Involution and anti involution on Hopf spaces (Q1389854)

Suppose an \(H\)-space \((X, \mu)\) admits an involution \(\tau \: X \to X\) that is an \(H\)-map. Then we say \(X\) is an \(H\)-space with involution. If the involution is an \(H\)-map from \((X, \mu)\) to \((X, \widetilde\mu)\), i.e., \(X\) but with the `opposite' multiplication given by switching coordinates before multiplying, then we say \(X\) is an \(H\)-space with anti-involution. This paper considers the case in which \(X\) is a \({\mathbb Z}_2\)-homology sphere. It is shown that under this hypothesis, the possibilities for displaying these kinds of structure are restricted quite severely. For the rest of this review, we focus only on the results for involution, since those for anti-involution have a similar flavour. If an \(H\)-space is a \({\mathbb Z}_2\)-homology \(n\)-sphere, then it follows from the well-known work of Adams, that \(n = 0, 1, 3\) or \(7\). If, in addition, \(X\) admits a \({\mathbb Z}_2\) action, then it follows from the so-called localization theorem of Hsiang/Quillen that the fixed-point set of the action must also be a \({\mathbb Z}_2\)-homology sphere. So now suppose \(X\) is a \({\mathbb Z}_2\)-homology \(n\)-sphere that admits a \({\mathbb Z}_2\) action, and that the fixed-point set is a \({\mathbb Z}_2\)-homology \(m\)-sphere. Then \(X\) is referred to as being of type \((n,m)\). Now an involution can be regarded as a \({\mathbb Z}_2\) action. Thus, a \({\mathbb Z}_2\)-homology sphere that is an \(H\)-space with involution is restricted to the types \((1, m)\), \((3, m)\) or \((7, m)\). The main result shows further that if \(X\) is a \({\mathbb Z}_2\)-homology sphere that is an \(H\)-space with involution, then the type is one of \((1, 0)\), \((1, 1)\), \((3, 1)\), \((3, 3)\), \((7, 3)\) or \((7, 7)\). In addition, for each of these possible types, a concrete example is given that displays the desired structure. The examples given are as in earlier work of \textit{K. Iriye} [J. Math. Kyoto Univ. 22, 719-727 (1983; Zbl 0518.55010)]. Each consists of the unit sphere of an orthogonal representation space of \({\mathbb Z}_2\). The restriction on the possible types comes from an argument that makes use of the Hopf construction, together with some Steenrod square calculations. The paper is written with a very narrow focus on the main result and its proof. This style makes it unnecessarily difficult to place the results in context. One interesting possible extension of the results that is not discussed is whether the given examples of \(H\)-spaces with involution are unique within their type? The author might have mentioned some of the various results known about closely-related questions [cf., for example, \textit{F. D. Williams}, Mich. Math. J. 19, 209-213 (1972; Zbl 0241.55014)].