A Criterion for Phantomness of dg-categories

1. Problem Statement and Context

The paper addresses a fundamental question in noncommutative algebraic geometry: Does the vanishing of additive invariants characterize "phantomness" for smooth proper dg-categories admitting geometric realizations?

• Phantom Categories: A dg-category $T$ is a quasi-phantom if its Hochschild homology vanishes ($HH_*(T/k)=0$) and its Grothendieck group $K_0(T)$ is finite. It is a universal phantom if $K_0(T \otimes Y) = 0$ for all smooth proper varieties $Y$.
• Motivic Quasi-Phantom: The author introduces a rational version: $T$ is a motivic quasi-phantom if $K_0(T \otimes Y)_{\mathbb{Q}} = 0$ for all smooth proper $Y$. This is equivalent to the vanishing of the rational noncommutative motive of $T$.
• The Core Question: If an additive invariant (such as $K$-theory, Hochschild homology, or topological $K$-theory) vanishes rationally, does the category become a motivic quasi-phantom? This question was partially raised by Sosna [Sos20].

2. Methodology

The author employs motivic homotopy theory and the theory of noncommutative motives to bridge the gap between algebraic invariants and the structural properties of dg-categories.

A. Construction of the Commutative Motive $M(T)$

For a smooth proper dg-category $T$ admitting a geometric realization $T \hookrightarrow \text{perf}^{\text{dg}}(X)$, the author constructs a specific motive $M(T) \in DM(k, \mathbb{Q})$ (the category of mixed motives over $k$).

• Functorial Construction: The construction utilizes the embedding of the category of noncommutative motives $SH^{\text{nc}}(k)_{\mathbb{Q}}$ into the category of modules over the rational algebraic K-theory spectrum $KGL_{\mathbb{Q}}$.
• Definition: $M(T)$ is defined via the composition of functors:
  $$M(T) = H^{-1} \circ I \circ M_{KGL_{\mathbb{Q}}} (U(T))$$
  where $U(T)$ is the noncommutative motive of $T$, $M_{KGL_{\mathbb{Q}}}$ is the right adjoint to the embedding of commutative motives into noncommutative motives, and $I$ is the forgetful functor.

B. Realization Theorems

The central technical achievement is proving that the l-adic realization (and analogously Betti and de Rham realizations) of this constructed motive $M(T)$ recovers specific algebraic invariants of $T$.

• l-adic Realization: The author proves that the cohomology of the l-adic realization of $M(T)$ is isomorphic to the homotopy groups of the $K(1, l)$-local algebraic K-theory of $T$:
  $$H^i(R_l(M(T))) \cong \pi_i LK_{(1,l)} K(T)$$
• Proof Strategy:
  1. Construct two symmetric monoidal functors, $F_1$ and $F_2$, from the category of $KGL_{\mathbb{Q}}$-modules to the derived category of $\mathbb{Q}_l[\beta^{\pm}]$-modules.
  2. $F_1$ is derived from the l-adic realization of motives.
  3. $F_2$ is derived from the $K(1, l)$-local K-theory spectrum.
  4. Using Thomason's descent spectral sequence and the fact that both functors preserve colimits and are determined by their action on smooth varieties, the author proves $F_1 \simeq F_2$.

C. The Role of Kimura-Finiteness

The proofs rely heavily on the assumption that the Chow motive of the ambient variety $X$ is Kimura-finite.

• Kimura-Finiteness: A property implying that the motive decomposes into finite-dimensional pieces, allowing for the application of nilpotence arguments.
• Application: If a morphism in the category of motives is numerically trivial (detected by cohomology/realization), Kimura-finiteness implies it is nilpotent. This allows the author to deduce that if an invariant vanishes, the underlying motive must vanish.

3. Key Contributions and Results

Main Theorem (Theorem 1.2 & 3.3)

Let $T$ be a non-zero smooth proper dg-category admitting a geometric realization in a variety $X$ with a Kimura-finite Chow motive. The following implications hold:

1. K-theory Criterion: If $LK_{(1,l)} K(T)_{\mathbb{Q}} = 0$ (for a prime $l$ invertible in $k$), then $T$ is a motivic quasi-phantom.
2. Hochschild Homology Criterion: If $\text{char}(k) = 0$ and $HH_*(T/k) = 0$, then $T$ is a motivic quasi-phantom.
3. Topological K-theory Criterion: If $k = \mathbb{C}$ and $K_{\text{top}}(T)_{\mathbb{Q}} = 0$, then $T$ is a motivic quasi-phantom.

Significance: This establishes a converse to the standard implication (vanishing motive $\implies$ vanishing invariants). It provides a concrete criterion for detecting phantomness using computable invariants.

Deformation Invariance (Corollary 1.3 & 4.2)

The paper establishes that the property of admitting a phantom category is deformation-invariant in smooth proper families, provided the fibers have Kimura-finite motives.

• Result: If a fiber $X_0$ in a family $f: X \to S$ admits a motivic quasi-phantom $T_0$, then there exists an étale neighborhood of $0$ where every fiber admits a motivic quasi-phantom.
• Condition: This holds if $LK_{(1,l)} K(T_0) \neq 0$ (ensuring the phantom is "detectable" via K-theory). Under the Belmans–Okawa–Ricolfi conjecture, this non-vanishing condition is unnecessary.

4. Significance and Impact

1. Partial Answer to Sosna's Question: The paper provides a rigorous "yes" to whether the vanishing of specific additive invariants implies the vanishing of the noncommutative motive, under the Kimura-finiteness assumption.
2. Unification of Realizations: It unifies the study of l-adic, Betti, and de Rham realizations in the context of noncommutative motives, showing they all recover the corresponding K-theoretic or homological invariants of the dg-category.
3. New Tool for Classification: The constructed motive $M(T)$ serves as a powerful tool for classifying dg-categories. If $M(T) = 0$, the category is "phantom" in the strongest rational sense.
4. Generalization of Previous Work: The deformation invariance result generalizes the work of Böhmning et al. [BGvBKS15] on Godeaux surfaces, extending the strategy from specific numerical examples to a broad class of families via the Kimura-finiteness assumption.

5. Limitations and Assumptions

• Kimura-Finiteness: The main results require the Chow motive of the ambient variety $X$ to be Kimura-finite. While conjectured to hold for all smooth projective varieties, it is currently only proven for specific cases (e.g., surfaces with vanishing geometric genus).
• Geometric Realization: The results apply specifically to dg-categories that are admissible subcategories of $\text{perf}^{\text{dg}}(X)$ for some smooth proper $X$.
• Field Characteristics: Different results apply depending on the characteristic of the base field (e.g., Hochschild homology vanishing requires characteristic 0).

In summary, Matsumoto's paper provides a deep structural link between the vanishing of classical invariants (K-theory, Hochschild homology) and the vanishing of noncommutative motives, offering a robust criterion for identifying phantom categories in algebraic geometry.