Counterexample to a conjecture about braces. (Q5965114)

The paper provides a counterexample for a conjecture which arose in connection with braces, a ring-like structure generalizing radical rings. The adjoint group of a brace is connected with the additive group by a 1-cocycle, and conversely, every bijective 1-cocycle comes from a brace. For a finite brace, the adjoint group is solvable. So the question arose whether conversely, every finite solvable group is the adjoint group of a brace. For \(\mathbb R\)-linear braces, a similar question is Milnor's 1977 conjecture, equivalent to the assertion that every simply connected solvable Lie group is isomorphic to the adjoint group of an \(\mathbb R\)-brace [\textit{J. W. Milnor}, Adv. Math. 25, 178-187 (1977; Zbl 0364.55001)]. Milnor proved that the Lie groups in question are exactly the connected Lie groups which admit a free action of the affine group on a finite space. Such groups are solvable. However, the converse was believed to be true until \textit{Y. Benoist} constructed a counterexample [J. Differ. Geom. 41, No. 1, 21-52 (1995; Zbl 0828.22023)] by means of a filiform Lie algebra of dimension 11, which led to a nilvariety without an affine structure. The paper extends this to the discrete case where some battle with the characteristic has to be fought.