🔢 mathematics

Generalized K-theoretic invariants and wall-crossing via non-abelian localization

This paper introduces generalized $K$ -theoretic invariants and a new $K$ -Hall algebra structure to prove wall-crossing formulas via non-abelian localization, thereby extending Joyce and Liu's cohomological results to non-standard hearts of $D^b(X)$ where framing functors are not known to exist.

Original authors: Ivan Karpov, Miguel Moreira

Published 2026-04-08

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Ivan Karpov, Miguel Moreira

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Counting Shapes in a Shifting Landscape

Imagine you are an explorer trying to count the number of unique islands in a vast ocean. These islands represent mathematical objects (like bundles of strings or geometric shapes). However, the ocean is tricky: the water level (which mathematicians call a stability condition) keeps rising and falling.

When the water is low: Some islands are separate. You can count them easily.
When the water rises: Two islands might merge into one giant landmass.
When the water falls: That giant landmass might split back into two.

The central problem in this field is: How do we count these islands accurately when the water level changes? If we just count them at one moment, our count is wrong for the next moment. We need a formula that tells us exactly how the count changes as the water rises or falls. This is called Wall-Crossing.

The Problem: The "Strict" vs. "Loose" Rules

In the past, mathematicians had two ways to handle this:

The "Motivic" Way (The Motive): This worked well for simple counting. It was like counting islands by looking at their shadows. It was flexible but didn't capture deep structural details.
The "Cohomological/K-theoretic" Way (The Deep Dive): This tried to count the islands by analyzing their internal structure (like counting the atoms inside the rock). This was more powerful but much harder.

The Catch: To use the "Deep Dive" method, mathematicians previously needed a special tool called a Framing Functor. Think of this as a special pair of glasses that forces every island to have a unique, rigid shape so it never merges with another.

The Problem: These "glasses" only existed for very simple, standard types of islands (like vector bundles on a curve).
The Gap: For more complex, modern islands (called Bridgeland stable objects, which appear in string theory and advanced geometry), nobody knew if these "glasses" even existed. Without them, the deep counting formulas were just guesses.

The Solution: A New Way to Count Without Glasses

Karpov and Moreira (the authors) have invented a new method that doesn't need the special glasses. They developed a way to count the islands directly, even when they are messy, merging, and splitting.

Here is how they did it, step-by-step:

1. The "K-Hall Algebra": A New Language for Mixing

Imagine you have a bag of Lego bricks.

Old Way: You could only count the bricks if they were already snapped together in a specific way.
New Way (K-Hall Algebra): The authors created a new "grammar" or language (an algebra) that allows you to describe how to snap bricks together and how to break them apart.
They defined a special multiplication rule. If you have a pile of "Island A" and a pile of "Island B," this rule tells you how to mathematically combine them into a "Merged Island."

2. The "Logarithm" Trick: From Messy to Clean

In their new language, they first define a raw count called a $\delta$ -invariant. This is like counting every single island, including the messy ones that are half-merged.

The Issue: These raw counts are "noisy." They have mathematical "poles" (infinite spikes) that make them hard to use for deep formulas.
The Fix: The authors take the "logarithm" of these raw counts. In math, taking a logarithm is like filtering out the noise.
This gives them a new, clean count called the $\epsilon$ -invariant. These are the "true" invariants that behave nicely. It's like taking a blurry photo of a crowd and using a filter to make every person distinct and countable.

3. The Magic Tool: Non-Abelian Localization

How do they prove that their new counting method works when the water level changes? They use a powerful theorem called Non-Abelian Localization (developed by Halpern-Leistner).

The Analogy: Imagine you want to know the total weight of a complex machine. Instead of weighing the whole thing, you realize the machine is made of a stable core and some vibrating parts.
The Theorem: It says, "The total weight of the whole machine is equal to the weight of the stable core PLUS the weight of the vibrating parts, calculated in a specific way."
In their math, the "stable core" is the islands that don't change, and the "vibrating parts" are the ones that are merging or splitting. This theorem allows them to calculate the change in the count exactly when the water level (stability condition) shifts.

