Summation of unordered arrays (Q1646313)

An array \(a = \{a_t\}_{t \in T}\), consisting of complex numbers, is said to converge to zero if, for every \(\varepsilon > 0\), the set \(T(a, \varepsilon) = \{t \in T : |a_t| > \varepsilon\}\) is finite. Such an array is said to be greedily summable if there exists a limit of \(\sum_{t \in T(a, \varepsilon)} a_t\) as \(\varepsilon \to 0\). This limit is called the greedy sum and is denoted by \(\sum_{t \in T} a_t\). The main result of the paper says that, for arbitrary greedily summable arrays \(\{a_t\}_{t \in T}\) and \(\{b_s\}_{s \in S}\), if the product array \(\{a_t b_s\}_{(t, s) \in T \times S}\) is also greedily summable, then \[ \sum_{(t, s) \in T \times S} a_t b_s = \sum_{t \in T} a_t \sum_{s \in S} b_s. \] The demonstration is based on the properties of the associated generalized Dirichlet series \(A(z) = \sum_{t \in T} a_t |a_t|^z\) of the greedily summable array \(a\), namely, that \(A\) is holomorphic in the right half-plane and \(\lim_{z \to 0, z \in \mathbb R ^+} A(z) = \sum_{t \in T} a_t\). The result is generalized to arrays whose elements are nonsingular square matrices if the greedy summation is defined with respect to the absolute values of determinants.