As in my previous two posts (1 & 2), a unitary two-dimensional TQFT is specified by a set of real numbers $\{\lambda_i\}$ such that the partition function is $$ Z(\lambda)=\sum_{i=1}^n\lambda_i^{g-1}\tag1 $$

My main question is whether there is a natural, "first principles" construction of such a TQFT other than by specifying these real numbers. This question is inspired by the similar situation in $d=3$, where the "instrinsic data" is a modular tensor category, but there is also the Chern-Simons construction which supposedly describes an arbitrary three-dimensional TQFT. In particular, Chern-Simons actions are specified by some Lie Group. So the intrinsic data is the MTC, but it can be conveniently packaged into a choice of Lie Group. I wonder whether a similar situation holds in two-dimensional TQFTs:

  1. Can every TQFT$_2$ be obtained in the path-integral formalism? (By this I mean continuous field theory rather than a lattice prescription). What is the corresponding Lagrangian?

  2. Is there a set of basic "building blocks" such that any other TQFT can be obtained from them via a set of simple operations? These would be akin to the simple groups in three-dimensions, such that any other Lie Group can be obtained through direct product, extensions, and quotients.


1 Answer 1


Well I guess in $2d$ we can write a continuum topological term, that is a metric independent term which also has a zero Hamiltonian this is the definition of the topological term after all, as

$$S=\int_M dA$$ where $A$ is 1-form $U(1)$ gauge field. You can extend this to non-abelian case as well I think that would answer your question on packing MTC to Lie Group(for the non-abelian case we would have $S=\text{tr}(\int_M dA+A\wedge A$ i am not extremely sure about the non-abelian case though ).

We also can extend this abelian case to all even dimensions. In $4d$ this would be just theta term. In odd dimensions, the topological term is just d dimensional chern simons term.

We can also write a general topological term that is valid for the general dimension whether it is even or odd. For example;

For $d=n+1>2$ we can have a general topological term now in the lattice as

\begin{equation} S_{\text{top}} = i \pi \int_Y (A_{n-1} \cup_{n-3} A_{n-1} + A_{n-1} \cup_{n-2} \delta A_{n-1}), \end{equation} where $A_{n-1}$ is $(n-1)$-form gauge fields, a $(n-1)$-cochain $A_{n-1} \in C^{n-1}(Y,\mathbb{Z}_2)$, and $Y$ is $(n+1)$-dimensional spacetime manifold. The continuum limit of this is quite trivial for d=3 for higher dimensional case I have no idea how to take continuum limit.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.