Why This Matters

It Works Where Others Failed: Because they don't need the "special glasses" (framing functors), they can now count complex, modern mathematical objects (like those in string theory) that were previously impossible to count rigorously.
It Unifies the Field: They showed that their new "clean" counts ( $\epsilon$ ) are actually the same as the old "glasses" counts when the glasses do exist. This proves their new method is a generalization, not a contradiction.
It's Shorter and Cleaner: Their proof is more direct. Instead of building a complex scaffold (the framing functor) to reach the answer, they built a bridge (the K-Hall algebra) that goes straight there.

Summary Metaphor

Imagine you are trying to track a flock of birds.

Old Method: You needed a net (Framing Functor) to catch the birds so you could count them. But for some rare, fast birds, the net didn't exist.
New Method: Karpov and Moreira invented a new type of radar (K-Hall Algebra) that can track the birds even when they are flying in a chaotic swarm. They developed a formula (Wall-Crossing) that predicts exactly how the flock splits or merges as the wind changes, without ever needing to catch a single bird.

This breakthrough opens the door to solving long-standing problems in geometry and physics that were stuck waiting for a way to count these "un-catchable" mathematical objects.

1. Problem Statement

The paper addresses the problem of defining and computing generalized K-theoretic invariants for moduli spaces of objects in abelian categories (such as sheaves on varieties or representations of quivers) under varying stability conditions.

Context: In enumerative geometry, invariants (motivic, cohomological, or K-theoretic) are attached to moduli spaces $M^\mu_\alpha$ of $\mu$ -semistable objects. A central challenge is understanding how these invariants change as the stability condition $\mu$ crosses a "wall" in the stability manifold (wall-crossing).
The Gap:
- Motivic Setting: Well-understood via the work of Joyce, Kontsevich–Soibelman, and others using the motivic Hall algebra.
- Cohomological Setting: Joyce developed a theory using "framing functors" to construct auxiliary moduli spaces of pairs (Joyce–Song pairs) to handle strictly semistable objects.
- K-theoretic Setting: Liu adapted Joyce's cohomological approach to K-theory, also relying on the existence of a framing functor.
- The Limitation: Framing functors are not known to exist for many important stability conditions, particularly weak Bridgeland stability conditions on derived categories (e.g., on Fano 3-folds or for tilt stability). Consequently, wall-crossing formulas in these contexts remained conjectural.

2. Methodology

The authors develop a new, intrinsic approach that does not require framing functors. Their methodology relies on three main pillars:

The K-Hall Algebra:
- They define an associative algebra structure on the K-homology of the rigidified stack of objects, denoted $K(A) = K_*(\mathcal{M}^{rig}_A)_\mathbb{Q}$ .
- This algebra is constructed using a product $\ast$ defined via a push-pull formula involving the direct sum map $\Sigma$ and a specific complex $\Gamma_-$ (related to the inverse Euler class of the normal bundle to the direct sum map).
- This structure serves as the K-theoretic analog of the motivic Hall algebra or the quantum torus.
Generalized Invariants ( $\delta$ and $\epsilon$ ):
- $\delta$ -invariants: Defined directly from the stack of semistable objects $\mathcal{M}^\mu_\alpha$ as pushforwards of the virtual fundamental class. These correspond to the "naive" invariants.
- $\epsilon$ -invariants: Defined by taking a formal logarithm of the $\delta$ -invariants within the K-Hall algebra. This mirrors the motivic "no-pole" theorem, where the logarithm removes singularities associated with strictly semistable objects.
- The relation is given by:
  $\epsilon^\mu_\alpha = \sum_{k \ge 1} \frac{(-1)^{k-1}}{k} \sum_{\alpha_1 + \dots + \alpha_k = \alpha} \delta^\mu_{\alpha_1} \ast \dots \ast \delta^\mu_{\alpha_k}$
Non-Abelian Localization:
- Instead of using auxiliary pair moduli spaces, the authors utilize the Virtual Non-Abelian Localization Theorem of Halpern-Leistner.
- They employ $\Theta$ -stratifications (generalizing Harder–Narasimhan filtrations) on the moduli stack.
- The wall-crossing formula is derived by comparing the Euler characteristics of the semistable locus under two different stability conditions via the stratification of the unstable locus.

