On the constancy regions for mixed test ideals (Q405908)

Let \(R\) be an \(F\)-finite regular ring of characteristic \(p>0\), and let \({\mathfrak a}={\mathfrak a}_1,\ldots, {\mathfrak a}_n\) be ideals of \(R\). Let \({\mathbf c}=(c_1,\ldots,c_n)\) be a nonnegative real vector. The \textit{mixed generalized test ideal} \(\tau({\mathfrak a}^{\mathbf c})\) was introduced by \textit{N. Hara} and \textit{K.-i. Yoshida} [Trans. Amer. Math. Soc. 355, No. 8, 3143--3174 (2003; Zbl 1028.13003)], and in this setting it can be defined as follows, according to [\textit{M. Blickle}, \textit{M. Mustaţă}, and \textit{K. E. Smith}, Mich. Math. J. 57, 43--61 (2008; Zbl 1177.13013)]. For each \(e\in {\mathbb N}\), let \(I_e=\big({\mathfrak a}_1^{\lceil p^ec_1\rceil }\cdots{\mathfrak a}_n^{\lceil p^ec_n\rceil}\big)^{[1/p^e]}\), where \(J^{[1/p^e]}\) denotes the \([1/p^e]^{\text{th}}\) power of the ideal \(J\), introduced by \textit{M. Blickle et al.} in the aforementioned paper. The ideals \(I_e\) form an ascending chain, and therefore stabilize; its stable ideal is the mixed test ideal \(\tau({\mathfrak a}^{\mathbf c})\). Mixed test ideals share many properties with standard test ideals and multiplier ideals. For instance, \(\tau({\mathfrak a}^{\mathbf c})\) varies discretely with the parameter \({\mathbf c}\), remaining constant as each component \(c_i\) increases by a sufficiently small quantity. The article under review studies the regions of the parameter space \({\mathbb R}^n_{\geq 0}\) on which \(\tau({\mathfrak a}^{\mathbf c})\) is constant. The first main point of the paper is to show that, in contrast to its characteristic zero analogue -- the constancy regions of mixed multiplier ideals, which are finite unions of rational polytopes -- the constancy regions of mixed test ideals can be rather intricate and have fractal-like, self-similar behavior. This point is made through the careful analysis of an example. The second main point of the paper is to show that there is \textit{some} order behind those intricate structures -- that when \(R\) is essentially of finite type over a finite field, those regions can be described by a finite set of rules. This point is formalized through the use of the notion of \(p\)-fractals, introduced by \textit{P. Monsky} and \textit{P. Teixeira} [J. Algebra 280, 505--536 (2004; Zbl 1082.13016)], which are functions that possess certain self-similarity properties with respect to magnification by a factor of \(p\). Using the fact that over any bounded domain \(D\subseteq {\mathbb R}_{\geq 0}^n\) the set \(\{\tau({\mathfrak a}^{\mathbf c}):{\mathbf c}\in D\}\) is finite, the author shows that the characteristic function of a constancy region is a \(p\)-fractal. Since \(p\)-fractals can be described by finitely many initial values and rules, so can the constancy regions.