nLab manifold diagram




Manifold nn-diagrams generalize string diagrams to higher dimensions—they recover string diagrams in dimension n=2n = 2. For n=3n = 3, they specialize to the notion of surface diagrams.

The idea of manifold diagrams has its roots in the relation between stratified manifolds and higher categories. One way to think about this relation is by noting that stratified manifolds arise after dualizing (in the sense of Poincaré duality) pasting diagrams of directed cells in higher categories.

A well-known instance of this relation materializes in the generalized tangle hypothesis: indeed, tangles can (in a canonical way) be regarded as examples of manifold diagrams.

Figure 1: m-Tangles in dimension n can be regarded as manifold diagrams (picture reproduces part of Fig. 30 in BaezDolan95); note that k = n-m is the codimension of the tangle).

From the perspective of the tangle hypothesis, manifold diagrams can be used to express the structure of coherently dualizable objects. More precisely, this translates between “manifold singularities” and relations satisfied by the higher morphisms associated to a coherently dualizable objects. For instance, the manifold diagram in Figure 2 expresses the “swallowtail identity” satisfied by coherent biadjunctions (this is a manifold 4-diagram: think about the two surfaces as embedded in 3-dimensional space, and in the 4th spatial dimension we deform the top surface into the bottom one). The terminology derives from Thom’s classification of classical singularities (cf. this picture).

Figure 2: The swallowtail identity as a manifold 4-diagram (reproducing the title image of MSRI program 323: higher categories and categorification).

Many other interesting categorical and higher-algebraic identities can be expressed in manifold diagrams in a similar way. As an example, the Yang–Baxter equation (or, relatedly, the Reidemeister moves) can be expressed in manifold 4-diagrams as well, see Figure 3.

Figure 3: The Yang-Baxter identity as a manifold 4-diagram, illustrated as movie of 1-tangles in dimension 3 read from left to right.

In higher dimensions the types of identites become more and more complicated: Figure 4 shows a manifold 5-diagram representing the Zamolodchikov tetrahedron equation). Much diagrammatic algebra of surfaces along these lines as been developed by Carter and collaborators.

Figure 4: The Zamolodchikov tetrahedron equation as a manifold 5-diagram, illustrated as a movie from left to right (but the movie only has 2 frames).

Manifold diagrams are meant to provide a definitional framework for higher algebraic laws such as those above in all dimensions. Moreover, while the above examples only illustrate such laws for tangles (i.e. for embedded ordinary manifolds), in full generality manifold diagrams are diagrams of stratified manifolds.



Several definitional variants exist in dimensions 4\leq 4 (see also at surface diagrams). In general dimensions, manifold diagrams can be defined as conical, compactly triangulable stratifications of directed euclidean space.


A manifold nn-diagram is a conical, compactly triangulable stratification of standard nn-dimensional directed space.

Here, “standard directed space” can be taken to mean standard n \mathbb{R}^n with standard n n -framing. The term “conical” means a directed version of usual conicality; namely, in the context of directed spaces one requires tubular neighborhoods to interact nicely with the directions of the underlying directed space. “Compact triangulability” means, roughly, that up to isomorphism a description by finite data can be given.


We provide details for the preceding definition.


The standard nn-framing of n\mathbb{R}^n is the chain of oriented \mathbb{R}-fiber bundles π i: i i1\pi_i : \mathbb{R}^i \to \mathbb{R}^{i-1} (1in1 \leq i \leq n) with π i\pi_i defined to be the map that forgets the last coordinate of i\mathbb{R}^i (and fibers carry the standard orientation of \mathbb{R} after identifying i= i1×\mathbb{R}^i = \mathbb{R}^{i-1} \times \mathbb{R}).

When considering n\mathbb{R}^n we tacitly always think of it as ‘standard framed n\mathbb{R}^n’ and, thus, we stop mentioning the standard framing as an explicit structure all-together. Indeed, more important than defining the standard nn-framing is to define the maps that preserve it.


