A remark on the invariance of $K$-theory under duality

This paper clarifies that the invariance of $K$-theory under duality is a purely formal property, contrasting it with the non-formal nature of the general statement for arbitrary localizing invariants, and provides a counterexample to the claim that the universal localizing invariant is invariant under taking opposite categories.

The Gist

Imagine you are an architect designing a building. In mathematics, specifically in a field called K-theory, we don't just look at the building itself; we look at its "blueprint" or its "essence." This essence is what mathematicians call an invariant. It's a number or a shape that stays the same no matter how you rearrange the furniture inside, as long as the structure remains intact.

This paper by Georg Lehner is a short, sharp note about a specific question: If you flip a building inside out (take its "opposite" or "dual"), does its essence change?

Here is the breakdown of the paper using simple analogies:

1. The Two Sides of the Coin (The "Good" News)

The author starts by looking at two specific types of mathematical structures (like two specific types of buildings).

• Example A: A space that is "locally compact" (think of a city with well-defined neighborhoods).
• Example B: A "continuous poset" (think of a hierarchy or a flowchart where things are connected smoothly).

In these two specific cases, the author confirms a happy truth: If you flip the building inside out, the essence (K-theory) stays exactly the same.

• The Analogy: Imagine a perfectly symmetrical snowflake. If you flip it over, it looks identical. The "K-theory" is like the pattern of the snowflake; it doesn't care which way is up.

The author explains that for K-theory (the most famous type of invariant), this symmetry is almost automatic. It's like saying, "If you have a list of all the rooms in a house, flipping the house upside down doesn't change the list of rooms." This is a formal, logical certainty.

2. The Trap (The "Bad" News)

However, the author warns us: Don't get too comfortable.

While K-theory is symmetrical, there are other, more complex "essences" (called universal localizing invariants) that are not symmetrical.

• The Analogy: Imagine a house that looks the same from the outside, but the inside is a maze. If you flip the house inside out, the maze might become impossible to navigate. The "essence" of the maze has changed, even though the K-theory (the list of rooms) hasn't.

The paper proves that for these more complex invariants, flipping the structure (taking the opposite category) actually changes the result.

3. The Smoking Gun (The Counterexample)

To prove this, the author uses a famous concept from algebra called the Brauer Group. Think of this as a "club" for special types of number systems (algebras).

• There is a rule in this club: Every member has a "partner" (its opposite). Usually, a member and its partner are considered the same.
• The Twist: The author finds a specific "member" (a division algebra over the rational numbers, $\mathbb{Q}$) that is not its own partner.
• The Metaphor: Imagine a glove. A left glove and a right glove are "opposites." In most cases, if you flip a left glove, it becomes a right glove, and they are distinct. But in K-theory, we pretend they are the same. In this specific "universal" case, the author shows that the left glove and the right glove are actually different in a way that matters.

By using a theorem about "Hasse invariants" (which are like local weather reports for different parts of a number system), the author shows that for a specific algebra $A$, the "essence" of $A$ is mathematically distinct from the "essence" of its opposite $A^{op}$.

4. The Big Takeaway

The paper concludes with a challenge for other mathematicians:

• We know K-theory is always symmetrical (flipping it doesn't matter).
• We know the "Universal" invariant is not always symmetrical.
• The Open Question: What is the rule? When can we flip a structure and expect the result to be the same? The author gives us two examples where it works, but we need a general rule to know when it works for everything else.

Summary in One Sentence

"While the most famous mathematical 'essence' (K-theory) is perfectly symmetrical and doesn't care if you flip a structure inside out, more complex essences do care, and this paper proves it using a clever trick with number systems."

Technical Summary

Here is a detailed technical summary of the paper "A Remark on the Invariance of K-Theory under Duality" by Georg Lehner.

1. Problem Statement

The paper addresses a fundamental question in the theory of localizing invariants for stable $\infty$-categories: Is a localizing invariant $F$ invariant under the operation of taking the dual category?

Specifically, for a dualizable $\infty$-category $\mathcal{C}$, let $\mathcal{C}^\vee = \text{Fun}^L(\mathcal{C}, \text{Sp})$ denote its dual (the category of continuous functors to spectra). The author investigates whether the equivalence $F(\mathcal{C}) \simeq F(\mathcal{C}^\vee)$ holds generally.

• Context: While it is known that $\mathcal{C} \not\simeq \mathcal{C}^\vee$ in general, specific examples (such as sheaves on stably locally compact spaces or continuous posets) suggest invariance might hold for certain classes of categories.
• The Conflict: There is a tension between the behavior of Algebraic K-theory (which appears to be invariant under duality) and the behavior of the universal localizing invariant $U$ (which the author argues is not invariant under duality).

2. Methodology

The paper employs a combination of formal categorical arguments, explicit calculations in K-theory, and counterexamples derived from the theory of Brauer groups.

