Commuting traces of multiadditive mappings (Q1364314): Difference between revisions

Let \(R\) be a prime ring. The problem of determining those \(n\)-additive maps \(M\colon R^n\to R\) whose trace is commuting (that is, \([M(x,\dots,x),x]=0\) for all \(x\in R\)) was solved for \(n=1\) [J. Algebra 156, No. 2, 385-394 (1993; Zbl 0773.16017)] and \(n=2\) [Trans. Am. Math. Soc. 335, No. 2, 525-546 (1993; Zbl 0791.16028)] by the reviewer. It has turned out that this problem is closely related to the problem of characterizing Lie isomorphisms (derivations) of (associative) rings (see also [J. Algebra 169, No. 1, 304-327 (1994; Zbl 0813.16020)] by \textit{K. I. Beidar, W. S. Martindale III} and \textit{A. V. Mikhalev}.) In the paper under review the problem is solved for an arbitrary positive integer \(n\). It is shown that if the characteristic of \(R\) is 0 or greater than \(n\), then there exist mappings \(\mu_i\) of \(R\) into the extended centroid \(C\) of \(R\) such that \(M(x,\dots,x)=\sum^n_{i=0}\mu_i(x)x^{n-i}\) for all \(x\in R\). Moreover, if \(R\) is not algebraic of bounded degree \(n\) over \(C\), then each \(\mu_i\) is of the form \(\mu_i(x)=M_i(x,\dots,x)\) for some \(i\)-additive mapping \(M_i\colon R^i\to C\).