On the unit conjecture for supersoluble group algebras. (Q2350637)

Let \(G\) be a torsion-free group, \(K\) any field. The long-standing conjectures due to I. Kaplansky, namely that the group ring \(KG\) is without zero divisors and, the stronger one, that the group algebra \(KG\) has trivial units only, were investigated intensively in the past decades. By the classical result of D. S. Passman and S. K. Sehgal, the trivial unit conjecture holds true for up-groups. A famous example \(P=\langle x,y\mid xy^2x^{-1}=y^{-2},\;yx^2y^{-1}=x^{-2}\rangle\) of a torsion-free non-up group was given by S. D. Promislow, it is known that \(P\) is super-solvable and satisfies the zero divisor conjecture. Generalizations of the Promislow group have been studied recently by \textit{W. Carter} [J. Group Theory 17, No. 3, 445-464 (2014; Zbl 1300.20041)]. Promislow found a 14-element subset \(\mathcal P\) such that \(\mathcal P\cdot\mathcal P\) has no unique product. The main result of the authors is that no subset of at least two elements of the Promislow set \(\mathcal P\) may provide the support of a unit over \(K\). A group \(G\) is an \((X,Y,N)\)-group if there exist subgroups \(X,Y\) and a normal subgroup \(N\) with \(G/N=X/N*Y/N\) the infinite dihedral group. If \(KX\) and \(KY\) have no proper divisors of zero and \(KN\) is an Ore domain then \(KG\) is said to be an \((X,Y,N)\)-group algebra. Let \(X/N\) and \(Y/N\) be generated by the involutions \(Nx\) and \(Ny\), and let \(W\) be the set of all alternating words in \(x\) and \(y\). The length \(L(w)\) is defined to be the number of factors \(x,y\), for \(g\in Nw\) \(L(g)=L(w)\); for \(0\neq\alpha\in KG\) set \(L(\alpha)=\max\{L(g)\mid g\in\text{Supp\,}\alpha\}\), \(L(0)=-\infty\). A linear term in \(KG\) is defined to be an expression of form \(\Lambda=\alpha+\beta u\) for some \(\alpha,\beta\in KN\) and \(u\in\{x,y\}\). The product \(\Lambda_1\cdots\Lambda_s\) of non-zero linear terms is called a splitting in \(KG\), it is reduced if its length is either 0 or \(s\). The authors reprove the theorem that an \((X,Y,N)\)-group algebra satisfies the zero divisor conjecture by establishing important properties such as that if \(0\neq\sigma,\tau\in KG\) with \(\sigma\tau\in KN\) then \(L(\sigma)=0\) if and only if \(L(\tau)=0\). If \(N\) is finitely generated abelian, \(\sigma\) is a unit of length \(s\geq 1\) in \(KG\), then \(\varepsilon\sigma=(\alpha_1+\beta_1u_1)\cdots(\alpha_s+\beta_su_s)\) is a reduced splitting for some \(\varepsilon\in KN\setminus\{0\}\) such that \(\alpha_i\) and \(\beta_i\) are relatively prime for all \(i\). If \(H\) is a subgroup of finite index in \(G\) then there is a regular embedding \(\rho\) of \(KG\) in \(M_n(KH)\). If \(H\) is torsion free abelian, normal, \(Z(G)\) is trivial then \(\sigma\in U(KG)\) if and only if \(|\rho(\sigma)|\in U(K)\). As \(P\) is an \((X,Y,N)\)-group with \(X=\langle x,y^2\rangle\), \(Y=\langle x^2,y\rangle\) and \(N=\langle x^2,y^2\rangle\), these (and the numerous other) results of the paper can be used to investigate \(KP\) and prove the above main result, and that there are trivial units only of length at most 3. Moreover, the authors show that if \(\sigma\) is a unit of length 4 in \(KP\) with reduced split-form \(\sigma=(\alpha_1 +\beta_1x)(\alpha_2 +\beta_2y)(\alpha_3+\beta_3x)(\alpha_4+\beta_4y)\), \(\beta_2\) and \(\beta_3\) are units in \(KN\) then \(\sigma\) is a trivial unit, and provide additional guidelines to prove or refute the trivial unit conjecture for \(KP\).