Initial monomial invariants of holomorphic maps (Q907932): Difference between revisions

This interesting paper introduces and develops new biholomorphic invariants based on the notion of a generic initial ideal (\textit{gin}) from commutative algebra. The invariants are applied to holomorphic maps between domains in complex spaces of possibly different dimensions, CR mappings into hyperquadrics, as well as holomorphic mappings between complex manifolds. Moreover, the results apply to \textit{gin}s of real-analytic functions. The authors also prove an infinite-dimensional version of Galligo's theorem extending the finite-dimensional case obtained by the authors in [Math. Ann. 358, No. 3--4, 1059--1089 (2014; Zbl 1398.32037)]. Briefly, the \textit{gin} concept is developed along three different lines; for homogenizations of rational (proper) maps, for holomorphic decompositions of quotients of rational maps and for vector subspaces of holomorphic functions. An advantage of \textit{gin} invariants is that they are especially amenable to computer algebra systems. Many examples are given including a verification of Faran's classification of proper holomorphic maps from \(\mathbb B^2\) to \(\mathbb B^3\). More specifically, let \(U\subset\mathbb C^n\) and \(V\subset\mathbb C^m\) be domains. Define \(\mathcal{O}(U,V)\) to be the set of holomorphic maps \[ f:U\to V. \] An important problem in several complex variables is to understand \(\mathcal O(U,V)\) up to automorphism. Two such maps \(f\) and \(g\) in \(\mathcal O(U,V)\) are considered equivalent if there exist automorphisms \(\tau\) and \(\chi\) of \(U\) and \(V\) respectively such that \[ f\circ \tau=\chi\circ g. \] For \(f\) and \(g\) proper holomorphic maps that extend continuously to the boundary, there are corresponding questions in CR-geometry for the boundary maps \[ f,g:\partial U\to \partial V. \] Of particular interest are ball domains of signature \(b\) for integers \((a,b)\) where \(a\geq 1\) and \[ \mathbb B_b^{a+b}:=\bigg\{ z=(z_1,\dots,z_{a+b})\in\mathbb C^{a+b}:-\sum_{j=1}^b|z_j|^2+\sum_{j=1+b}^{a+b}|z_j|^2<1\bigg\} \] whose boundaries are the hyperquadrics \[ Q(a,b)=\bigg\{z\in\mathbb C^{a+b}:-\sum_{j=1}^b|z_j|^2+\sum_{j=1+b}^{a+b}|z_j|^2=1\bigg\}. \] \textit{Gin}s are defined as follows. Given homogeneous coordinates \(Z_0,Z_1,\dots, Z_n\), and a homogeneous polynomial \(P\), its initial monomial, in\((P)\), is defined as the maximal monomial of \(P\) in a multiplicative monomial order. For a given homogeneous ideal \(\mathcal I\), the initial monomial ideal, in \((\mathcal I)\), is defined as the smallest ideal containing all initial monomials of polynomials \(P\) in \(\mathcal I\). The flavor of the results include the following: Let \[ f_1 :\mathbb P^n\to\mathbb P^N\text{ and }f_2 :\mathbb P^n\to\mathbb P^N \] be rational proper maps that are equivalent in the sense that there exists a \(\tau\in \mathrm{Aut}(\mathbb P^n)\) and \(\chi\in\mathrm{Aut}(\mathbb P^N)\) such that \[ F\circ\tau=\chi\circ G. \] Then \[ gin(\mathcal I(F))=gin(\mathcal I(G)).\tag{*} \] As a corollary the authors obtain that (*) holds for the case of rational holomorphic maps between balls \[ f_1 :\mathbb B^n\to\mathbb B^N\text{ and }f_2 :\mathbb B^n\to\mathbb B^N \] that are spherically equivalent and have respective homogenizations \(F\) and \(G\). The proofs are based on techniques for generic initial ideals introduced in [\textit{M. L. Green}, Prog. Math. 166, 119--186 (1998; Zbl 0933.13002)] and developed further in the authors' recent paper [loc. cit.].