On phantom maps into suspension spaces (Q1890299)

A phantom map \(f\colon X\to Y\) is a continuous map with the property that the composite \(W\to X\to Y\) is nullhomotopic whenever \(W\) is a finite complex (this is called a phantom map of the second kind by \textit{C. A. McGibbon} [Handbook of algebraic topology, North-Holland, Amsterdam,1209--1257 (1995; Zbl 0867.55013)]. In this paper, all spaces are assumed to have finite skeleta, so it is equivalent to ask that \(f| _{X_n}\simeq *\) for all \(n\). It follows from work of \textit{B. I. Gray} and \textit{C. A. McGibbon} [Topology 32, No. 2, 371--394 (1993; Zbl 0774.55012)] that there are no essential phantom maps \(\Omega S^n \to Y\) for any space \(Y\). J. Roitberg asked if the Eckmann-Hilton dual statement is also true: is it true that there are no essential phantom maps \(X \to\Sigma K(\mathbb{Z},n)\) for any space \(X\)? This paper shows that the answer is no. There are essential phantom maps \(X\to \Sigma K(\mathbb{Z},n)\) for some space \(X\). This answer is perhaps surprising, but much more surprising and pleasing is the main theorem from which it follows: for any rationally nontrivial space \(Y\) with finite skeleta, there are essential phantom maps \(X\to \Sigma Y\) for some space \(X\). The proof uses the computation by \textit{D. C. Ravenel} and \textit{W. S. Wilson} [Am. J. Math. 102, 691--748 (1980; Zbl 0466.55007)] of the Morava \(K\)-theory of Eilenberg-Mac Lane spaces. The main theorem is proved as a corollary of a similar result on phantom maps into \(p\)-localized suspensions.