A combinatorial approach to the exponents of Moore spaces (Q667613)

Let $p\geq 2$ be a prime and let $X$ be a connected space with non-trivial reduced $\mathbb{Z}_{(p)}$-coefficients homology. Then it is known that any power map $p^t:\Omega\Sigma X\to \Omega\Sigma X$ is essential and this seems to discourage the study of the $p$ primary homotopy exponents of the single loop space. However, when the space $\Omega \Sigma X$ has various decompositions, one can consider the following question: \par When $T$ is the atomic retract of $\Omega\Sigma X$ containing the bottom cell, is it possible that there is a choice of projection map $\pi:\Omega\Sigma X\to T$ such that the composite \[ \Omega\Sigma X \stackrel{p^t}{\longrightarrow}\Omega \Sigma X \stackrel{\pi}{\longrightarrow}T \] is null homotopic for some integer $t\geq 1$? \par In fact, the authors study this problem and they obtain a positive answer for Moore spaces. \par Let $p\geq 3$ be an odd prime with $n\geq 2$ and recall that there is the following homotopy decomposition (obtained by Cohen-Moore-Neisendorfer) \[ \Omega P^{2n+1}(p^r)\simeq T^{2n+1}\{p^r\} \times \Omega P(n,p^r), \] where $P^m(p^r)=S^{m-1}\cup_{p^r}e^{m}$ is the $m$-dimensional mod $p^r$ Moore space, $P(n,p^r)$ is the wedge of mod $p^r$ Moore spaces, and $T^{2n+1}\{p^r\}$ is the atomic retract of $\Omega P^{2n+1}(p^r)$. \par In this paper the authors prove that there is a choice of the projection $\partial :\Omega P^{2n+1}(p^r)\to T^{2n+1}\{p^r\}$ such that the composite \[ \Omega P^{2n+1}(p^r) \stackrel{p^{r+1}}{\longrightarrow} \Omega P^{2n+1}(p^r) \stackrel{\partial}{\longrightarrow} T^{2n+1}\{p^r\} \] is null homotopic for $p\geq 5$ and $r>1$. As an application they also show that $T^{2n+1}\{p^r\}$ already has relative exponent $p^{r+1}$, where an $H$-space $X$ is said to have a relative exponent less than or equal to $p^t$ if there exists an $H$-space $Y$ and a retraction map $r:Y\to X$ such that the composite $r\circ p^t$ is null homotopic. Since it is known that $\Omega T^{2n+1}\{p^r\}$ has exponent $p^{r+1}$, this also generalizes the former classical result. The authors obtain the above results by using a combinatorial approach based on a careful analysis of the Cohen groups $K_n^{\mathbb{Z}/p^r}$.