Gröbner bases via linkage (Q2437458): Difference between revisions
From MaRDI portal
Latest revision as of 10:43, 7 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Gröbner bases via linkage |
scientific article |
Statements
Gröbner bases via linkage (English)
0 references
3 March 2014
0 references
The authors of the paper under review indicate Gröbner bases for some types of determinantal ideals using the (known) fact that those ideals can be obtained from an ideal generated by indeterminates by a finite number of elementary biliaisons. Let us, firstly, state a simple lemma on which their method is based. Let \(R\) denote the polynomial ring \(K[x_0, \dots , x_n]\) over a field \(K\) which, for the purposes of the paper, can be assumed to be algebraically closed without loss of generality. Fix a monomial order \(\sigma\) on \(R\). Let \(I\) be a homogeneous ideal of \(R\) such that \(\text{ht}(P) = c\), \(\forall \, P \in \text{Ass}_R(R/I)\), for some integer \(c\) with \(2 \leq c \leq n-1\). One says that a homogeneous ideal \(J\) of \(R\) is obtained from \(I\) by an \textit{elementary biliaison} of height \(\ell \in {\mathbb Z}\) if there exists a homogeneous ideal \(N \subset I \cap J\) of \(R\), of pure height \(c - 1\), with \(R/N\) Cohen-Macaulay, and an isomorphism of graded \(R/N\)-modules \(\phi : (I/N)(-\ell) \overset\sim\rightarrow J/N\). If \(S\) is the set of homogeneous non-zero divisors of \(R/N\) then \(\phi\) is defined by multiplication with an element \(\alpha \in S^{-1}(R/N)\), homogeneous of degree \(\ell\). Assume, now, that the initial ideals (with respect to \(\sigma\)) \(\text{in}(I)\) and \(\text{in}(N)\) satisfy the same properties as \(I\) and \(N\). The above alluded lemma asserts that if there exists a monomial \(m \in R\), of degree \(\ell\), such that \(\text{in}(N) + m\cdot \text{in}(I) \subseteq \text{in}(J)\) then \(\text{in}(J) = \text{in}(N) + m\cdot \text{in}(I)\). This follows from a simple Hilbert function computation. As a simple illustration of their method, the authors consider, in the second section of the paper under review, the case of the ideal generated by the maximal minors of a matrix of indeterminates. Let \(X = (x_{ij})\) be an \(m \times n\) matrix of indeterminates, \(m \leq n\), \(R = K[X]\), \(G_m(X)\) the set of \(m \times m\) minors of \(X\), and \(I_m(X)\) the ideal of \(R\) generated by \(G_m(X)\). Let \(\sigma\) be a \textit{diagonal} monomial order on \(R\) (this means that the initial term of any minor of \(X\) is the product of the elements from its diagonal). Then \(G_m(X)\) is a Gröbner basis for \(I_m(X)\). The proof runs by induction on \(mn\) as follows. If \(A\) (resp., \(B\)) is a subset of \([1,m] := \{1, \ldots , m\}\) (resp., \([1,n]\)), let \(X_{A,B}\) denote the submatrix \((x_{ij})_{i\in A,j\in B}\) of \(X\). Then \[ \frac{I_{m-1}(X_{[1,m-1],[1,n-1]})}{I_m(X_{[1,m],[1,n-1]})}(-1) \simeq \frac{I_m(X)}{I_m(X_{[1,m],[1,n-1]})} \] (which can be proved by showing that the image of \[ \text{det}\, X_{[1,m],B\cup \{n\}}/\text{det}\, X_{[1,m-1],B}\, , \;B \subseteq [1,n-1]\, ,\;\text{card}\, B = m-1\, , \] into the fraction field of \(R/I_m(X_{[1,m],[1,n-1]})\) does not depend on \(B\)) and \[ \text{in}(I_m(X_{[1,m],[1,n-1]})) + x_{mn}\text{in}(I_{m-1}(X_{[1,m-1],[1,n-1]})) \subseteq \text{in}(I_m(X)) \] and one can complete the proof of the induction step by applying the above lemma. The authors also show, using a theorem of \textit{U. Nagel} and \textit{T. Römer} [J. Pure Appl. Algebra 212, No. 10, 2250--2258 (2008; Zbl 1160.13019)], that the simplicial complex associated to the squarefree monomial ideal \(\text{in}(I_m(X))\) is \textit{vertex decomposable}, hence shellable. In the third section of the paper under review, the authors prove an analogous result for Pfaffian ideals of ladders. We only reproduce the definition of such an ideal. Let \(X\) be a \(n \times n\) skew-symmetric matrix of indeterminates, \(R = K[X]\), and \(\sigma\) an anti-diagonal monomial order on \(R\). Consider integers \(1 \leq a_1 \leq \cdots \leq a_s \leq n\), \(1 \leq b_1 \leq \cdots \leq b_s \leq n\) and \(t_1 \geq 1, \ldots , t_s \geq 1\) such that \(a_k < b_k\), \(\forall \, k\). The authors apply their method to the ideal \(I_{2t_1}({\mathcal X}_1) + \cdots + I_{2t_s}({\mathcal X}_s)\), where \({\mathcal X}_k = X_{[a_k,b_k],[a_k,b_k]}\) and \(I_{2t_k}({\mathcal X}_k)\) denotes the ideal generated by the \(2t_k\)-Pfaffians of the skew-symmetric matrix \({\mathcal X}_k\). In this case, the biliaison arguments required by the proof can be found in the paper of \textit{E. De Negri} and \textit{E. Gorla} [J. Algebra 321, No. 9, 2637--2649 (2009; Zbl 1184.13036)]. In the fourth section of the paper under review, the author apply their above described method to \textit{symmetric mixed ladder determinantal ideals} using the results of \textit{E. Gorla} [Prog. Math. 280, 49--62 (2010; Zbl 1203.13015)], while in the fifth section they do the same thing for \textit{mixed ladder determinantal ideals} using the results of \textit{E. Gorla} [J. Pure Appl. Algebra 211, No. 2, 433--444 (2007; Zbl 1128.14035)] which are, actually, reproved, in a different manner, in the appendix to the paper.
0 references
Gröbner basis
0 references
liaison
0 references
determinantal ideal
0 references
Pfaffian ideal
0 references
vertex decomposable simplicial complex
0 references
0 references