On subsemigroups of \(\beta\mathbb{N}\) and absolute coretracts (Q5952247)

A semigroup \(S\) is called (weak) absolute coretract if for any continuous homomorphism \(h\) from a (finite) compact Hausdorff right topological semigroup \(T\) onto a compact Hausdorff right topological semigroup containing \(S\) algebraically there exists a homomorphism \(s:S\to T\) such that \(h\circ s=id_S\). According to Corollary 6.5 of [\textit{N. Hindman} and \textit{D. Strauss}, Algebra in the Stone-Čech compactification, De Gruyter, Berlin (1998; Zbl 0918.22001)], the semigroup \(\beta\mathbb N\) contains an isomorphic copy of each countable absolute coretract. The author defines a class \(\mathcal C\) of countable absolute coretracts and proves that each finite weak absolute coretract semigroup of idempotents belongs to \(\mathcal C\). This implies that a finite weak absolute coretract semigroup of idempotents is an absolute coretract and consequently embeds into \(\beta\mathbb N\). This result answers a question of [\textit{N. Hindman}, Math. Jap. 44, 581-625 (1996; Zbl 0917.54031)] and yields many non-trivial examples of finite subsemigroups in \(\beta\mathbb N\). The class \(\mathcal C\) is defined as a collection of some special subsemigroups of the semigroup \(V\) whose elements are ordered pairs of words of the form \((i_1i_2\cdots i_p,\lambda_p\lambda_{p-1}\cdots\lambda_1)\), where \(i_k,\lambda_k\in\omega\), with the semigroup operation defined for pairs \(x=(i_1\cdots i_p,\lambda_p\cdots\lambda_1)\), \(y=(j_1\cdots j_q,\rho_q\cdots\rho_1)\) by \(xy=(i_1\cdots i_p,\rho_p\cdots\rho_1)\) if \(p=q\), \(xy=(i_1\cdots i_p,\lambda_p\cdots\lambda_{q+1}\rho_q\cdots\rho_1)\) if \(p>q\) and \(xy=(i_1\cdots i_pj_{p+1}\cdots j_q,\rho_q\cdots\rho_1)\) if \(p<q\). A subset \(C\subset V\) is said to satisfy property \(Z\), if for each \((i_1\cdots i_t0\cdots 0i_p,\lambda_p\cdots\lambda_1)\in C\) (where \(t\) can be equal to \(p-1\)) we have \(i_p\neq 0\) and if \(i_t\neq 0\), then \((i_1\cdots i_t,1\cdots 1)\in C\). Denote by \(V_p\subset V\) the semigroup of pairs of words of length \(p\) and for a pair \(x=(i_1\cdots i_p,\lambda_p\cdots\lambda_1)\) and a subset \(C\subset V\) let \(\overline x=(\lambda_1\cdots\lambda_p,i_p\cdots i_1)\) and \(\overline C=\{\overline x:x\in C\}\). By definition, the class \(\mathcal C\) consists of all subsemigroups \(C\) of \(V\) such that (i) \(C\cap V_p\) is finite for all \(p\), (ii) both \(C\) and \(\overline C\) satisfy property \(Z\), and (iii) for every \(t\in\mathbb N\) either \(C\subset 1_t\) or \(\overline C\subset 1_t\), where \(1_t=\{(i_1\cdots i_p,\lambda_p\cdots\lambda_1)\in V: i_t=1\) if \(p\geq t\}\).