On the class-breadth conjecture for algebras and \(\mathcal{T}\)-groups (Q7294307)

The paper under review develops an algorithm for computing the breadth of a finite-dimensional algebra \(L\) over an arbitrary field from its structure constants. If \(B\) is a basis and \(M_B(X)\) is the matrix of right multiplication by a generic element \(X\), then \(\operatorname{br}(L)\) is the largest size of a minor of \(M_B(X)\) that does not vanish on the ground field. Over a finite field of order \(q\), polynomial vanishing can be tested by reduction modulo the Gröbner basis \(\{X_i^q-X_i\mid 1\leq i\leq n\}\).\N\NFor a nilpotent algebra of class \(c\), a basis adapted to the lower central series gives a block upper-triangular form of \(M_B(X)\) and yields a nonzero \((c-1)\times(c-1)\) minor. Consequently, the class-breadth inequality \(\operatorname{cl}(L)\leq\operatorname{br}(L)+1\) holds over every infinite field. Over a finite field, it holds after extending scalars to a sufficiently large finite extension. Counterexamples over finite fields are examined by determining when the relevant minors vanish.\N\NFinally, for a finitely generated torsion-free nilpotent group \(G\), the authors consider the rational Lie algebra \(\Lambda(G)\) associated with its \(\mathbb{Q}\)-powered hull and prove that \(\operatorname{br}(G)=\operatorname{br}(\Lambda(G))\). This makes the breadth of \(G\) accessible to the algebraic algorithm and gives an alternative proof of the class-breadth conjecture for \(\mathcal{T}\)-groups. The possible types and breadths for \(\mathcal{T}\)-groups of Hirsch length at most \(6\) are also determined.