A model of spaces inside every strict higher category dimension

← All digests

Family 317 · Topology
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-22

1. The question

A category consists of objects and composable arrows. Surprisingly, categories can encode arbitrary homotopy types: the chains of composable arrows form a simplicial space called a nerve. This allows questions about spaces to be studied through algebraic diagrams.

A higher category adds arrows between arrows, then further cells between those, and so on. Family 317 asks whether strict globular higher categories admit the full machinery of a model category matching the homotopy theory of spaces, in every positive dimension.

M1 claims an affirmative answer, including infinitely many dimensions. The new content is not merely that the homotopy categories agree; that was already known. It is the prescribed system of weak equivalences, fibrations, cofibrations, factorizations, and lifting properties on the entire category of strict higher categories.

2. Background and essential definitions

In an ordinary category, arrows compose associatively and have identity arrows. A strict 2-category also has 2-cells between arrows, with compatible compositions. A strict globular $n$-category continues this through dimension $n$. “Globular” means a cell has a source and target of one lower dimension. “Strict” means the associativity, identity, and interchange rules are equalities.

Cells are not required to be invertible. This distinguishes the objects here from strict higher groupoids. A strict $\omega$-category has cells in every finite dimension; a finite $n$-category can be viewed as one whose higher cells are identities.

A simplicial set is a system of vertices, edges, triangles, and higher simplices with face and degeneracy maps. It need not be a simplicial complex embedded in Euclidean space. Geometric realization turns it into a topological space.

The Street nerve $N_nX$ records maps from oriented simplex categories, called orientals, into a strict higher category $X$. It generalizes the ordinary nerve. Its left adjoint $c_n$ freely turns simplicial data into a strict $n$-category.

A model category organizes homotopy theory using three classes of maps. Weak equivalences are maps to be inverted; fibrations and cofibrations control lifting problems and replacements. Two required factorizations split every map into suitable combinations of these classes. A Quillen equivalence means that the resulting derived homotopy theories agree through a compatible adjoint pair.

Subdivision $\operatorname{Sd}$ refines simplices into smaller pieces. Its right adjoint $\operatorname{Ex}$ changes a simplicial set in a way that improves its filling behavior while preserving homotopy type. Applying each twice is part of the specific construction here, not an arbitrary cosmetic choice.

3. Development history and previous partial results

Thomason established the ordinary-category case in 1980. His argument transferred the simplicial model structure through a twice-subdivided nerve adjunction. A central issue was how nerves behave when categories are glued by pushouts. Cisinski later corrected a retract argument while retaining the model structure and its properness.

Street’s orientals supplied the higher-dimensional nerve. Ara and Maltsiniotis developed an abstract transfer framework and established the proper combinatorial structure in dimension two. Together with Chiche’s comparison result, this gave the corresponding Quillen equivalence. Earlier proposed two-dimensional constructions by Worytkiewicz, Hess, Parent, and Tonks required corrections, as discussed in the cited literature.

For all dimensions, Ara–Maltsiniotis reduced the problem to two conditions: a unit comparison on nerves of partially ordered sets, and a compatibility statement for certain categorical pushouts with arbitrary targets. They proved the unit condition. Gagna later gave a direct simplicial approach and showed that localizing strict higher categories at nerve weak equivalences already recovers homotopy types.

Thus a basic equivalence of homotopy categories was available before M1. What remained was the pushout control needed to construct the specified lifting and factorization systems everywhere, including attachments to categories that are neither free nor cofibrant.

Fiore and Paoli also established Thomason structures for $n$-fold categories. Those are built by iterating the internal-category construction and have a different organization of cells. Their theorem does not directly supply the strict globular structure considered here. M1 explicitly distinguishes these historical achievements and claims to settle the remaining globular conjecture.

4. What the new manuscripts claim

M1, dated 25 September 2026, considers every

$$n\in\{1,2,3,\ldots\}\cup\{\omega\}.$$

On small strict globular $n$-categories, it defines

$$R_n=\operatorname{Ex}^2N_n, \qquad L_n=c_n\operatorname{Sd}^2.$$

A functor is a weak equivalence exactly when its Street nerve is a weak homotopy equivalence of simplicial sets. Equivalently, this can be tested after applying $\operatorname{Ex}^2$. A functor is a fibration exactly when $R_n$ sends it to a Kan fibration, the simplicial lifting condition for horns.

Cofibrations are determined by the left lifting property against maps that are both fibrations and weak equivalences. The manuscript claims that these classes form a proper combinatorial model structure, and that

$$L_n:\mathrm{sSet}\rightleftarrows n\mathrm{Cat}:R_n$$

is a Quillen equivalence with the usual homotopy theory of simplicial sets.