3. Key Contributions

Definition of K-Hall Algebra: The construction of an associative algebra structure on the K-homology of the moduli stack of objects, which allows for the definition of generalized invariants without auxiliary data.
Framing-Independent Wall-Crossing: The proof of wall-crossing formulas for $\delta$ and $\epsilon$ invariants that holds for any stability condition satisfying standard technical assumptions (finite type, quasi-smooth, existence of good moduli spaces), including cases where framing functors are unknown (e.g., Bridgeland stability).
Equivalence with Joyce–Liu: A rigorous proof that when a framing functor does exist, the authors' $\epsilon$ -invariants coincide with the previously defined Joyce–Liu invariants. This validates the new approach as a genuine generalization.
Lie Algebra Structure: Identification of a Lie subalgebra within the K-Hall algebra (generated by commutators) where the wall-crossing formulas for $\epsilon$ -invariants take a purely commutator form, analogous to the motivic and cohomological settings.

4. Main Results

Theorem A (K-Hall Algebra): The space $K(A)$ equipped with the product $\ast$ is an associative algebra.
Theorem B (Wall-Crossing Formula): If two stability conditions $\mu$ and $\mu'$ are connected by a path crossing finitely many walls, the invariants satisfy:
$\epsilon^{\mu'}_\alpha = \sum \tilde{U}(\dots) \cdot [\dots [\epsilon^\mu_{\alpha_1}, \epsilon^\mu_{\alpha_2}], \dots, \epsilon^\mu_{\alpha_n}]$
where the sum involves combinatorial coefficients and the Lie bracket is the commutator in $K(A)$ .
Theorem C (Comparison with Pairs): When a framing functor $\Phi$ exists, the authors' $\epsilon$ -invariants are related to the moduli of Joyce–Song pairs via a specific formula involving the pushforward of the virtual fundamental class of the pair moduli space.
Theorem D (Lie Algebra Homomorphism): The natural map from Liu's vertex algebra Lie algebra to the K-Hall algebra is a Lie algebra homomorphism, sending Liu's classes to the authors' $\epsilon$ -classes.
Theorem E (Cohomological Lift): If a framing functor exists, Joyce's cohomological invariants are "homological lifts" of the authors' K-theoretic $\epsilon$ -classes.
Corollary 1.2 (Application to Tilt Stability): The theory establishes wall-crossing formulas for tilt stability on surfaces with nef anticanonical bundles and Fano 3-folds satisfying the BMT inequality, a context where framing functors were previously unavailable.

5. Significance and Impact

Generality: The theory extends the scope of wall-crossing formulas to non-standard hearts of derived categories (like those arising in Bridgeland stability) where the existence of framing functors is an open problem. This resolves long-standing conjectures in enumerative geometry.
Conceptual Clarity: The approach is conceptually closer to the motivic setting (extracting invariants directly from stacks) rather than the cohomological setting (relying on auxiliary pair spaces). This unifies the three settings (motivic, cohomological, K-theoretic) under a single framework.
Simplicity: The proof is shorter and relies on the powerful machinery of non-abelian localization, making the theory more accessible and easier to extend to new contexts (e.g., Calabi–Yau 4-folds or shifted symplectic stacks).
Future Applications: The authors suggest immediate applications to the rationality of Pandharipande–Thomas (PT) invariants for Fano 3-folds and the rank $r$ from rank 0/1 theorem, as well as potential extensions to topological K-theory and Hochschild homology.

In summary, Karpov and Moreira provide a robust, functorial framework for K-theoretic wall-crossing that removes the dependency on framing functors, thereby unlocking the calculation of invariants for a much broader class of stability conditions in modern algebraic geometry.