A generalized convolution for arithmetic functions (Q2546212)

A basic sequence \(B\) is a set of pairs \((a,b)\) of positive integers with the properties (i) \((a,b)\in B\) if and only if \((b,a)\in B\), (ii) \((a,bc)\in B\) if and only if \((a,b)\in B\) and \((a,c)\in B\), (iii) \((1,k)\in B\) \((k= 1,2, \ldots)\). One example of a basic sequence is the set \(M\) of all pairs of relatively prime positive integers. \(A\) is the set of all real-valued arithmetic functions and \(P\) is the set of those members \(f\) of \(A\) such that \(f(1) >0\). \(M(B)\) is the set of real-valued functions which are multiplicative with respect to \(B\), i.e., \(f(mn)=f(m)f(n)\) for all \((m,n)\) in \(B\). For two arithmetic functions \(f\) and \(g\), and a basic sequence \(B\), the convolution of \(f\) and \(g\) over \(B\) is defined by \[ (f \circ_B g)(n) = \sum_{\substack {dd'=n \\ (d,d')\in B}} f(d)g(d'). \] Proofs are given of the following theorems, which generalize some of the results of \textit{D. Rearick} [Duke Math. J. 35, 761--766 (1968; Zbl 0169.37201)]. Theorem 1. If \(B\) and \(E\) are basic sequences such that \(B\subset E\cap M\), then \(\{M(B), \circ_E\}\) is a group. Theorem 2. The following groups are all isomorphic: (1) \(\{A,+\}\), (2) \(\{P, \circ_B\}\) for every basic sequence \(B\), (3) \(\{M(B), \circ_E\}\) where \(B\) and \(E\) are any basic sequences for which \(B\subset E\cap M\).