A framed map F: n nF : \mathbb{R}^n \to \mathbb{R}^n is a map for which there exist (necessarily unique) maps F j: j jF_j : \mathbb{R}^j \to \mathbb{R}^j (0jn0 \leq j \leq n) with F n=FF_n = F such that π iF i=F i1π i\pi_i \circ F_i = F_{i-1} \circ \pi_i with F iF_i preserving orientations of fibers of π i\pi_i (i.e. mapping fibers strictly monotously).

A framed stratified map ( n,f)(,g)(\mathbb{R}^n,f) \to (\mathbb{R},g) is a stratified map whose underlying map n n\mathbb{R}^n \to \mathbb{R}^n is framed. Moreover, when working with products ( k,f)×( nk,g)(\mathbb{R}^k,f) \times (\mathbb{R}^{n-k}, g) we will identitify n k× nk\mathbb{R}^n \cong \mathbb{R}^k \times \mathbb{R}^{n-k} in the standard way; and, when working with cone stratifications (Cone(S n1),cone(l))(\mathrm{Cone}(S^{n-1}), \mathrm{cone}(l)), we will standard embed S n1into nS^{n-1} \into \mathbb{R}^n and identify Cone(S n1) n\mathrm{Cone}(S^{n-1}) \cong \mathbb{R}^n by mapping (xS n1,λ[0,1))(x \in S^{n-1},\lambda \in [0,1)) to λ1λx n\frac{\lambda}{1 - \lambda}x \in \mathbb{R}^n.


A stratification ( n,f)(\mathbb{R}^n,f) is framed conical if each point x nx \in \mathbb{R}^n it has a framed stratified neighborhood (framed) homeomorphic to k×(Cone(S nk1),cone(l))\mathbb{R}^k \times (\mathrm{Cone}(S^{n-k-1}), \mathrm{cone}(l)) with x k×{0}x \in \mathbb{R}^k \times \{0\}, where 0kn0 \leq k \leq n and (S n1,l)(S^{n-1}, l) is some stratification.


A compactly-described triangulation KK of n\mathbb{R}^n is a finite stratification of n\mathbb{R}^n by open disks whose closures are the images of linear embeddings Δ k× 0 linto n\Delta^k \times \mathbb{R}^l_{\geq 0} \into \mathbb{R}^n (k+lnk + l \leq n). This now translates to the framed stratified case as follows: a stratification ( n,f)(\mathbb{R}^n,f) is framed compactly triangulable if it admits a framed stratified subdivision ( n,K)( n,f)(\mathbb{R}^n,K) \to (\mathbb{R}^n,f) of ff by a compactly-described triangulation KK.

We can now put these concepts together to obtain the following form definition of manifold diagrams.


(Formalization of earlier sketch). A manifold nn-diagram is a framed conical, framed compactly triangulable stratification of n\mathbb{R}^n.

The natural ‘isomorphism’ relation of manifold diagrams is framed stratified homeomorphism.


The definition has several important consequences.

  • While the compact triangulability condition requires the existence of some combinatorial representation, there is in fact a canonical combinatorial representation for isomorphism classes of manifold diagrams. This leads to a theory of combinatorial objects called “trusses”.

  • Strata in manifold diagrams are, indeed, manifolds. Moreover, they have canonical smooth structures. This follows since framed PL manifolds have unique framed smooth structure.

  • Links of tubular neighborhoods in manifold diagrams are in fact well-defined. That is, there is, up to isomorphism, a unique choice of link. This stands in contrast to classical topological conical stratifications, where non-homeomorphic choices of links exist.

  • Manifold diagrams have canonical geometric duals (in the sense of Poincaré duality). This relates manifold diagrams to diagrams of cells, and this may be considered as classical pasting diagrams (but with cell shapes that generalize the usual shapes, such as ‘globular’, ‘simplicial’, or ‘opetopic’ shapes).

These consequences have been worked out in Dorn and Douglas 22. (Note the primary definition in the latter text differs from that given here, but a comparison is given.)


A fast introduction to manifold diagrams can be found in:

Last revised on March 29, 2023 at 08:18:08. See the history of this page for a list of all contributions to it.