The prime spectrum of algebras of quadratic growth. (Q2470409)

Given a field \(k\) and a finitely generated \(k\)-algebra \(A\), a \(k\)-subspace \(V\) of \(A\) is called a `frame' of \(A\) if \(V\) is finite-dimensional, \(1\in V\), and \(V\) generates \(A\) as a \(k\)-algebra. Then \(A\) is of `quadratic growth' if there exists a frame \(V\) and constants \(C_1,C_2>0\) such that \[ C_1n^2\leq\dim_k(V^n)\leq C_2n^2,\quad n\geq 1. \] Recall that the Gelfand-Kirillov dimension (or GK-dimension) of a finitely generated \(k\)-algebra \(A\) is defined as \[ \text{GKdim}(A)=\limsup_{n\to\infty}\log(\dim(V^n))/\log n, \] [for an exposition see \textit{G. R. Krause, T. H. Lenagan}, Growth of algebras and Gelfand-Kirillov dimension. Graduate Stud. Math. 22. Providence, RI: AMS (2000; Zbl 0957.16001)]. While algebras of quadratic growth have GK dimension 2, it is not the case that an algebra of GK dimension 2 necessarily has quadratic growth. The authors consider prime monomial algebras of quadratic growth. A \(k\)-algebra \(A\) is a `monomial algebra' if \(A\cong k\{x_1,\dots,x_d\}/I\), for some ideal \(I\) generated by monomials in \(x_1,\dots,x_d\). The first result is the following Theorem 1. Let \(k\) be a field and let \(A\) be a prime monomial \(k\)-algebra of quadratic growth. Then the set of primes \(P\) such that \(\text{GKdim}(A/P)=1\) is finite; moreover, all such primes are monomial ideals. In particular, \(A\) has bounded matrix images. If a finitely generated \(k\)-algebra \(A\) has quadratic growth, the authors define the `growth constant' of \(A\) to be \[ \text{GC}(A):=\inf_V\limsup_{n\to\infty}\dim(V^n)/n^2, \] where the infimum is taken over all frames \(V\) of \(A\). The authors suggest the conjecture that the number of primes in the theorem above is determined as a function of \(\text{GC}(A)\). Next, the authors study graded algebras of quadratic growth. In this case, they give an analogue of Bergman's gap theorem, which states that there are no algebras of GK dimension strictly between 1 and 2 for ideals. The wording of that results is rather lengthy, as an application the authors obtain the following result about prime ideals in graded algebras of quadratic growth. Theorem 2. Let \(k\) be a field, and let \(A=\bigoplus_{i=0}^\infty A_i\) be a prime affine graded non-PI \(k\)-algebra, generated in degree one. If \(A\) has quadratic growth then the intersection of all non-zero prime ideals \(P\) such that \(A/P\) has GK dimension 2 is non-zero, where we take an empty intersection to be all of \(A\). Both theorems yield the following Corollary. Let \(A\) be a finitely generated prime monomial algebra of quadratic growth. Then \(A\) has bounded matrix images and either \(A\) is primitive or has non-zero locally nilpotent Jacobson radical. Finally, the authors consider a more general setting of algebras of GK dimension 2. In this case, they show that prime algebras can have unbounded matrix images, thus the quadratic growth assumption is necessary to conclude that there are only finitely many prime ideals such that \(A/P\) has GK-dimension 1. Theorem 3. There exists a finitely generated prime monomial algebra \(A\) of GK dimension 2 with unbounded matrix images.