Matrix factorization and the generic plane projection of curve a singularity

J. Elias · 2025

Read Matrix factorization and the generic plane projection of curve a singularity by J. Elias (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-18926.txt.

Introduction

Joan Elias studies matrix factorizations tied to curve singularities in a 2025 mathematics paper. A singularity is a point where a curve stops being smooth, and such points carry deep algebraic structure. Projecting a space curve onto a plane gives a plane curve, and the projection is called generic when it avoids special positions.

Elias examines the matrix factorizations defined by those generic projections. Turning the view around, he starts from a plane singularity and studies all space curves that project onto it, with their factorizations as a family. Matrix factorizations encode how singularities resolve and classify. The results link the algebra of a curve to that of its shadow, a bridge used in modern geometry and physics.

Catalog record

CREATORJ. Elias
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-18926

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Matrix factorization and the generic plane projection of curve a singularity J. Elias Abstract We study the matrix factorizations defined by the generic plane projections of a curve singularity of ( ℂ n , 0 ) (\mathbb{C}^{n},0) . On the other hand, given a plane curve singularity ( Y , 0 ) ⊂ ( ℂ 2 , 0 ) (Y,0)\subset(\mathbb{C}^{2},0) we study the family of matrix factorizations defined by the space curve singularities ( X , 0 ) ⊂ ( ℂ n , 0 ) (X,0)\subset(\mathbb{C}^{n},0) such that ( Y , 0 ) (Y,0) is the generic plane projection of ( X , 0 ) (X,0) . 2020 Mathematics Subject Classification 14B05, 13D02, 14B07. Key words and Phrases: . 1 Introduction The description of the structure and the computation of the length of minimal resolutions of finitely generated modules are central problems in several areas of mathematics. Hilbert’s syzygies theorem and the results of Auslander, Buchsbaum and Serre prove that regular rings are those for which all finitely generated modules have finite minimal resolutions. Koszul complexes and the theorems of Hilbert-Burch and Buschbaum-Eisenbud give explicit structures of the (finite) minimal resolutions of complete intersections, codimension two perfect ideals and codimension three Gorenstein ideals, respectively. In all these cases the information provided by the minimal resolution is finite and essentially unique. On the other hand, infinite minimal resolutions naturally arise in algebra since we consider modules over non-regular rings. See the contribution of Avramov in [ 16 ] for a good introduction to infinite resolutions. Eisenbud proved in [ 8 ] that finitely generated modules over a complete intersections of codimension one, i.e. R=S/(F) with R a local regular ring, eventually becomes periodic of period two. The matrices A,B of the morphisms defining this periodic structure is a matrix factorization of F , that means AB=BA=F\rm{Id} . This construction was generalized to complete intersection rings by Eisenbud and Peeva in [ 9 ] . Although the minimal resolution is infinite in such cases, the information defining it remains finite. See [ 9 , Chapter 1] for a introduction and survey to matrix factorizations, see also [ 10 ] , [ 11 ] and the references therein. The aim of this paper is to study matrix factorizations that naturally appear when we consider generic plane projections of an irreducible curve singularity of (\mathbb{C}^{n},0) . Given a curve singularity (X,0)\subset(\mathbb{C}^{n},0) the generic plane projections (Y,0) are well defined up to equisingularity type, see Definition 3.8 , inducing finite ring