The structure of the inverse system of Gorenstein \(k\)-algebras (Q2357473)

This is a seminal paper! \textit{F. S. Macaulay} [The algebraic theory of modular systems. Cambridge: University Press (1916; JFM 46.0167.01)] introduced his ``inverse systems'' in 1916 -- one hundred years ago. Here is one version of the Macaulay correspondence in modern language. Let \(V\) be a finite dimensional vector space over the field \({\pmb k}\), \(R\) be the polynomial ring \[ R=\bigoplus_{0\leq i}{\text{Sym}}_i^{{\pmb k}}V \] and \(D\) be the graded dual \(D=\bigoplus_{0\leq i}D_i^{{\pmb k}}(V^*)\) of \(R\). (So, \(D_i^{{\pmb k}}(V^*)\) is another name for \({\text{Hom}}_{{\pmb k}}({\text{Sym}}_i(V),{\pmb k})\).) View \(D\) as an \(R\)-module as follows. If \({r_i\in R_i}\) and \(\phi_j\in D_j\), then \(r_i(\phi_j)\) is the element of \(D_{j-i}V^*\) which sends \(r_{j-i}\in R_{j-i}\) to \(\phi_j(r_ir_{j-i})\in {\pmb k}\). Macaulay proved that there is a one-to-one correspondence between the \(R_+\)-primary ideals \(I\) of \(R\) and the finitely generated \(R\)-submodules \(M\) of \(D\) given by \[ I\rightarrow 0:_DI \quad \text{and}\quad {0:_RM\leftarrow M}. \] The correspondence carries homogeneous ideals of \(R\) to homogeneous submodules of \(D\) (and vice-versa). The correspondence also carries ideals of \(R\) which define Gorenstein rings to cyclic submodules of \(R\) (and vice-versa). The Macaulay correspondence is a duality; indeed, it is an example of Matlis Duality because \(D\) is the injective envelope of \({\pmb k}\) as an \(R\)-module and \[ {\text{Hom}}_R(R/I,D)=0:_DI \quad\text{and}\quad {\text{Hom}}_R(D/M,D)= 0:_RM. \] A minimal generating set for \(0:_DI\) is called a \textit{Macaulay inverse system} for \(I\). It took a while before the commutative algebra community came to appreciate Macaulay inverse systems. MathSciNet only records three uses of the concept in the period 1950--1975; but soon thereafter, Tony Iarrobino was using them (see, for example, [\textit{A. Iarrobino}, Trans. Am. Math. Soc. 285, 337--378 (1984; Zbl 0548.13009)]), and now-a-days it is impossible to attend a commutative algebra conference without hearing about Macaulay inverse systems. People now give structure theorems for Artinian \({\pmb k}\)-algebras based on the form of the corresponding Macaulay inverse system. Furthermore, a convenient way to describe a Gorenstein ideal is to give its Macaulay inverse system, a convenient way to study a \textbf{family} of Artinian rings is to describe properties of their Macaulay inverse systems. Even the computer program Macaulay2 comes complete with ``toDual'' and ``fromDual'' and a many people use these commands and these ideas. Now, let \(I\) be an ideal of \(R\) which defines a Gorenstein quotient of \(R\) of \textbf{positive} Krull dimension \(d\). It follows from Macaulay's theorem that \(0:_DI\) is an infinitely generated submodule of \(D\). Elias and Rossi have the had the audacity to ask, ``Can the generators of \(D\) be described in some coherent manner?'' and ``Are there situations where in fact \textbf{finite} data taken from a generating set for \(0:_DI\) describes \(I\)?''. The answers are YES! and YES!. This is a game changer. This paper will have many applications and spark many extensions and improvements. Will it be possible to use the Elias-Rossi extension to Macaulay inverse systems to describe the structure of Gorenstein algebras of positive dimension? Can the Elias-Rossi ideas be used on non-Gorenstein algebras? The Elias-Rossi approach is straightforward. Why didn't any one think of it before? Pick \(z_1,\dots, z_d\) in \(R_1\) which represent a system of parameters for \(R/I\). Let \(H_{(1,\dots,1)}\) be an element of \(D\) which is a Macaulay inverse system for \(R/(I, z_1,\dots, z_d)\). Of course, there is a unique choice for \(H_{(1,\dots,1)}\) up to scalar multiplication. Now slowly and carefully integrate to find \[ \{H_{(\ell_1,\dots,\ell_d)}|1\leq \ell_i \text{ for \(1\leq \ell\leq d\)}\} \] with {\parindent=0.7cm\begin{itemize}\item[--] \(H_{(\ell_1,\dots,\ell_d)}\) a Macaulay inverse system for \(R/(I, z_1^{\ell_1},\dots, z_d^{\ell_d})\), and \item[--] \(z_i(H_{(\ell_1,\dots,\ell_d)})=H_{(\ell_1,\dots,\ell_{i}-1,\dots \ell_d)}\), whenever \(2\leq \ell_i\). \end{itemize}} The work is involved in axiomatizing the submodules of \(D\) that are produced by the above procedure, proving the necessary one-to-one correspondence between the ideals of \(R\) and the submodules of \(D\), and using Castelnuovo-Mumford regularity to extract all of the ideal from a finitely generated submodule of \(D\). This is an awesome project.