Absolute retracts and essential extensions in congruence modular varieties (Q2377313)

In this paper the author studies absolute retracts in congruence modular varieties. An algebra \(\mathbb{ R}\) of a variety \(\mathcal{V}\) is said to be an \textit{absolute retract} in \(\mathcal{V}\) if for any embedding \(e:\mathbb{ R}\hookrightarrow \mathbf{ A}\) , \(\mathbf{ A}\in \mathcal{V}\), there is a surjective homomorphism \(p:\mathbf{ A}\twoheadrightarrow \mathbb{ R}\) such that \(p\circ e=\mathrm{id}_{R}\). If \(\mathbf{ A}\) is an algebra and \(\alpha\in \text{Con}(\mathbf{ A})\), \[ \begin{aligned} \alpha^{*}:=\{(a,b)\in A^{2}:\forall t\in \text{Clo}_{n+1}(\mathbf{ A}) \forall (\mathbf{ c},\mathbf{ d})\in \alpha, \\ t(a,\mathbf{ c})=t(a,\mathbf{ d}) \Longleftrightarrow t(b,\mathbf{ c})=t(b,\mathbf{ d})\}\end{aligned} \] is a congruence called the \textit{centralizer} of \(\alpha\). The \textit{center} of \(\mathbf{ A}\) is the congruence \(\zeta_{\mathbf{ A}}:=(1_{\mathbf{ A}})^{*}\). \(\mathbf{ A}\) is said to be \textit{abelian} if \(\zeta_{\mathbf{ A}}=1_{\mathbf{ A}}\) and \textit{centerless} if \(\zeta_{\mathbf{ A}}=0_{\mathbf{ A}}\). For \(\alpha, \beta\in \text{Con}(\mathbf{ A})\), the congruence \[ [\alpha,\beta]:=\bigcap\{\gamma\in \text{Con}(\mathbf{ A}): \gamma\subseteq\alpha\wedge \beta \text{ and } (\alpha/\gamma)\subseteq(\beta/\gamma)^{*}\} \] is called the \textit{commutator} of \((\alpha,\beta)\). A variety is said to satisfy the \textit{commutator identity} if the congruences of their algebras satisfies the following: \[ \alpha\wedge[\beta,\beta]=[\alpha\wedge \beta,\beta]. \] The main result of the paper is this: If \(\mathbf{ A}\) is an absolute retract in a congruence modular variety and if Con(\(\mathbf{ A}\)) is of finite height then \(\mathbf{ A}\) is a product of subdirectly irreducible algebras. Further, if the variety satisfies the commutator identity then \(\mathbf{ A}\) is the product of a centerless algebra and an abelian algebra. In particular, if \(\mathbf{A}\) is directly indecomposable, then \(\mathbf{ A}\) is either centerless or abelian. Since a lattice is finite-dimensional iff the lattice is modular and of finite height (see Theorem 2.42 of [\textit{R. N. McKenzie} et al., Algebras, lattices, varieties. Volume I. Monterey, California: Wadsworth \& Brooks/Cole Advance Books \& Software (1987; Zbl 0611.08001)]), the main result of the paper, as stated in the abstract of the paper, is correct. But I think it better as stated in this review since, after all, what is used in the proof is the fact that the lattice congruence is of finite height.