On certain cohomological invariants of groups. (Q607350)

\textit{T. V. Gedrich} and \textit{K. W. Gruenberg} [in Topology Appl. 25, 203-223 (1987; Zbl 0656.16013)], considered the invariants \(\mathrm{silp\,}\mathbb ZG\), the supremum of the injective lengths of projective \(\mathbb ZG\)-modules, and \(\mathrm{spli\,}\mathbb ZG\), the supremum of the projective lengths of injective \(\mathbb ZG\)-modules, in their study of complete cohomological functors. They showed that for any group \(G\) \(\mathrm{silp\,}\mathbb ZG\leq\mathrm{spli\,}\mathbb ZG\), and that if \(\mathrm{spli\,}\mathbb ZG<\infty\) then \(\mathrm{silp\,}\mathbb ZG=\mathrm{spli\,}\mathbb ZG\). In this paper the author shows that for any group \(G\), \(\mathrm{spli\,}\mathbb ZG\leq\mathrm{silp\,}\mathbb ZG\). Actually, he shows that if \(\mathrm{sfli\,}\mathbb ZG\) denotes the supremum of the flat lengths of injective \(\mathbb ZG\)-modules, then \(\mathrm{sfli\,}\mathbb ZG\leq\mathrm{silp\,}\mathbb ZG\). That \(\mathrm{spli\,}\mathbb ZG\leq\mathrm{silp\,}\mathbb ZG\) is now implied from \textit{C. U. Jensen}'s result [Math. Scand. 18, 97-105 (1966; Zbl 0145.26605)] that if \(\mathbb ZG\) has finite finitistic dimension then every flat \(\mathbb ZG\)-module has finite projective dimension. Related results are also obtained for rings which are left and right \(\aleph_0\)-Noetherian. As an application the author shows a conjecture stated by \textit{B. M. Ikenaga}, [in J. Pure Appl. Algebra 40, 125-134 (1986; Zbl 0586.20028)], that a group \(G\) is finite if and only if the generalized cohomological dimension of \(G\), \(\underline{\mathrm {cd}}\,G\), is zero, where \[ \underline{\mathrm{cd}}\,G=\sup\{n\in\mathbb N\mid\mathrm{Ext}^n_{\mathbb ZG}(M,F)\neq 0,\;M\;\mathbb Z\mathrm{-free},\;F\;\mathbb ZG\mathrm{-free}\}. \] To prove this conjecture, the author uses a result of \textit{F. Dembegioti} and \textit{O. Talelli} [J. Algebra 319, No. 1, 267-271 (2008; Zbl 1232.20051)] which says that the finiteness of a group \(G\) is equivalent to the equality \(\mathrm{spli\,}\mathbb ZG=1\). The author's main tools are the Mittag-Leffler condition and a certain natural transformation which relates \(\mathrm{Tor}^R_*\) with \(\mathrm{Ext}^*_R\).