Nonlinear identities with skew derivations (Q298037)

Let \(R\) be a prime ring with left Utumi ring of quotients \(U\) and denote by \(\mathbf{c}\) the extended centroid of \(R\) (i.e., the center of \(U\)). A skew derivation of \(R\) (also called \((g,h)\)-derivation) is a map \(\delta: x\to x^\delta\in U\) such that \((x+y)^\delta=x^\delta+y^\delta\) and \((xy)^\delta=x^gy^\delta+x^\delta y^h\) for all \(x,y\in R\) where \(g,h\) are automorphisms of \(R\). Let \(\phi\) be a polynomial with coefficients in \(R\) and variables acted by composition products of finitely many automorphisms and skew derivations of \(R\). Then \(\phi\) is an identity with skew derivations of \(R\) if it vanishes identically for any evaluation of its variables in \(R\). This paper is focused on the study prime rings satisfying generalized identities with skew derivation in the line of Kharchenko's theory of differential identities with automorphisms, taking advantage of the notion of expansion closed word sets previously introduced by the author [J. Algebra 224, No. 2, 292--335 (2000; Zbl 0952.16025)]. An expansion closed word set is a set \(\Omega\) of symbols such that: {\parindent=8mm \begin{itemize}\item[(i)] \(\Omega=\bigcup_{n\geq0} \Omega_n\), where the subsets \(\Omega_n\) satisfy \(\Omega_{n-1}\subseteq \Omega_n\) for all \(n\geq0\). \item[(ii)] Each \(g\in \Omega_0\) is associated with an automorphism of \(R\) denoted by \(g:x\to x^g\) and the polynomial \(\pi_g(x,y)=x^gy^g\) called the expansion formula of \(g\in \Omega_0\). \item[(iii)] Let \(n\geq1\). Each symbol \(\Delta \in \Omega_n - \Omega_0\) is associated to a map \( R\to U\), denoted \(\Delta:x\to x^\Delta\), and a polynomial \[ \pi_\Delta(x,y) = x^\Delta y^h+x^gy^\Delta+\sum_i a_ix^{\Delta_i} b_i y^{\Delta_i'}c_i, \] where \(g,h\in\Omega_0\), \(a_i,b_i,c_i\in U\) and \(\Delta_i,\Delta_i'\in\Omega_{n-1}\), such that it holds \((x+y)^\Delta=x^\Delta+y^\Delta\) and \((xy)^\Delta=\pi_\Delta(x,y)\) for all \(x,y\in R\). Here \((g,h)\) is called the type of \(\Delta\) and \( \pi_\Delta(x,y)\) the expansion formula of \(\Delta \in \Omega_n - \Omega_0\). \end{itemize}} Examples of expansion closed word sets include (for a proper definition of the maps and expansion formulas related to its elements): {\parindent=8mm \begin{itemize}\item[(i)] \(\Omega\) with \(\Omega_0\) a semigroup \(G\) of automorphisms of \(R\) and \(\Omega_n\) the set of products of finitely many \(g\in \Omega_0\) and \(\nu\) factors in \(L_G\), for some \(0\leq \nu\leq n\), where \(L_G\) denotes the set of all \((g,h)\)-derivations of \(R\) with \(g,h\in G\). \item[(ii)] Given a (pointed) coalgebra \(C\) over a field \(\mathbf{k}\), take \(\Omega_0= \mathbf{k} G\), where \(G\) denotes the set of group-like elements of \(C\) and \(\Omega_n=\bigcup_{\sigma,\tau,n} C_{\sigma,\tau,n}\) where \(\sigma,\tau\in G\) and, for \(n\geq1\), \(C_{\sigma,\tau,n}=\{h\in C_n\mid \Delta(h)= \sigma\otimes h+h\otimes\tau+C_{n-1}\otimes C_{n-1}\}\) with \(C_n=\Delta^{-1}(C\otimes C_{n-1}+C_0\otimes C)\) being \(C_0(=\mathbf{k} G=\Omega_0)\) the coradical of \(C\). \end{itemize}} For a subset \(\Sigma\) of an expansion closed word set \(\Omega\), let \( \mathfrak{p}(\Sigma)\) be the set of all generalized polynomials with coefficients in \(U\) and variables acted by \(\Delta\in \Sigma\). The elements of \( \mathfrak{p} (\Sigma)\) are called differential polynomials. Linear differential polynomials are those involving only one variable \(x\) and therefore of the form \(\sum_ia_ix^{\Delta_i} b_i\), with \(a_i,b_i\in U\) and \(\Delta_i \in\Omega\). Similarly multilinear differential polynomials are those linear in all the appearing variables. A differential polynomial \( \varphi\in \mathfrak{p} (\Sigma)\) is a differential identity in \(R\) if it vanishes for any evaluation of its variables in \(R\). A subset \(\Sigma\subseteq \Omega\) is a basis of \( \Omega\) if \( \mathfrak{p}(\Sigma)\cap \mathfrak{J} =\emptyset\) and \( \mathfrak{p}(\Sigma)+\mathfrak{J}= \mathfrak{p}(\Omega)\), where \(\mathfrak{J}\) denotes the ideal of \( \mathfrak{p}(\Omega)\) generated by all linear identities. Given a basis \(\Sigma\) of an expansion closed word set \(\Omega\) (bases for expansion closed word sets were proven to always exist), the elements \(\varphi\in \mathfrak{p}(\Omega)\) have a unique (so-called) \(\Sigma\)-reduced form in \( \mathfrak{p}(\Sigma)\) uniquely written as \(f(x_i^{\delta_j})\) with \(\delta_j\in \Sigma\) all distinct, \(x_i\) distinct variables and \(f(z_{ij})\) a generalized polynomial in distinct non-commuting variables \(z_{ij}\) with coefficients in \(U\). Clearly a differential polynomial \(f(x_i^{\delta_j})\) is nonzero if and only if so is the generalized polynomial \(f(z_{ij})\). This paper proves that (under the above assumptions), given \(f(x_i^{\delta_j})\) an identity of \(R\), with distinct \(\delta_j\in \Sigma\), distinct non-commuting variables \(x_i\) and such that \(f(z_{ij})\) is a nonzero generalized polynomial in distinct non-commuting variables \(z_{ij}\) with coefficients in \(U\): {\parindent=8mm \begin{itemize}\item[(i)] If the extended centroid \(\mathbf{c}\) of \(R\) is a perfect field, then \( \Sigma\subseteq \Omega_0\) and \(f(x_i^{\delta_j})\in \wp(\Omega_0)\). \item[(ii)] If the extended centroid \(\mathbf{c}\) is not a perfect field, then \(f(z_{ij})\) is also an identity of \(R\). \end{itemize}}