extensions \pi^{*}:\mathcal{O}_{(Y,0)}\longrightarrow\mathcal{O}_{(X,0)} . Since \mathcal{O}_{(Y,0)}=\mathbb{C}\{x,y\}/(F) is a complete intersection ring, by [ 8 ] there exists a matrix factorization of F defined by a pair of square matrices A , B , with entries in (x,y)\mathbb{C}\{x,y\} , such that AB=BA=F\mathrm{Id} , inducing an exact sequence 0\longrightarrow\mathbb{C}\{x,y\}^{b}\stackrel{{\scriptstyle A}}{{\longrightarrow}}\mathbb{C}\{x,y\}^{b}\longrightarrow\mathcal{O}_{(X,0)}\longrightarrow 0, and a minimal periodic free resolution of \mathcal{O}_{(Y,0)} -modules \cdots\longrightarrow\mathcal{O}_{(Y,0)}^{b}\stackrel{{\scriptstyle B}}{{\longrightarrow}}\mathcal{O}_{(Y,0)}^{b}\stackrel{{\scriptstyle A}}{{\longrightarrow}}\mathcal{O}_{(Y,0)}^{b}\stackrel{{\scriptstyle B}}{{\longrightarrow}}\mathcal{O}_{(Y,0)}^{b}\stackrel{{\scriptstyle A}}{{\longrightarrow}}\mathcal{O}_{(Y,0)}^{b}\longrightarrow\mathcal{O}_{(X,0)}\longrightarrow 0, see Theorem 4.2 . Moreover, we know that the family of generic plane projections is flat, see section 3 . This flat family induces a global matrix factorization specializing to the above matrix factorization, Theorem 4.2 . In the section 5 we consider the family of matrix factorizations defined by the space curve singularities (X,0)\subset(\mathbb{C}^{n},0) with a given a generic plane projection (Y,0)\subset(\mathbb{C}^{2},0) . In Proposition 5.7 we characterize the matrix factorizations of F\in\mathbb{C}\{x,y\} , equation defining (Y,0) , induced by a space curve singularity (X,0)\subset(\mathbb{C}^{n},0) . Equivalently, we characterize the dimension b square matrices A such that the cokernel of \mathbb{C}\{x,y\}^{b}\stackrel{{\scriptstyle A}}{{\longrightarrow}}\mathbb{C}\{x,y\}^{b} is a curve singularity with generic plane projection (Y,0) . In Theorem 5.12 we establish a bijection between the orbits of the action of the general linear group in the family of matrix factorizations induced by curve singularities with generic plane projection (Y,0) and the clases of isomorphism of \mathcal{O}_{(Y,0)} -algebras of the rings \mathcal{O}_{(X,0)} , where (X,0)\subset(\mathbb{C}^{n},0) is a curve singularity with (Y,0)\subset(\mathbb{C}^{2},0) as a generic plane projection. The explicit computations of this paper are performed by using the computer algebra system Singular [ 7 ] . 2 Preliminaries Let (X,0) be a an analytic reduced curve singularity of (\mathbb{C}^{n},0) , from now on curve singularity. The local ring {\mathcal{O}}_{(X,0)} is a one-dimensional reduced local ring with maximal ideal {\bf m}_{(X,0)} . We write by I_{(X,0)}\subset{\mathcal{O}}_{(\mathbb{C}^{n},0)}={\mathbb{C}}\{x_{1},\dots,x_{n}\} the radical ideal defining (X,0) , i.e. {\mathcal{O}}_{(X,0)}={\mathbb{C}}\{x_{1},\dots,x_{n}\}/I_{(X,0)} . The Hilbert function of (X,0) is the numerical function {\operatorname{H\!F}}_{(X,0)}:\mathbb{N}\longrightarrow\mathbb{N} such that {\operatorname{H\!F}}_{(X,0)}(n)={\mathrm{Length}}_{{\mathcal{O}}_{(X,0)}}({\bf m}_{(X,0)}^{n}/{\bf m}_{(X,0)}^{n+1}) for all n\in\mathbb{N} , The Hilbert-Samuel function of (X,0) is defined by {\operatorname{H\!F}}^{1}_{(X,0)}(n)=\sum_{j=0}^{n}{\operatorname{H\!F}}_{(X,0)}(j)={\mathrm{Length}}({\mathcal{O}}_{(X,0)}/{\bf m}_{(X,0)}^{n+1}), n\in\mathbb{N} . It is well known that there exist integers e_{0}(X,0),e_{1}(X,0)\in\mathbb{N} such that \operatorname{H\!P}^{1}_{(X,0)}(T)=e_{0}(X,0)(T+1)-e_{1}(X,0) is the Hilbert-Samuel polynomial of (X,0) , i.e. {\operatorname{H\!F}}^{1}_{(X,0)}(n)=\operatorname{H\!P}^{1}_{(X,0)}(n) for n\gg 0 . The integer e_{0}(X,0) is the multiplicity of (X,0) and the embedding dimension of (X,0) is {\operatorname{H\!F}}_{(X,0)}(1)=\operatorname{dim}_{\mathbb{C}}({\bf m}_{(X,0)}/{\bf m}_{(X,0)}^{2}) . Let \nu:\overline{X}={\bf Spec}(\overline{{\mathcal{O}}_{(X,0)}})\longrightarrow(X,0) be the normalization of (X,0) , where \overline{{\mathcal{O}}_{(X,0)}} is the integral closure of {\mathcal{O}}_{(X,0)} on its full ring of fractions {\mathrm{tot}}({\mathcal{O}}_{(X,0)}) . The singularity order of (X,0) is \delta(X,0)=\operatorname{dim}_{\mathbb{C}}\left(\overline{{\mathcal{O}}_{(X,0)}}/{\mathcal{O}}_{(X,0)}\right). We denote by \mathcal{C}(X,0) the conductor of the finite extension \nu^{*}:{\mathcal{O}}_{(X,0)}\hookrightarrow\overline{{\mathcal{O}}_{(X,0)}} and by c(X,0) the dimension of \overline{{\mathcal{O}}_{(X,0)}}/\mathcal{C}(X,0) . Let \omega_{(X,0)}={\mathrm{Ext}}^{n-1}_{{\mathcal{O}}_{(\mathbb{C}^{n},0)}}({\mathcal{O}}_{(X,0)},\Omega^{n}_{(\mathbb{C}^{n},0)}) be the dualizing module of (X,0) . We can consider the composition morphism of {\mathcal{O}}_{(X,0)} -modules \gamma_{X}:\Omega_{(X,0)}\longrightarrow\nu_{*}\Omega_{\overline{X}}\cong\nu_{*}\omega_{\overline{X}}\longrightarrow\omega_{(X,0)}. Let d:{\mathcal{O}}_{(X,0)}\longrightarrow\Omega_{(X,0)} the universal derivation. The Milnor number of (X,0) is, [ 3 ] , \mu(X,0)=\operatorname{dim}_{\mathbb{C}}(\omega_{(X,0)}/(\gamma_{X}\circ d){\mathcal{O}}_{(X,0)}) Notice that (X,0) is non-singular iff \mu(X,0)=0 iff \delta(X,0)=0 iff c(X,0)=0 . In the following result we collect some basic results on \mu(X,0) and other numerical invariants that we will use later on. Proposition 2.1 . Let (X,0) be a curve singularity of embedding dimension n . Then (i) \mu(X,0)=2\delta(X,0)-r(X,0)+1 , where r(X,0) is the number of branches of (X,0) . (ii) It holds e_{0}(X,0)-1\leq e_{1}(X,0)\leq\delta(X,0)\leq\mu(X,0) and e_{1}(X,0)\leq\binom{e_{0}(X,0)}{2}-\binom{n-1}{2} . (iii) If X is singular then \delta(X,0)+1\leq c(X,0)\leq 2\delta(X,0) , and c(X,0)=2\delta(X,0) if and only if {\mathcal{O}}_{(X,0)} is a Gorenstein ring. Proof. (i) [ 3 , Proposition 1.2.1] . (ii) [ 3 , Proposition 1.2.4 (i)] , [ 19 ] , [ 14 ] , [ 15 ] . (iii) [ 20 , Proposition 7, pag. 80] , and [ 1 ] . ∎ Given a local ring (A,{\bf m}) and a finitely generated A -module M we denote by \beta_{A}(M) the minimal number of generators of M , i.e. \beta_{A}(M)=\operatorname{dim}_{A/{\bf m}}(M/{\bf m}M) . 3 On the generic projection of a curve singularity Throughout this paper we only consider irreducible curve singularities. Let (X,0) be a curve singularity of (\mathbb{C}^{n},0) , i.e. (X,0) is an irreducible curve singularity of (\mathbb{C}^{n},0) . Then \overline{{\mathcal{O}}_{(X,0)}}\cong{\mathbb{C}}\{t\} for some uniformation parameter t\in{\mathrm{tot}}({\mathcal{O}}_{(X,0)}) and the morphism \nu^{*}:{\mathcal{O}}_{(X,0)}\hookrightarrow\overline{{\mathcal{O}}_{(X,0)}}\cong{\mathbb{C}}\{t\} can be described by a family of series \nu^{*}(\overline{x_{i}})=f_{i}(t)\in{\mathbb{C}}\{t\} , i=1,\dots,n . We say that (X,0):\left\{\begin{array}[]{l}x_{1}=f_{1}(t)\\ \quad\quad\vdots\\ x_{n}=f_{n}(t)\\ \end{array}\right. is a parametrization of (X,0) . Notice that the \mathbb{C} -algebra morphism defined by \begin{array}[]{cccc}\gamma:&{\mathbb{C}}\{x_{1},\dots,x_{n}\}&\longrightarrow&{\mathbb{C}}\{t\}\\ &x_{i}&\mapsto&f_{i}(t)\end{array} has as kernel the ideal I_{(X,0)} and induces the normalization map \nu^{*} . The conductor of this ring extension is principal \mathcal{C}(X,0)=(t^{c(X,0)}) . After a linear change of variables and choosing an appropriate uniformation parameter, i.e. an appropriate generator of the maximal ideal of {\mathbb{C}}\{t\} , we may assume that the parametrization of (X,0) is defined by: f_{1}(t)=t^{e_{0}(X,0)} and f_{i}(t)=\sum_{j\geq e_{0}(X,0)+1}b_{j,i}t^{j} , i=2,\dots,n , with b_{j,i}\in\mathbb{C} . We say that such a parametrization is

Read the complete text