# 4-dimensional TQFT with/without requiring spin structure

My questions here are focused on $D$-dimensional topological quantum field theories (TQFTs) which are unitary and which have finite dimensional Hilbert space on a closed spatial manifold $M^{d-1}$. Say the partition function $$Z(M^{d-1} \times S^1)=\dim \mathcal{H}$$ is finite. The spacetime $M^{d-1} \times S^1$ can be viewed as a closed spatial manifold $M^{d-1}$ and a compact time.

We know that there are a large classes of 3-dimensional (3d) TQFT defined on a spacetime 3-manifold. Many of them are some version of Chern-Simons TQFT. Some of them can defined on any closed manifold, other TQFTs require to have the spin-manifold to be defined. For example, even for the Chern-Simons thoery, we have the $3d$ spin Chern-Simons theory requiring the spin-manifold.

My question: What are some examples of 4d TQFT with or without requiring spin structure, say the $4d$ spin-manifold? There are a large classes of TQFTs described by twisted discrete gauge theories, say the Dijkgraaf-Witten theory. What are other 4d TQFT with or without requiring spin structure / spin-manifold? Do we have the continuous Lagrangian formulation or the lattice formulations of the theories?

• The spin-statistics theorem asserts that, if you insist that your (T)QFTs be unitary, then you need spin structures iff you have fermions. So super Dijkgraaf–Witten theories, for example, require spin structures. See for example arxiv.org/abs/1505.05856. – Theo Johnson-Freyd Dec 19 '16 at 5:23
• Heegard Floer homology is an example. – Soutrik Jan 31 '17 at 14:41
• Heegard Floer homology is an example for what? – miss-tery Feb 1 '17 at 3:57

Crane-Yetter is defined for a ribbon fusion (premodular) category $\mathcal{C}$. As far as I know, it is unitary if $\mathcal{C}$ is. The Crane-Yetter theory doesn't require spin structures, although it can be defined as a spin TQFT if your ribbon fusion category has a $\mathbb{Z}_2$-grading. Such a grading can arise when you have a nontrivial twist on the transparent objects (basically a finite supergroup symmetry).
Recently, Crane-Yetter has been generalised from ribbon fusion categories to $G$-crossed braided fusion categories, see this article by Shawn Cui.