Integral closures of powers of sums of ideals (Q6171763)

Let \(I \subseteq R = k[x_1,\ldots,x_n]\) be a homogeneous ideal over a standard graded polynomial ring \(R\) over a field \(k\). The integral closure of \(I\), denoted by \(\bar{I}\), consists of elements \(y \in R\) satisfying an equation of integral dependence of the form \[y^r + m_1 y^{r-1} + \cdots + m_{r-1} y + m_r = 0,\] where \(m_i \in I^i\). For a general ideal \(I\), computing its integral closure is very difficult. On the other hand, when \(I\) is a monomial ideal, \(\bar{I}\) is generated by monomials whose exponents are integer points of the Newton polyhedron of \(I\). When \(I = J + K\) is the sum of two ideals in different sets of variables, then \(I^t\) and \(I^{(t)}\) have the binomial expansion \begin{align*} I^{t} & = J^t + J^{t-1} K + \cdots + J K^{t-1} + K^t\\ I^{(t)} & = J^{(t)} + J^{(t-1)} K + \cdots + J K^{(t-1)} + K^{(t)}, \end{align*} where \(I^{(t)}\) is the \(t\)th symbolic power of \(I\). Based on these binomial expansions, [\textit{H. D. Nguyen} and \textit{T. Vu}, J. Pure Appl. Algebra 223, No. 7, 3081--3111 (2019; Zbl 1422.13015)] showed that the depth and regularity of \(I^t\) can be computed in terms of the depth and regularity of \(J^i\) and \(K^{j}\). Similarly, [\textit{H. T. Ha} et al., Math. Z. 294, No. 3--4, 1499--1520 (2020; Zbl 1437.13024)] showed the same for the symbolic powers of \(I\). The integral closure of powers of \(I\) does not have such a simple binomial expansion. A main observation of the paper is that when one replaces the regular powers by rational powers then one has a kind of binomial expansion for \(I\). For a real number \(r \ge 0\), the \(r\)th power of \(I\), denoted by \(I_r\), is the monomial ideal generated by monomials whose exponents are integer points of \(r NP(I)\), where \(NP(I)\) is the Newton polyhedron of \(I\). [\textit{P. Dongre} et al., J. Symb. Comput. 116, 39--57 (2023; Zbl 1499.13067)] showed that there is a discrete set of rational numbers \(t_0 = 0 < t_1 < \cdots < \) such that \(I_r = I_{t_j}\) for all \(r \in (t_{j-1},t_j]\). They called them the jumping numbers of \(I\). Furthermore, given an interval \([a,b]\), there are only finitely many jumping numbers of \(I\) within this interval. With this notation, one of the main result of the paper under review is Theorem 2.5 stating that for monomial ideals \(I\) and \(J\) in two different set of variables, one has \[(I+J)_r = \sum_{w} I_w J_{r-w},\] where the sum is taken over all jumping numbers \(w\) of \(I\) in \([0,r]\). As a consequence, they deduce bounds for the depth and regularity of integral closure of \((I+J)^t\) in terms of the depth and regularity of rational powers of \(I\) and \(J\).