Pith's one-line read
Uniformly locally finite metric spaces with isomorphic Roe algebras are coarsely equivalent; the same rigidity holds for the uniform and controlled-propagation algebras, and outer automorphisms are exactly coarse equivalences up to…
desk verdict
Theorem A is the real result and its proof is solid; Theorem B has a small, repairable gap in Claim 4.6, so the paper needs a fix before publication.
read the letter →
A machine-rendered reading of the paper's core claim, the
machinery that carries it, and where it could break.
The reading
Two uniformly locally finite metric spaces with isomorphic Roe algebras must be coarsely equivalent: that is the paper's Theorem A, solving the C*-rigidity problem for bounded geometry metric spaces. The same conclusion holds for the uniform Roe algebra and the controlled-propagation algebra. Since every bounded geometry space is coarsely equivalent to a uniformly locally finite one, the theorem covers all spaces of bounded geometry, so the coarse type is a complete invariant of these operator algebras. Theorem B goes further and shows the outer automorphism group of the Roe algebra is canonically isomorphic to the group of coarse equivalences up to closeness, making the correspondence functorial.
What carries the argument
The machinery is Proposition 3.2, a concentration inequality for unitaries $U:\ell^2(X;H)\to\ell^2(Y;H)$. It says: if for a fixed ball $B(y;R)$ every basis vector $U\chi_x$ has norm at most $\delta$ on that ball, then some subset $A\subseteq X$ has $\|(1-\chi_{B(y;R)})U\chi_A U^*\chi_y\|\ge \tfrac12(1-\delta^2)^{1/2}$, a corner of large norm that jumps over a distance $>R$. The proof averages over Rademacher signs and uses the Hilbert-space identity $\mathbb{E}\|\sum\varepsilon_n v_n\|^2=\sum\|v_n\|^2$. Combined with weak approximate control of implementing unitaries from Theorems 2.14 and 2.15, the inequality produces the controlled maps that witness coarse equivalence.
What would settle it
A concrete refutation would be a pair of uniformly locally finite metric spaces $X$ and $Y$ (for instance, graphs with different growth or expander behaviour) admitting a *-isomorphism $C^*_{\mathrm{Roe}}(X) \cong C^*_{\mathrm{Roe}}(Y)$ while no coarse equivalence between them exists; alternatively, an explicit outer automorphism of a Roe algebra not induced by a coarse self-equivalence would disprove Theorem B.
The central discovery is that the coarse geometry of a uniformly locally finite space can be read off from the fine matrix-coefficient structure of any unitary implementing an isomorphism of its Roe algebra. The authors show that every such unitary is weakly approximately controlled, and then prove a concentration inequality that forces the unitary to have uniformly large matrix entries along some map $g:Y\to X$; symmetry gives a map $f:X\to Y$, and a lemma on weakly controlled operators upgrades the large entries into the statement that $f$ and $g$ are controlled and mutually close. Theorem A states this for $C^*_{\mathrm{Roe}}$, $C^*_u$, and $C^*_{cp}$. Theorem B refines the construction: any implementing unitary is a norm limit of operators coarsely supported on the constructed coarse equivalence, which yields the canonical isomorphism $\tau:\mathrm{CE}(X)\to \mathrm{Out}(C^*_{\mathrm{Roe}}(X))$.
Load-bearing premise
The load-bearing hypothesis is uniform local finiteness of the metric spaces, with the bounded-geometry extension relying on the asserted reduction of every bounded geometry space to that case by coarse equivalence.
Editorial extensions
If this is right
The C*-rigidity problem for bounded geometry metric spaces is settled: the Roe algebra, the uniform Roe algebra, and the controlled-propagation algebra each determine the coarse type of the space.
The outer automorphism group of the Roe algebra is a complete invariant of the coarse equivalence class, canonically isomorphic to the group of coarse self-equivalences up to closeness.
Every automorphism of the Roe algebra is outer-equivalent to an automorphism implemented by a unitary covering a coarse equivalence, so non-inner automorphisms are exactly the nontrivial coarse self-equivalences.
The automorphism groups and outer automorphism groups of $C^*_{\mathrm{Roe}}(X;H)$ and $C^*_{cp}(X;H)$ coincide, so rigidity results transfer between the two algebras.
The rigidity statements hold for arbitrary coefficient Hilbert spaces $H$, not only the standard separable one.
Reading between the lines
Editorial extensions of the paper, not claims the author makes directly.
Inference: the concentration inequality uses only Hilbert-space coefficients and the metric, so the same mechanism should prove rigidity for other Roe-like algebras, such as quasi-local operator algebras, whose isomorphisms are spatially implemented and weakly approximately controlled.
Inference: Theorem 4.5 explicitly uses infinite-dimensionality of $H$, so a finite-dimensional analogue of the norm-limit support statement would need new ideas; whether finite-dimensional coefficients give the same canonical description of outer automorphisms is a natural test.
Inference: the inequality is a quantitative statement about how unitaries localize vectors, and it can be read as an uncertainty-type bound for coarse-like operators, potentially useful outside C*-rigidity.