Unital Classification Theorem

• The unital classification theorem is a framework that uses explicit algebraic invariants and geometric data to distinguish unital operator and algebraic structures.
• Key invariants include ordered K–theory, trace information, and filtered data that establish complete isomorphism or Morita equivalence criteria.
• Geometric realizations via specific graph moves offer practical methods to verify equivalences, mirroring techniques seen in knot theory.

A unital classification theorem identifies a complete invariant that determines equivalence (usually *-isomorphism or Morita equivalence) between algebraic or operator-theoretic objects when these carry a distinguished unit, unit class, or point, or arise as "unital" objects in a broader category. In the context of $C^*$-algebras, Leavitt path algebras, Cuntz–Krieger algebras, and incidence geometries, the unital classification theorem asserts that unital structures are classified by explicit algebraic invariants (often $K$–theoretic, tracial, or filtered invariants) together with geometric data, and often that every invariant isomorphism arises from a concrete model transformation (e.g., graph moves).

1. Formulation and Abstract Setting

The paradigm of the unital classification theorem originated in operator algebras but is now foundational in both algebraic and geometric classifications. For unital simple separable nuclear $C^*$–algebras satisfying the Universal Coefficient Theorem (UCT) and Jiang–Su stability, the theorem is stated: $A \cong B \iff \Ell(A) \cong \Ell(B)$ where $\Ell(A)$ is the ordered $K$–theory and trace invariant: $\Ell(A) = \left(K_0(A),\, K_0(A)^+,\,[1_A],\,K_1(A),\,T(A),\,\rho_A\right)$ with $K_0(A)$ the Grothendieck group of projections, $K_0(A)^+$ the positive cone, $[1_A]$ the distinguished order unit, $K_1(A)$ the unitary class group, $T(A)$ the tracial simplex, and $\rho_A$ the natural pairing between $K_0$ and tracial states (White, 2023, Bouwen et al., 2024, Carrión et al., 2023, Elliott et al., 2015, Lin et al., 2008).

This formulation unifies both purely infinite and stably finite cases (Bouwen et al., 2024, White, 2023), as detailed below, and is echoed in related settings (e.g., Leavitt path algebras (Ruiz et al., 2012), graph $C^*$-algebras (Eilers et al., 2016, Eilers et al., 2015, Arklint et al., 2019), and finite geometry (Grundhöfer et al., 2022)).

2. Key Invariants and Structural Ingredients

Table: Classification Invariants Across Major Classes

Object/Class	Invariant(s) Used	Notes
Simple nuclear $C^*$-algebras	$\Ell(A)$ as above	Complete when unital, UCT, Jiang–Su stable (White, 2023, Bouwen et al., 2024)
Purely infinite Cuntz–Krieger	Filtered $K$–theory + unit class	Reduced filtered $K$–theory, unital class (Carlsen et al., 2016, Eilers et al., 2016)
Graph $C^*$-algebras (real rank 0)	Ordered, filtered $K$–theory	Realized via graph moves (Eilers et al., 2015, Arklint et al., 2019)
Leavitt path algebras (infinite)	$K_0^\text{alg}$ and $K_1^\text{alg}$, singular vertices	Moves on graphs, field property (Ruiz et al., 2012)
Hermitian unitals	Combinatorial symmetry criteria	Fixed translations, involutions (Grundhöfer et al., 2022)

Crucially, invariants include both algebraic ($K$–theory, traces) and combinatorial (ideal lattice, filtered data, unit class) (Carrión et al., 2023, Carlsen et al., 2016). In certain cases, filtered data incorporates six-term exact sequences, action of moves, or pointed structure to encode the unital aspect.

3. Geometric and Move-Based Realization

A distinctive feature in the classification of graph $C^*$-algebras and Leavitt path algebras is the geometric realization of equivalence by explicit sequences of graph moves: outsplitting (O), insplitting (I), reduction (R), regular source removal (S), Cuntz splice (C), and in some classification schemes, the Pulelehua move (P) or variants (Eilers et al., 2016, Arklint et al., 2019, Eilers et al., 2015, Ruiz et al., 2012).

Moves correspond to elementary row/column operations on adjacency matrices, with determinant conditions handled by the Cuntz splice or refined expansion moves. Every Morita equivalence or isomorphism can thus be realized via a finite chain of moves, mirroring the combinatorial philosophy of Reidemeister moves in knot theory (Eilers et al., 2015, Arklint et al., 2019, Eilers et al., 2016).

4. Proof Strategies and Abstract Arguments

Modern approaches unify classification by abstract absorption and extension theorems, bypassing explicit finite-dimensional approximation structure. The Carrión–Gabe–Schafhauser–Tikuisis–White framework (Bouwen et al., 2024, White, 2023) proceeds via:

• Classification of maps via ultraproducts and $K$–theory: For stably finite algebras, the "trace-kernel extension" allows one to lift tracial data to *-homomorphisms into sequence algebras, classified via $KK$-theory and abstract absorption theorems (Elliott–Kucerovsky).
• Purely infinite side via state-kernel extensions: Corresponds to classification of embeddings into hereditary/corona algebras, synergizing with Kirchberg–Phillips machinery.
• Approximate intertwining: Once embeddings with matching invariants are constructed, the two-sided approximate intertwining argument yields genuine isomorphism (Carrión et al., 2023, Bouwen et al., 2024, Elliott et al., 2015).

In the purely infinite case, ideal-lattice geometry replaces full $KK$–theory, and selection-theoretic arguments yield existence of required embeddings purely from combinatorial invariants (Gabe, 2017).

5. Extensions and Range Theorems

A refined version identifies the range of the invariant, guaranteeing existence and uniqueness for arbitrary data tuples. For example, Lin–Niu’s range theorem for simple AH/ASH/ATD algebras establishes that for any ordered abelian group $G_0$, countable abelian group $G_1$, tracial simplex $S$, and affine pairing $r$, there exists a unique algebra realizing $(G_0,G_1,S,r)$ (Lin et al., 2008).

Similarly, for Leavitt path algebras, invariants like $K_0^\text{alg}$, $K_1^\text{alg}$, and singular vertex count (or their graded analogues under restriction on the field) classify up to Morita equivalence (Ruiz et al., 2012, Brix et al., 2024).

6. Applications, Impact, and Examples

Classification is now comprehensive for large classes:

• All simple separable nuclear $C^*$-algebras known "in nature" (AF, AH, ASH, AI, AT, irrational rotation algebras, groupoid $C^*$-algebras) fall under the unital classification theorem (White, 2023, Bouwen et al., 2024).
• For graph $C^*$-algebras, all Morita equivalence and isomorphism problems are decidable via algorithmic analysis of filtered $K$–theory (Eilers et al., 2016).
• Abrams–Tomforde conjecture (Morita equivalence for Leavitt path and graph $C^*$-algebras) holds in the unital setting (Eilers et al., 2016).
• Rigidity phenomena for Leavitt path algebras (full graded $K$–theory functor, nonexistence of unital graded maps in splice contexts) are confirmed (Brix et al., 2024).

7. Controversies and Open Questions

A remaining technical issue is the necessity of the "sign of determinant" condition in the classification of purely infinite simple Leavitt path algebras and finite graph algebras; in certain cases (finite graphs with purely infinite simple Leavitt path algebras), the invariant may need sign information to fully detect Morita equivalence, and its necessity is unresolved (Ruiz et al., 2012).

In summary, the unital classification theorem provides a uniform, explicit, and often geometric invariant for classifying unital structures across operator algebraic and algebraic contexts, backed by abstract absorption and ultraproduct techniques, combinatorial and K-theoretic invariants, and geometric realizations via moves.