Injective endomorphisms of \({\mathcal G}_ X\)-normal semigroups: Infinite defects (Q1188317)

Let \(X\) be an infinite set and \({\mathcal G}_ X\) be the symmetric group on \(X\). For a mapping \(f: X\to X\), the defect of \(f\) is defined by \(\text{def }f = | X - f(X)|\). A semigroup \(S\) of transformations of \(X\) is said to be \({\mathcal G}_ X\)-normal if for every \(h\in {\mathcal G}_ X\), \(hSh^{-1} \subseteq S\). Injective endomorphisms of \({\mathcal G}_ X\)-normal semigroups of one-to-one transformations with infinite defects smaller than \(| X|\) are described. To formulate this exactly, let us bring some more definitions. For \(f\in S\), denote \(\text{shift }f = | \{x\in X: f(x) \neq x\}|\). For \(T\subset S\), define \(\sigma\text{-def }T = \{\text{def }f: f\in T\}\). Denote \(\eta = \min\{\alpha: \alpha > \text{def }f\), for all \(f\in S\}\). If \(\alpha\) is an infinite cardinal, let \(\Delta_ \alpha\) be the congruence on \(S\) such that \((f,g)\in \Delta_ \alpha\) if and only if \(|\{x\in X: f(x) \neq g(x)\}| < \alpha\). For \(\alpha \in \sigma\text{-def }S\), denote \(S(\alpha) = \{f\in S: \text{def }f = \alpha\}\) and \(\mu(\alpha) = \min\{\beta: \beta >\text{shift }f,\;f \in S(\alpha)\}\). Theorem. Let \(\varphi\) be an injective endomorphism of \(S\). There exist (i) a subset \(W\) of \(X\) with \(| W| = | X|\) and a partition \(\{X_ i: i \in I\}\) of \(W\) such that \(| X_ i| = | X|\), \(| I|\in \sigma\text{-def }S\) or \(| I| \leq \nu = \min \sigma\text{-def }S\), (ii) a homomorphism \(\tau: \Omega \to {\mathcal G}_ I\) such that for every \(\beta \not\in \ker \tau\), \(\mu(\gamma| I|) = | X|^ +\), for each \(\gamma \in \sigma\text{-def }V_ \beta\), (iii) a set of bijections \(h_ i: X\to X_ i\), \(i\in I\), and (iv) a group homomorphism \(\xi: S\to G_ \mu = \{h\in{\mathcal G}_{X-W}:\text{shift }h<\mu\}\) with the associated congruence \(\Delta_ \alpha\) on \(S\), \(\alpha \geq \eta\) satisfying \(\text{shift }\xi(g) < \mu(\gamma| I|)\), \(g\in V\), a \(\Delta_ \alpha\)-class, \(\gamma \in \sigma\text{-def }V\), such that given \(f\in V_ \beta\), \(x\in X\), \[ \varphi(f)(x) = \begin{cases} h_{r(\beta)(i)}fh_ i^{-1}(x), &\text{if \(x\in X_ i\),}\\\xi(f)(x), & \text{if \(x\in X- W\).}\end{cases}\tag{*} \] Conversely, given (i)-(iv) the mapping defined in (*) is an injective endomorphism of \(S\).