Dice Question Streamline Icon: https://streamlinehq.com

Stability of Cheng-type rigidity in RCD*(K,N) spaces

Establish a stability version of the rigidity in Cheng’s eigenvalue comparison (Theorem Cheng dir-Rigid) for RCD*(K,N) metric measure spaces. Specifically, determine whether for every epsilon > 0 there exists delta > 0 such that if (X, d, μ) is an RCD*(K, N) space containing a ball B(x0, r) with r in (r0 − delta, r0 + delta) and satisfying λ1^D(B(x0, r)) ≥ λ_{K, N, r0} − delta, then the pointed measured space (X, d, μ, x0) is epsilon-close in the pointed measured Gromov–Hausdorff (pmGH) sense to one of the three rigid configurations described in Theorem Cheng dir-Rigid (one-dimensional interval model, one-dimensional manifold, or local (K, N)-cone over an RCD*(N−2, N−1) space).

Information Square Streamline Icon: https://streamlinehq.com

Background

The paper proves a Cheng-type upper bound for the first Dirichlet eigenvalue of balls in essentially non-branching CD* (K, N) spaces and a rigidity result on RCD* (K, N) spaces: equality in the bound forces the local geometry around the ball to be one of three rigid models (1D interval, 1D manifold, or a (K, N)-cone over an RCD* (N−2, N−1) space).

The open problem asks for a quantitative stability: if the Dirichlet eigenvalue is close to the model value and the ball’s radius is close to r0, does the ambient space necessarily lie close (in pmGH topology) to one of the rigid models? A significant difficulty is that Dirichlet spectra are not generally continuous under pmGH convergence, complicating compactness and approximation arguments.

The authors note that this stability question is open even for smooth Riemannian manifolds; in that smooth setting, such stability is expected to follow from Cheeger–Colding’s estimates, suggesting a possible approach in the non-smooth synthetic framework as well.

References

Does stability hold in Theorem \ref{th: Cheng dir-Rigid}?\ More precisely, is it true that for every $\epsilon>0$ there exists $\delta>0$ such that if $(X,d,)$ is an $RCD{\star}(K,N)$ space, containing a ball $B(x_0, r)$, with $r\in (r_0-\delta, r_0+\delta)$, such that $\lambda_1D(B(x_0,r))\geq \lambda_{K,N,r_0} -\delta$, then $(X,d,, x_0)$ is necessarily $\epsilon$-close in pmGH sense to one among the rigid configurations 1.\;2.\;or 3.\;in Theorem \ref{th: Cheng dir-Rigid}? A challenge in establishing such a result is that, in general, the Dirichlet spectrum is not continuous under pmGH convergence (see for necessary and sufficient conditions). The stability in Cheng's inequality seems to be an interesting open question already in the framework of smooth Riemannian manifolds; however, in the latter setting, the stability should follow from Cheeger-Colding's estimates .

Cheng's eigenvalue comparison on metric measure spaces and applications (2507.23671 - Luca et al., 31 Jul 2025) in Open Problem [Stability], following Theorem Cheng dir-Rigid, Section 3