An upper bound on the LS category in presence of the fundamental group (Q2292411)

In this paper, the author establishes a meaningful improvement of the so-called Grossman-Whitehead estimates of the LS-category. The first approximation of this type states that \(cat_{LS}X \leq \frac{1}{2} \dim X\) for any simply connected CW-complex \(X\). But, in the presence of the fundamental group, it turns out (cf. Proposition 2.6) that, for any essential CW-complex \(X\), (i.e. one for which no map \(u_X: X\rightarrow B\pi _1(X)\) that classifies its universal cover can be deformed to the \((n-1)\)-skeleton \(B\pi_1(X)^{(n-1)}\)) one has \(cat_{LS}X = \dim X\) and inversely. Later, the author has shown that this Grossman-Whitehead estimate (whose existence was conjectured by Y. Rudyak) holds for complexes with the fundamental group having small cohomological dimension \(cd(\pi_1(X))\), namely, it was shown in [\textit{A. Dranishnikov}, Algebr. Geom. Topol. 10, No. 2, 917--924 (2010; Zbl 1194.55008)] that: \[cat_{LS}(X) \leq cd(\pi_1(X)) + 1 / 2\dim X.\] Nevertheless, this inequality is far from being optimal when \(cd(\pi_1(X))\geq 1 / 2\dim X\) as it is shown through the particular case \(cat_{LS}(M\times \mathbb{C}P^n)=m+n < 3/2m +n =cd(\pi _1(X)) + 1 / 2\dim X\) where \(M\) is an aspherical \(m\)-manifold. The desired upper bound \[ cat_{LS}(X) \leq 1 / 2[cd(\pi_1(X)) + \dim X] \] which gives the optimal estimate for the above example is then considered by the author, especially that it is already shown valid for the LS-systolic category [\textit{L. F. Wang}, Pac. J. Math. 254, No. 2, 449--464 (2011; Zbl 1253.53038)], a differential geometry relative of the LS category. The above inequality stated as Corollary 4.2 is obtained as a consequence of the following one (Theorem 4.1): \[ cat_{LS}(X) \leq 1 / 2[cat_{LS}(u_X) + \dim X] \] Based on a further modification of the Kolmogorov-Ostrand multiple cover technique [\textit{A. N. Dranishnikov}, Proc. Am. Math. Soc. 137, No. 4, 1489--1497 (2009; Zbl 1168.55002)], the technical proposition (Proposition 3.1) is proved and used together with Borel construction machinery to obtain the Lemma 3.2 used to achieve the proof of Theorem 4.1. To obtain Corollary 4.2, two cases had to be distinguished. First, when \(cd(\pi_1(X))\not =2\), this is a consequence of Theorem 4.1, then, when \(cd(\pi_1(X)) =2\), using obstruction theory and some intermediate results, the author was lead to replace \(u_X: X\rightarrow B\pi _1(X)\) by \(r\circ u_X: X\rightarrow B\pi _1(X)^{(2)}\) where \(r: B\pi _1(X)\rightarrow B\pi _1(X)^{(2)}\) is a map which is the identity on the \(1\)-skeleton and induces an isomorphism on fundamental groups. Thus, he applies Theorem 4.1 for \(r\circ u_X\) in this case. The author ends the paper by establishing more general results, namely Theorem 4.5 and Corollary 4.6 which give upper bounds of \(cat_{LS}(X)\) for any CW complex \(X\) with \(r\)-connected universal covering \(\tilde{X}\).