Almost all one-relator groups with at least three generators are residually finite. (Q621852)

The paper considers three natural models for choosing a random one-relator group. The main result is that under any of these models a random \(k\)-generator one-relator group, \(k\geq 3\), can be embedded into an ascending HNN extension of a finitely generated free group with probability 1. Therefore almost surely, such a group is residually finite, is a virtually residually (finite \(p\)-)group for every sufficiently large prime \(p\), and is coherent (i.e. every finitely generated subgroup is finitely presented). Quoting from the introduction: ``[The proof] uses some strong results from different areas of mathematics that are very rarely employed together: geometric group theory, algebraic geometry and probability theory (Brownian motion). Namely, we use the result of Feighn and Handel that ascending HNN extensions of free groups are coherent (that is mostly geometric group theory, more precisely discrete Morse theory), results by Borisov and Sapir that every ascending HNN extension of a free group is residually finite and even a virtually residually (finite \(p\)-)group for almost all primes \(p\) (that is essentially a result from algebraic geometry, namely the theory of quasi-fixed points of polynomial maps over finite fields), a result of Ol'shanskiĭ about subgroups with congruence extension property of hyperbolic groups (this is geometric group theory), a result of Kapovich, Schupp and Shpilrain about generic solvability of isomorphism problem for one-relator groups (this result uses many parts of geometric group theory including the Arzhantseva-Ol'shanskiĭ method, boundaries of hyperbolic groups, etc.), and a result by Cranston, Hsu, and March that the boundary of the convex hull of a Brownian trajectory is smooth almost surely in Wiener's measure (probability theory).'' The final section discusses the 2-generator case and poses some related questions.