The Hurewicz theorem in Homotopy Type Theory

J. Daniel Christensen, Luis Scoccola

TL;DR

Researchers have successfully proven the Hurewicz theorem within Homotopy Type Theory (HoTT), establishing a fundamental link between homotopy groups and homology groups. Because HoTT serves as a formal language for all $\infty$-toposes, this result generalizes a classic piece of algebraic topology to a vast array of mathematical universes beyond traditional topological spaces.

What Problem Does This Paper Address?

In classical topology, the Hurewicz theorem is a cornerstone result. It tells us that for a sufficiently "connected" space (one without low-dimensional holes), the first non-trivial homotopy group is essentially the same as the first non-trivial homology group. This is incredibly useful because homotopy groups are notoriously difficult to compute, while homology groups are much more tractable.

However, proving this in Homotopy Type Theory (HoTT)—a modern framework that merges logic, type theory, and geometry—is a significant challenge. HoTT is "constructive" and "synthetic," meaning researchers cannot rely on classical tools like the Law of Excluded Middle or the Axiom of Choice, which are often used in standard topological proofs. Furthermore, proving that these relationships are "natural" (consistent across different maps between spaces) requires sophisticated machinery to handle the "higher-dimensional" nature of equality in HoTT.

Methods

The authors employ several novel "synthetic" techniques to build the proof from the ground up:

  • Theorem S (Smash Product Theorem): A major prerequisite was proving how the "smash product" of two types affects their connectivity and homotopy groups. They showed that the lowest non-trivial homotopy group of a smash product $X \wedge Y$ is the tensor product of the homotopy groups of $X$ and $Y$.
  • The "Magma" Trick: To avoid the "coherence hell" of tracking infinite layers of algebraic laws (like associativity) in loop spaces, the authors used magmas—types with a binary operation but no required laws. They showed that "weak" magma morphisms are enough to define the necessary maps, simplifying the formalization significantly.
  • 1-Coherent Categories: They developed a framework for "1-coherent categories" to rigorously express naturality and use the Yoneda Lemma without needing the full, complex machinery of $(\infty, 1)$-categories.
  • Formalization in Coq: Much of the work was implemented and verified using the Coq proof assistant and the HoTT library, ensuring the logical correctness of the arguments.

Datasets

Not applicable. This is a theoretical paper in pure mathematics and computer science; no empirical datasets were used.

Key Insights and Ideas

  • Universality via $\infty$-toposes: Because HoTT has models in any $\infty$-topos, this proof isn't just about "spaces" in the way we usually think of them. It applies to any mathematical structure that behaves like a space, such as sheaves or higher stacks.
  • Homology via Prespectra: The authors define homology groups as the stable homotopy groups of a specific "prespectrum" (a sequence of types). This approach is native to the HoTT environment and avoids the need for traditional chain complexes.
  • Uniqueness up to Sign: The paper proves that the Hurewicz homomorphism is unique up to a sign change, provided it satisfies certain naturality and normalization conditions. This mirrors a deep result in classical stable homotopy theory.

Key Results

  • Theorem H (The Main Event): For a pointed, $(n-1)$-connected type $X$ and an abelian group $A$, there is a natural isomorphism: $$\pi_n(X)^{ab} \otimes A \cong \tilde{H}_n(X; A)$$ This relates the abelianization of the $n$-th homotopy group directly to the $n$-th reduced homology group.

  • Connectivity of Smash Products: They proved that if $X$ is $(n-1)$-connected and $Y$ is $(m-1)$-connected, then $X \wedge Y$ is $(n+m-1)$-connected.

  • Truncation Independence: They demonstrated that low-dimensional homology is unaffected by high-dimensional truncation—a vital property for practical calculations.

Limitations and Future Work

While this is a major milestone, the authors note several remaining gaps in the HoTT homology toolkit:

  • Missing Axioms: A full proof that this homology theory satisfies all the Eilenberg–Steenrod axioms (the "gold standard" for homology) is still a work in progress.
  • Spectral Sequences: Tools like the Serre spectral sequence, which are essential for advanced calculations in topology, have not yet been fully formalized in this framework.
  • Cellular Homology: The paper notes that we still lack a proof that the homology of a "cellular" space (built out of simple blocks) can be computed purely through its cellular structure in HoTT.

The authors suggest that future work should focus on these missing computational tools to make HoTT a fully-featured environment for algebraic topology.

Summary updated

Abstract

We prove the Hurewicz theorem in homotopy type theory, i.e., that for $X$ a pointed, $(n-1)$-connected type $(n \geq 1)$ and $A$ an abelian group, there is a natural isomorphism $π_n(X)^{ab} \otimes A \cong \tilde{H}_n(X; A)$ relating the abelianization of the homotopy groups with the homology. We also compute the connectivity of a smash product of types and express the lowest non-trivial homotopy group as a tensor product. Along the way, we study magmas, loop spaces, connected covers and prespectra, and we use $1$-coherent categories to express naturality and for the Yoneda lemma. As homotopy type theory has models in all $\infty$-toposes, our results can be viewed as extending known results about spaces to all other $\infty$-toposes.
BibTeX
@inproceedings{christensen2023hurewicz,
  author = {J. Daniel Christensen and Luis Scoccola},
  title = {{The Hurewicz theorem in Homotopy Type Theory}},
  booktitle = {Algebraic},
  year = {2023},
  doi = {10.2140/agt.2023.23.2107},
}
Figure