Derived Algebraic Cobordism

Toni AnnalaarXiv 2022 · DOI · arXiv

Abstract

We construct and study a theory of bivariant cobordism of derived schemes. Our theory provides a vast generalization of the algebraic bordism theory of characteristic 0 algebraic schemes, constructed earlier by Levine and Morel, and a (partial) non-$\mathbb{A}^1$-invariant refinement of the motivic cohomology theory $MGL$ in Morel–Voevodsky's stable motivic homotopy theory. Our main result is that bivariant cobordism satisfies the projective bundle formula. As applications of this, we construct cobordism Chern classes of vector bundles, and establish a strong connection between the cobordism cohomology rings and the Grothendieck ring of vector bundles. We also provide several universal properties for our theory. Additionally, our algebraic cobordism is also used to construct a candidate for the elusive theory of Chow cohomology of schemes.

Introduction

In algebraic topology, cobordism is a foundational method for classifying manifolds by determining if they can serve as the boundary of a higher-dimensional shape. Translating this concept into algebraic geometry—where shapes are defined by polynomial equations as schemes—has historically been a complex challenge.

Prior to this work, two primary frameworks dominated the field. The first, developed by Levine and Morel, offered a geometric model for algebraic bordism but was constrained by a reliance on the resolution of singularities, a technique primarily applicable in characteristic 0. The second, Voevodsky’s motivic theory ($MGL$), utilized the power of stable motivic homotopy theory but was inherently $\mathbb{A}^1$-invariant, meaning it could not distinguish between a scheme and its affine line bundle. This research bridges these gaps by constructing a bivariant theory of algebraic cobordism that functions across all characteristics and leverages the framework of derived algebraic geometry (DAG).

The Framework of Derived Algebraic Geometry

The primary engine of this new theory is Derived Algebraic Geometry. In classical algebraic geometry, the intersection of two schemes can lose critical information if the intersection is tangential or occurs at a singularity. DAG addresses this by replacing classical rings with derived rings (simplicial commutative rings). This allows the theory to track the "virtual" data of intersections, ensuring that pullbacks are well-behaved even in non-smooth settings.

By utilizing derived blowups—a technique recently advanced by Khan and Rydh—the theory avoids the necessity of resolving singularities. This methodological shift is what allows the construction to remain valid in arbitrary characteristics, including finite fields, where classical resolution techniques often fail.

Bivariant Formalism and Construction

The theory is built using the bivariant formalism of Fulton and MacPherson, which defines groups for morphisms between spaces rather than just the spaces themselves. This approach unifies homology (cycles) and cohomology (operators) into a single framework.

Constructed as a universal bivariant theory, it uses projective morphisms as generators and is governed by derived double point relations. Because every pullback is well-behaved in the derived setting, the construction of the theory becomes more transparent and universal than its classical predecessors. It effectively acts as a non-$\mathbb{A}^1$-invariant refinement of the motivic spectrum $MGL$, retaining geometric information that is typically lost in the homotopy-theoretic approach.

The Projective Bundle Formula and Chern Classes

A central achievement of this work is the proof of the projective bundle formula. This formula demonstrates that the cobordism of a projective bundle can be computed directly from its base space. This result is the essential prerequisite for defining cobordism Chern classes—the fundamental invariants used to identify and classify vector bundles.

Establishing this formula allows for the construction of a candidate for Chow cohomology on singular schemes. Chow cohomology has long been a sought-after goal in intersection theory, and this derived approach provides a rigorous path toward its realization.

Connections to K-Theory and Universal Properties

The research establishes a profound connection between cobordism and the Grothendieck ring of vector bundles ($K$-theory). By proving a derived version of the classic Conner–Floyd theorem, the author shows that algebraic $K$-theory can be recovered from cobordism by forcing the formal group law to be multiplicative ($x + y - xy$).

Furthermore, the theory is shown to be the universal bivariant theory satisfying a weak version of the projective bundle formula. This implies that any other theory sharing these fundamental properties must naturally map from this derived cobordism theory. In characteristic 0, the theory is shown to be isomorphic to the classical Levine-Morel theory, ensuring that it remains compatible with established geometric results while extending their reach.

Limitations and Future Directions

While the theory provides a robust foundation, several avenues for future research remain. Currently, the "localization exact sequence"—a vital tool for practical calculations—is proven only for specific cases, such as in characteristic 0 or with certain coefficients.

Additionally, this work focuses on the degree 0 truncation of the theory. The next logical step is the construction of a full spectrum-valued invariant that satisfies Nisnevich descent. Finally, while the theory refines motivic cobordism, the precise relationship between this geometric model and the abstract motivic spectrum $MGL$ across all characteristics remains an open question in the field.

Summary updated

Abstract

We construct and study a theory of bivariant cobordism of derived schemes. Our theory provides a vast generalization of the algebraic bordism theory of characteristic 0 algebraic schemes, constructed earlier by Levine and Morel, and a (partial) non-$\Ab^1$-invariant refinement of the motivic cohomology theory $MGL$ in Morel--Voevodsky's stable motivic homotopy theory. Our main result is that bivariant cobordism satisfies the projective bundle formula. As applications of this, we construct cobordism Chern classes of vector bundles, and establish a strong connection between the cobordism cohomology rings and the Grothendieck ring of vector bundles. We also provide several universal properties for our theory. Additionally, our algebraic cobordism is also used to construct a candidate for the elusive theory of Chow cohomology of schemes.
BibTeX
@misc{annala2022derived,
  author = {Toni Annala},
  title = {{Derived Algebraic Cobordism}},
  year = {2022},
  eprint = {2203.12096},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2203.12096},
}

Loading related content...

Figure