UCT problem for nuclear C*-algebras

A two-by-two complex matrix can keep the first coordinate and discard the second: use diagonal entries one and zero, and applying it twice changes nothing further. Can smaller records of what two algebras of such operators contain recover information about maps between them, together with a specified correction term? C*-algebras generalize such arithmetic to norm-closed algebras of operators. The question concerns separable nuclear algebras: they have a countable dense set, and any finite set of their elements can be approximated through finite matrix algebras by completely positive maps. Their K-groups record stable invariant data. The universal coefficient theorem, or UCT, asks for a precise relation from those groups to the two-algebra information called KK-theory, with every separable second algebra.[1] [3] [4]

Knowing how two algebras of operators look individually is not the same as knowing the ways they relate. Can smaller records of their elements determine that two-algebra information, together with a specified extension term? K-groups describe stable invariant data of individual algebras, while KK-theory records information relating two algebras. The finite matrix projection $p=\operatorname{diag}(1,0)$ has $p^2=p$, but even an exact finite matrix factorization does not illustrate the full two-algebra theorem. For each separable nuclear complex $A$ and every separable complex $B$, the natural map $\gamma:KK_*(A,B)\to\operatorname{Hom}(K_*(A),K_*(B))$ must be onto. The UCT identifies $\ker\gamma$ with $\operatorname{Ext}(K_*(A),K_{*+1}(B))$ through the specified natural exact sequence, with graded indices interpreted modulo two. Equivalently, $A$ must be KK-equivalent to a commutative C*-algebra. KK-equivalence is weaker than an algebra isomorphism; nuclearity is approximation by completely positive maps through matrices, not a finite-dimensionality assumption.[1] [3] [4]

Commutative operator algebras satisfy the UCT, and many more examples follow through established closure properties. Separable nuclear algebras with a Cartan subalgebra also satisfy it. What is missing is a reason why every separable nuclear algebra belongs to the UCT class. The 2026 homotopy-rigidity result drops UCT from a specified theorem; removing it as that theorem’s assumption is not proving it for all nuclear algebras.[1] [2] [3]

The Cartan-subalgebra criterion establishes UCT in a substantial nuclear class and relates the remaining problem to groupoid descriptions. The general implication from separability and nuclearity remains missing. The recent homotopy-rigidity theorem instead retains KK and other hypotheses in its specified Z₀-stable setting while avoiding a UCT assumption. Neither its lack of that assumption nor a reduction to a special family proves the general UCT exact sequence.[1] [2]

Classifying algebras by smaller lists of invariants is more informative than comparing every possible operator separately. Important classification theorems use UCT to pass between ordinary K-group data and the richer information relating two algebras. If nuclearity always ensured UCT, that extra assumption could be removed in the eligible classes. Their other requirements would remain: the claim is not that K-groups alone distinguish every nuclear algebra.[1] [3]

UCT controls how K-group data recover bivariant KK information; it is not a claim that K-groups alone classify all nuclear algebras. The homomorphism part describes what KK classes do to ordinary K-groups, while the extension term records information that can be invisible to those maps. This distinction matters in classification: eligible theorems can require a KK-equivalence or use UCT to obtain one from simpler invariant data. Understanding that passage is a concrete reason to study UCT, while simplicity, regularity, trace and other hypotheses of each classification theorem remain separate.[1] [2] [3]