The rational homotopy of the \(K(2)\)-local sphere and the chromatic splitting conjecture for the prime 3 and level 2 (Q2339190)

This paper is part of the modern chromatic approach towards the homotopy groups of spheres. Denote by \(L_n\) Bousfield localization with respect to the \(n\)-th Johnson-Wilson theory \(E(n)\) and by \(L_{K(n)}\) localization with respect to Morava K-theory for a fixed prime \(p\). Key to the chromatic picture of homotopy theory is the Hopkins--Ravenel chromatic convergence theorem, stating that one can recover the \(p\)-localization of a finite complex \(X\) as the homotopy inverse limit over all \(L_nX\). The \(L_nX\) can, in principle, be inductively computed via the \(L_{K(n)}X\) since \(L_nX\) is the homotopy pullback \(L_{K(n)}X \times^h_{L_{n-1}L_{K(n)}X} L_{n-1}X\). Hopkins' chromatic splitting conjecture [\textit{M. Hovey}, Contemp. Math. 181, 225--250 (1995; Zbl 0830.55004)] states rather precisely how this gluing process works in the case of the \(p\)-completed sphere: It predicts that the map \(L_{n-1}S^0_p \to L_{n-1}L_{K(n)}S^0_p\) is the inclusion of a direct summand and expresses the cokernel explicitly in terms of localizations of \(S^0_p\). It has been proven for \(n=1\) (all primes) and \(n=2\) and \(p>3\). The present paper verifies the conjecture for \(n=2\) and \(p=3\). Beaudry has shown since then that the strong form of the conjecture is wrong for \(n=2\) and \(p=2\), [\textit{A. Beaudry}, The Chromatic Splitting Conjecture at \(n=p=2\), {\url arXiv:1502.02190}]. From now on, let \(p=3\) be fixed and we will review the content of the present paper in more detail. The first main result is that \(\pi_*L_{K(2)}S^0\otimes \mathbb{Q}\) is an exterior algebra on \(2\) generators in degrees \(-1\) and \(-3\). The second main result strengthens this to a computation of \(L_1L_{K(2)}S^0\) as \(L_1(S^0_3\vee S^{-1}_3) \vee L_0(S^{-3}_3\vee S^{-4}_3)\). Note that \(L_1S^0\) is a direct summand as predicted by the chromatic splitting conjecture and that the homotopy groups of \(L_1\) and \(L_0\) of the sphere are completely known. The methods rely on earlier work by \textit{P. G. Goerss} et al. [Ann. Math. (2) 162, No. 2, 777--822 (2005; Zbl 1108.55009)] and \textit{H.-W. Henn} et al. [Math. Z. 275, No. 3--4, 953--1004 (2013; Zbl 1294.55003)], who produce a resolution of the \(K(2)\)-local sphere and a corresponding algebraic resolution and obtain calculations for \(\pi_*L_{K(2)}S^0/3\). This is combined with the spectral sequence \(H^s(\mathbb{G}_2, (E_2)_t)\Rightarrow \pi_{t-s}L_{K(2)}S^0\), where \(\mathbb{G}_2\) denotes the Morava stabilizer group and \(E_2\) Morava \(E\)-theory.