On anticommutative semirings (Q1118671)

Let \((R,+,\cdot)\) be a semiring, i.e. \((R,+)\) and \((R,\cdot)\) are semigroups connected by ring-like distributivity, and assume that \((R,.)\) is anti-commutative (or nowhere commutative, i.e. \(xy = yx\) for all \(x = y\) of \(R\)). It is well known that the letter holds iff \((R,\cdot)\) is isomorphic to the direct product \((R_1,\cdot) \times (R_2,\cdot)\) of the left [right] zero subsemigroups \(R_1 = Ra\) [\(R_2 = aR\)] of \((R,\cdot)\) for any fixed element \(a\) in \(R\), where \(x \to (xa,ax)\) for all \(x \in R\) defines such an isomorphism \(f\). Applying this to the semiring \((R,+,\cdot)\), clearly \(R_1\) and \(R_2\) are subsemirings and \(f\) is an isomorphism of semirings. Moreover, \((R_1,+)\) and \((R_2,+)\) are idempotent (e.g. by \(r + r = rs + rs = r(s + s) = r\) for all \(r,s \in R_1\), which yields that \((R,+)\) is idempotent. These two remarks contain and prove all results of the paper except the following one: A commutative semigroup \((R,+)\) provides a semiring \((R,+,\cdot)\) if one defines \(x\cdot y = x\) for all \(x,y \in R\). But the latter is only true iff \((R,+)\) is idempotent, and independent on the assumption that \((R,+)\) is commutative, which is superfluous throughout the paper.