“Combinatorial” means the underlying category has suitable presentability properties and the model structure is generated by sets of maps. Here the generators come from twice-subdivided simplex boundaries and horns. “Proper” means weak equivalences behave well under the relevant pushouts along cofibrations and pullbacks along fibrations.

The statement applies to all small strict categories of the specified dimension. It is not restricted to free presentations or cofibrant inputs, and it is a full model structure, rather than only a semi-model structure.

5. Why the advance matters

Importance: 5/5, provisionally. If correct, the result completes a central model-structure problem in strict higher-category theory, with one uniform construction across all positive finite dimensions and dimension $\omega$.

A model structure gives more than a list of equivalent objects. It provides replacements, derived constructions, and lifting tools compatible with how objects are attached. These are the mechanisms used to calculate homotopy limits and colimits and to compare different algebraic presentations.

The arbitrary-target scope is especially significant. A construction restricted to free inputs might miss the attachments that arise during a factorization or a derived gluing problem. The claim supplies the structure on the full category where those operations actually take place.

6. The main idea

Make a contraction stationary on the attaching part

The central pushout problem begins with a partially ordered set $E$ and a downward-closed full subset $A$. Suppose the inclusion has a right adjoint retraction. After turning their nerves into strict higher categories, the manuscript seeks a weak equivalence after attaching the larger object along any map out of the smaller one.

The geometric idea is to contract the added part while fixing the old part throughout the contraction. If that equality holds strictly on the attaching subcategory, the contraction can be glued to the constant cylinder on an arbitrary target category.

Ordinary homotopy intuition is insufficient because a categorical pushout creates new composites. The cylinder must control those composites as well as the named generators.

Construct positive directed prisms

The manuscript works with directed chain complexes, where selected positive chains encode allowed higher-cell directions. Under specific monotonicity hypotheses, it writes an explicit prism map between two maps of posets. Repeated vertices become zero after normalization, and the resulting map preserves positivity.

Through Steiner’s description of orientals and the Gray tensor product, this chain map gives a strict cylinder. Its restriction to the old subcategory remains constant in the interval direction. This is the algebraic reason it survives arbitrary attachments.

When the newly added part has a greatest element, two such cylinders yield a relative contraction by a zigzag. For a general finite poset, the argument removes different maximal new elements and covers the attachment by smaller pieces. Their nerves glue along monomorphisms, so the local weak-equivalence conclusions combine.

The contractions on overlapping pieces need not be chosen compatibly. What glues is the homotopical conclusion, using the precise sieve and intersection descriptions. Passing through finite subposets stable under the retraction treats arbitrary posets.

Transfer from simplicial attachments

Two subdivisions put simplex boundaries and horns into the required poset form. The relative contraction theorem then supplies the comparison between a pushout of nerves and the nerve of the categorical pushout.

That comparison permits the small-object constructions producing both factorization systems. Additional presentability and finite-presentation arguments ensure that the generating objects are small enough for those constructions. The paper then establishes properness and the adjunction-unit comparison for arbitrary simplicial sets, giving the claimed Quillen equivalence.

The infinite-dimensional argument uses finite orientals directly. Reflection to finite dimension carries the same construction to each finite $n$, while respecting the necessary relations in the top retained dimension.

7. Scope, limitations, and open questions

The theorem concerns strict globular categories with possibly noninvertible cells. It does not contradict the limitations of strict globular groupoids discussed in family 312. Noninvertible categorical data can encode a general homotopy type through its nerve, even though a strictly invertible globular model is more restrictive.

A finite category dimension does not bound the dimensions in which its nerve can have homotopy. Already ordinary categories can model arbitrary homotopy types. Thus “$n$-category” here is not synonymous with “space having no homotopy above degree $n$.”

The weak equivalences are detected by the Street nerve, not by literal isomorphism or the usual equivalence relation between higher categories. Different choices of weak equivalence lead to different homotopy theories.

M1 compares its method with the weak-groupoid expansion argument in family 312, but states that the latter is not an input. The results address related foundations with different objects and different model-structure conclusions. Explicit computational models and comparisons with other higher-category frameworks remain further applications to develop.

8. Sources and further reading

M1 supplies the theorem and stationary-cylinder construction. Primary background includes Thomason’s original model structure, Ara–Maltsiniotis’s transfer framework and dimension-two result, Gagna’s equivalence with homotopy types, and Steiner’s directed-chain description. These references distinguish the previously known homotopy-category comparison from the additional structure claimed here.

Source manuscripts

• M1. OpenAI, Thomason Model Structures in Every Strict Higher Dimension (September 25, 2026).

Sources consulted on 22 October 2026 at repository revision 1581d4c. Download the manuscript citations.