Rigid division algebras. (Q2342108): Difference between revisions

Let \(R\) be a ring (necessarily associative) and \(R_{\text{smgr}}\) its multiplicative semigroup. The ring \(R\) is said to be rigid, if the operation of addition in \(R\) can be reconstructed from the definition of multiplication in \(R\). Rigidity holds, if \(R\) is an isomorphically rigid ring (a UA-ring), i.e. for any ring \(S\), every semigroup isomorphism \(R_{\text{smgr}}\to S_{\text{smgr}}\) is determined by some ring isomorphism \(R\to S\). The study of UA-rings dates back at least to the middle of the 20th century (see \textit{C. E. Rickart} [Bull. Am. Math. Soc. 54, 758-764 (1948; Zbl 0032.24904)], \textit{R. E. Johnson} [Proc. Am. Math. Soc. 9, 57-61 (1958; Zbl 0081.26304)] and \textit{L. M. Gluskin} [Izv. Akad. Nauk SSSR, Ser. Mat. 23, 841-870 (1959; Zbl 0090.32801)]). As shown by \textit{W. Stephenson} [Can. J. Math. 21, 1455-1461 (1969; Zbl 0188.08301)], for any ring \(A\) with a unit, both the full matrix ring \(M_n(A)\), where \(n\geq 2\), and its subring of upper triangular matrices are UA-rings (for other types of UA-rings, see \textit{A. V. Mikhalev} [Mat. Sb., Nov. Ser. 135(177), No. 2, 210-224 (1988; Zbl 0645.16024)]). The paper under review proves the existence of finite-dimensional central division algebras \(D\) over some fields \(E\), which are not UA-rings. It shows that one may take as \(D\) any division \(E\)-algebra of the considered type, if the multiplicative group \(E^*\) of \(E\) has a torsion-free nilpotent endomorphism. The author observes that \(E^*\) has the described property in several frequently used special cases: in particular, such is the case where \(E\) is a finitely-generated extension of its prime subfield \(\mathbb F\) and either \(\text{char}(E)=0\) or \(E\) is transcendental over \(\mathbb F\). He also proves that \(D\) is not a UA-ring, if \(D\) is a generalized quaternion division \(E\)-algebra and \(E\) is any field with \(\text{char}(E)\neq 2\). The results of the reviewed paper agree with the author's conjecture that a finite-dimensional central division algebra over an arbitrary field is a UA-ring if and only if it is a field with \(2\), \(3\) or \(4\) elements.