors require that the configurations be arithmetically Cohen-Macaulay (ACM), i.e., that their associated coordinate rings be Cohen-Macaulay. Expressed in the language of points in \(\mathbb{P}^1\times \mathbb{P}^1\), the main result is: { Theorem.} Let \(I = I(X)\) be the ideal of a finite reduced ACM subscheme \(X\) in \(\mathbb{P}^1\times \mathbb{P}^1\). Then \(I^{(m)} = I^m\) for all \(m\geq 1\) if and only if \(I^{(3)} =I^3\). A question raised by C. Huneke asks if, for a homogeneous ideal \(I\) of big height \(c\), it is true that \(I^{(m)} = I^m\) for all \(m\geq 1\) if \(I^{(c)} = I^c\). The authors, in this paper, show that the answer in general is no. In the Example 3.1, the authors show a set of points \(X\in \mathbb{P}^1\times \mathbb{P}^1\) such that its ideal \(I(X)\) is an unmixed ideal of height two (in particular, its big height is two) that has \(I(X)^{(2)} =I(X)^2\). The authors show that \(I(X)^{(3)}\not= I(X)^3\), then, by the main Theorem, it fails to have \(I(X)^{(m)} =I(X)^m\) for all \(m\geq 1\).