2208.11624
Faithful Invariant Random Subgroups in Acylindrically Hyperbolic Groups
Yair Glasner, Anton Hase
incompletemedium confidence
- Category
- Not specified
- Journal tier
- Strong Field
- Processed
- Sep 28, 2025, 12:56 AM
- arXiv Links
- Abstract ↗PDF ↗
Audit review
The paper proves Theorem 3.4 via Sun’s Cohen–Lyndon property and co-induction to obtain a weakly mixing, nontrivial, faithful IRS for every acylindrically hyperbolic group, with a complete and correct proof sketch (see the abstract and Theorem 3.4 together with Proposition 3.3 and its construction of µG via induction/co-induction, and the kernel/closure calculations) . For Theorem 3.6, the paper constructs an IRS inside Aut(G) supported in Inn(G) and then states that “we can see it as a characteristic random subgroup of G,” calling it faithful; this step is under-specified unless one assumes G is center-free, because passing from Inn(G) ≅ G/Z(G) back to Sub(G) by pullback forces the kernel to contain Z(G). The paper’s text gives no center-free hypothesis at that point . The candidate solution reproduces the core pipeline (Cohen–Lyndon + induction/co-induction) but uses a different base IRS (random selection of free factors) and explicitly handles the center issue, yielding a faithful characteristic IRS on G when Z(G)=1 and otherwise on G/Z(G), which fixes the missing hypothesis.
Referee report (LaTeX)
\textbf{Recommendation:} minor revisions \textbf{Journal Tier:} strong field \textbf{Justification:} The paper gives a clean, conceptually natural synthesis of Sun’s Cohen–Lyndon theorem with Kechris–Quorning co-induction, establishing the existence of weakly mixing, nontrivial, faithful IRSs in all acylindrically hyperbolic groups, and extends this to characteristic IRSs for hyperbolic groups. The main construction (decreasing chain (Γ\_k), induction, and co-induction) is clear and correct. One presentational point needs clarification: when passing from an IRS in Aut(G) supported in Inn(G) to a characteristic IRS on G, faithfulness on G requires Z(G)=1; otherwise, the natural pullback has kernel Z(G). Addressing this would strengthen the statement of Theorem 3.6 without altering the core ideas.