The second relative homology of Leibniz algebras (Q2318384)

Studied in the paper under review are the relative Leibniz homology groups in dimension two of a pair \(( \mathfrak{g}, \, \mathfrak{n})\), where \(\mathfrak{g}\) is a right Leibniz algebra over a field \(\mathbf{F}\) and \(\mathfrak{n}\) is a two-sided ideal of \(\mathfrak{g}\). Recall that a right Leibniz algebra in this case is a vector space over \(\mathbf{F}\) equipped with a bilinear map \(\circ :\mathfrak{g} \times\mathfrak{g} \to \mathfrak{g}\) that is a derivation from the right, meaning that \[ ( x \circ y) \circ z = x \circ ( y \circ z) + ( x \circ z) \circ y, \quad x, y, z \in\mathfrak{g}. \] The paper adopts the older notation with \(x \circ y\) denoted by \([x, y]\), although the bracket is not necessarily skew-symmetric. The Leibniz homology of \(\mathfrak{g}\) with coefficients in \(\mathbf{F}\) is denoted \(HL_* (\mathfrak{g} )\), and is the homology of the chain complex \[ CL_k (\mathfrak{g} ) := ( \mathfrak{g}^{\otimes k}, \, d), \quad k \geq 0, \] where the boundary map \(d\) is defined in [\textit{J.-L. Loday} and \textit{T. Pirashvili}, Math. Ann. 296, No. 1, 139--158 (1993; Zbl 0821.17022)]. Let \(\mathfrak{n}\) be a two-sided ideal of \(\mathfrak{g}\) and let \(\pi :\mathfrak{g} \to\mathfrak{g}/\mathfrak{n}\) be the quotient map of Leibniz algebras. There is chain map \[ \pi : (\mathfrak{g}^{\otimes k}, \, d) \to ( (\mathfrak{g}/ \mathfrak{n})^{\otimes k}, \, d) \] with mapping cone \[ M_k( \pi ) = (\mathfrak{g}^{\otimes (k-1)} \oplus (\mathfrak{g}/\mathfrak{n})^{\otimes k}, \, \delta ). \] Then \(( M_* ( \pi ), \, \delta )\) becomes a chain complex with the authors using the sign convention \[ \delta (a, \, b) = ( -d(a), \, \pi(a) + d(b)), \ \ \ a \in\mathfrak{g}^{\otimes (k-1)}, \ b \in (\mathfrak{g}/\mathfrak{n})^{\otimes k}. \] By standard results in homological algebra, there is a long exact sequence in homology: \[ \longrightarrow HL_k (\mathfrak{g}/ \mathfrak{n}) \longrightarrow H_k (M_*( \pi )) \longrightarrow HL_{k-1} (\mathfrak{g}) \overset{\pi_*}{\longrightarrow} HL_{k-1}(\mathfrak{g}/\mathfrak{n}) \longrightarrow \] By definition the relative homology groups are \(HL_k (\mathfrak{g}, \, \mathfrak{n} ) := H_{k+1} ( M_* (\pi) )\). A basic result about \(HL_2 (\mathfrak{g}, \, \mathfrak{n} )\) is given in [\textit{G. Donadze} et al., Rev. Mat. Complut. 31, No. 1, 217--236 (2018; Zbl 1403.18015)], where the non-abelian exterior product \(\mathfrak{m} \curlywedge \mathfrak{n}\) of two ideals \(\mathfrak{m}\), \(\mathfrak{n}\) of \(\mathfrak{g}\) is given, and the kernel of the commutator map \([ \ , \ ] : \mathfrak{g} \curlywedge \mathfrak{n} \to\mathfrak{n}\) is proven to be isomorphic to \(HL_2 (\mathfrak{g}, \, \mathfrak{n} )\). Let \(\mathfrak{g}^2 = [\mathfrak{g}, \, \mathfrak{g}]\) be the derived subalgebra of \(\mathfrak{g}\) and let \(Z( \mathfrak{g} )\) denote the center of \(\mathfrak{g}\). Proven in the paper under review is that if \(\mathfrak{n} \subseteq Z(\mathfrak{g} ) \cap \mathfrak{g}^2\), then \[ HL_2 (\mathfrak{g}, \, \mathfrak{n} ) \simeq ( \mathfrak{g}/\mathfrak{g}^2 \otimes \mathfrak{n}) \oplus (\mathfrak{n} \otimes \mathfrak{g}/\mathfrak{g}^2 ) . \] Suppose that \(\mathfrak{r} \to\mathfrak{f} \overset{\pi}{\to} \mathfrak{g}\) is a free presentation of \(\mathfrak{g}\), where \(\mathfrak{f}\) is the free Leibniz algebra on the set \(\mathfrak{g}\). Suppose further that \(\eta\) is an ideal of \(\mathfrak{f}\) generated by \(\mathfrak{n}\). Proven is the ``Hopf formula'' \[ HL_2 (\mathfrak{g}, \, \mathfrak{n} ) \simeq \frac{ \mathfrak{r} \cap [ \eta , \, \mathfrak{f} ]} { [\mathfrak{r}, \, \eta] + [\mathfrak{f}, \,\mathfrak{r} \cap \eta]}. \] Furthermore, the notions of a stem extension and a stem cover from [\textit{J. M. Casas} and \textit{M. Ladra}, Georgian Math. J. 9, No. 4, 659--669 (2002; Zbl 1051.17001)] are generalized to the relative case, and it is proven that the pair \((\mathfrak{g}, \,\mathfrak{n})\) admits a relative stem cover. Let \(\mathfrak{g}\) be a Leibniz algeba with \(\dim_{\mathbf{F}} (\mathfrak{g}/\mathfrak{n}) = m\), \(\dim_{\mathbf{F}} ( \mathfrak{n} / Z(\mathfrak{g}, \,\mathfrak{n}) ) = n\). Proven is that (i) \(\dim_{\mathbf{F}} ( [\mathfrak{g}, \,\mathfrak{n}] ) \leq n(n + 2m)\), and (ii) if \(\mathfrak{g}\) is nilpotent and \(\dim_{\mathbf{F}} ( [\mathfrak{g}, \,\mathfrak{n} ]) = n(n + 2m)\) or \(n(n + 2m) - 1\), then \( \mathfrak{n}/ Z(\mathfrak{g}, \, \mathfrak{n})\) is a central ideal of \(\mathfrak{g} / Z( \mathfrak{g}, \,\mathfrak{n})\). Let \(d(\mathfrak{g})\) denote the minimal number of generators of the Leibniz algebra \(\mathfrak{g}\). If \(\mathfrak{g}\) is finite dimensional and nilpotent, then \(d(\mathfrak{g} ) = \dim_{\mathbf{F}} (\mathfrak{g} / \mathfrak{g}^2 )\). For such a Leibniz algebra \(\mathfrak{g}\) with an ideal \(\mathfrak{n}\) of codimension 1, proven is that \[ \dim_{\mathbf{F}} ( HL_2 ( \mathfrak{g}, \, \mathfrak{n} ) ) \leq\dim_{\mathbf{F}} (HL_2 (\mathfrak{n} )) + 2 d(n). \] Equality of the bound is attained if \(Z( \mathfrak{g} ) \nsubseteq \mathfrak{n}\).