• Formal Categorical Arguments: The author utilizes the universal property of K-theory as a localizing invariant. By analyzing the involution $(-)^{op}$ on the category of perfect $\infty$-categories ($\text{Cat}^{perf}$), the author demonstrates that K-theory's invariance is a "formal" consequence of its definition.
• Extension to Dualizable Categories: The argument is extended from $\text{Cat}^{perf}$ to the larger category of dualizable stable $\infty$-categories ($\text{Pr}^L_{\text{dual}}$) using the Ind-completion functor and the relationship between a category and its dual.
• Counterexample Construction: To disprove the general claim for arbitrary localizing invariants, the author constructs a specific counterexample using:
  • Brauer Groups: Utilizing the structure of the Brauer group $\text{Br}(\mathbb{Q})$ of the rational numbers.
  • Albert–Brauer–Hasse–Noether Theorem: Using local-global principles to construct a central division algebra $A$ over $\mathbb{Q}$ of order 3.
  • Motivic Homotopy Theory: Analyzing the mapping spaces in the category of non-commutative motives ($\text{Mot}$) to show that the universal invariant distinguishes between a category and its opposite.

3. Key Contributions and Results

A. Formal Invariance of K-Theory (Theorems 3 & 4)

The author proves that Algebraic K-theory is invariant under duality.

• Theorem 3: There is a natural equivalence of functors $K \simeq K \circ (-)^{op}$ on $\text{Cat}^{perf}$.
  • Reasoning: The groupoid core functor $(-)^\simeq$ is invariant under taking opposites. Since K-theory is the universal localizing invariant receiving a map from the groupoid core, it inherits this invariance.
• Corollary 4: This extends to continuous K-theory ($K_{\text{cont}}$) on dualizable categories: $K_{\text{cont}} \simeq K_{\text{cont}} \circ (-)^\vee$.
• Implication: For any dualizable category $\mathcal{C}$, $K(\mathcal{C}) \simeq K(\mathcal{C}^\vee)$. This explains why Examples 1 (sheaves on spaces) and 2 (sheaves on posets) in the literature exhibit this invariance; they are instances of K-theory, not general localizing invariants.

B. Failure of Invariance for the Universal Invariant (Theorem 6)

The author refutes the claim that the universal localizing invariant $U$ is invariant under duality.

• Theorem 6: There exists a stable, idempotent complete $\infty$-category $\mathcal{C}$ such that $U(\mathcal{C}) \not\simeq U(\mathcal{C}^{op})$.
• Proposition 9 (The Core Counterexample): Let $A$ be a finite-dimensional central division algebra over $\mathbb{Q}$ that is not of order 2 in the Brauer group $\text{Br}(\mathbb{Q})$. Then $U_{\mathbb{Q}}(A) \not\simeq U_{\mathbb{Q}}(A^{op})$ in the category of motives $\text{Mot}_{\mathbb{Q}}$.
  • Proof Sketch: The proof relies on the structure of the Brauer group. If $A$ has order $>2$, then $A \otimes A$ is not equivalent to a matrix algebra over $\mathbb{Q}$ (it involves a division algebra $D$ of dimension $d > 1$).
  • The composition of morphisms in the motive category corresponds to a map on $K_0$ groups:
    $$K_0(A \otimes A) \times K_0(A^{op} \otimes A^{op}) \to K_0(A^{op} \otimes A) \cong K_0(\mathbb{Q}) \cong \mathbb{Z}$$
  • This map is identified with multiplication by $d^2$ (where $d = \dim(D)$). Since $d > 1$, the image of this map cannot contain $1 \in \mathbb{Z}$. Therefore, the identity morphism on$U(A)$cannot be factored through$U(A^{op})$, proving they are not equivalent.
• Concrete Example: Using the Albert–Brauer–Hasse–Noether theorem, the author constructs a specific element $\alpha \in \text{Br}(\mathbb{Q})$ of order 3 (defined by local invariants at primes 2 and 3). The corresponding division algebra $A$ serves as the counterexample.

4. Significance

1. Clarification of K-Theory vs. General Invariants: The paper resolves a potential confusion in the literature. It clarifies that the invariance observed in specific geometric and topological examples (like sheaves on spaces) is a special property of K-theory (and functors factoring through it), not a universal property of all localizing invariants.
2. Distinction between $K$ and $U$: It establishes a sharp distinction between the universal localizing invariant $U$ and K-theory. While $K$ "forgets" enough information to be invariant under duality, $U$ retains enough structural data (specifically related to the Brauer group) to distinguish a category from its dual.
3. Connection to Brauer Groups: The work highlights a deep connection between the theory of non-commutative motives and the classical Brauer group of fields. It shows that the map from the Brauer group to the Picard group of motives is injective, and that the operation of taking the opposite category corresponds to taking the inverse in the Brauer group.
4. Open Question: The paper concludes by posing a significant open question (Question 7): What is the precise criterion for a dualizable $\infty$-category $\mathcal{C}$ such that $U(\mathcal{C}) \simeq U(\mathcal{C}^\vee)$? The author notes that the examples where this holds (sheaves on spaces/posets) satisfy it, but the general condition remains unknown.

Summary Conclusion

Georg Lehner's remark demonstrates that while Algebraic K-theory is formally invariant under the duality of categories (due to its universality and relation to groupoid cores), the universal localizing invariant is not invariant under duality. This is proven by constructing a counterexample using central division algebras of order greater than 2, where the universal invariant distinguishes between an algebra and its opposite. This result underscores that K-theory is a "coarser" invariant that smooths over certain non-commutative distinctions preserved by the universal invariant.