Cross number invariants of finite Abelian groups. (Q2637181): Difference between revisions

The cross-number \(k(S)\) of a sequence \(S\) of non-zero elements of a finite Abelian group \(A=\bigoplus_{i=1}^rC_{p_i^{\alpha_i}}\) is defined by \[ k(A)=\sum_{a\in A}{v_a(S)\over\text{ord}(a)}, \] where \(v_a(S)\) denotes the number of occurrences of \(a\) in \(S\). The little cross-number of \(A\) is defined by \(k(A)=\max\{k(S):S\text{ is zero sum free}\}\), and the cross-number of \(A\) equals \(K(A)=\max\{k(S):S\text{ is irreducible}\}\). \(S\) is called irreducible if it does not contain a proper subsequence having zero sum. Moreover one puts \[ K_1(A)=\max\{k(S):S\text{ has unique factorization}\}, \] where a sequence \(S\) is said to have unique factorization, if it is of zero sum and there is essentially only one way to write it as a disjoint union of irreducible zero-sum subsequences. Finally put \[ k^*(A)=\sum_{i=1}^r\left(1-p_i^{-\alpha_i}\right),\quad K^*(A)=k^*(A)+1/\exp(A),\quad\text{and}\quad K_1^*(A)=\sum_{i=1}^r{p_i^{\alpha_i}-1\over p_i^{\alpha_i}-p_i^{\alpha_i-1}}. \] The equality \(K(A)=K^*(A)\) has been conjectured by \textit{U. Krause} and \textit{C. Zahlten} [Mitt. Math. Ges. Hamb. 12, No. 3, 681-696 (1991; Zbl 0756.20010)], and \textit{W. Gao} and \textit{L. Wang} [Integers 12, No. 4, 677-687, A14 (2012; Zbl 1343.11034)] conjectured the equality \(K_1(A)=K_1^*(A)\). These conjectures were established for certain families of groups (see \textit{A. Geroldinger} [J. Number Theory 48, No. 2, 219-223 (1994; Zbl 0814.20033)], \textit{A. Geroldinger} and \textit{R. Schneider} [Eur. J. Comb. 15, No. 4, 399-405 (1994; Zbl 0833.20061)], \textit{D. Kriz} [J. Number Theory 133, No. 9, 3033-3056 (2013)]). The author calls a prime \(p\) \textit{wide with respect to} \(n=\prod_{i=1}^sq_i^{a_i}\) if one has \[ {p\over p-1}\geq\prod_{i=1}^s{q_i^{a_i+1}-1\over q_i^{a_i+1}-q_i^{a_i}}, \] and shows that if \(p\) is wide with respect to \(\exp(A)\), then \[ k(C_{p^\alpha}\oplus A)=k(A)+k(C_{p^\alpha}), \] and a similar result is also obtained for \(K_1\), with a slightly different definition of wideness. This permits to extend the families of groups for which the above conjectures hold.