My question is about a claim which I suspect is true in some generality, but that I have not seen stated anywhere in the literature, and for which I do not know of any counter-examples. Most of the content of this question arose within a collaboration with Dominic Williamson.

Setup: Consider some theory $\mathcal{T}$ in $D$ spacetime dimensions (could be a QFT or a quantum lattice model) with a non-anomalous finite Abelian $q$-form global symmetry group $A^{(q)}$. Feel free to add whatever other assumptions you would like about $A^{(q)}$. It is well-known that if one gauges such a symmetry, then the gauged theory $\mathcal{T}/A^{(q)}$ enjoys an emergent ``quantum'' $(D-2-q)$-form global symmetry group $\hat{A}^{(D-2-q)}$ with $\hat{A}$ the Pontryagin dual of $A$.

Claim: If $A^{(q)}$ is spontaneously broken in the theory $\mathcal{T}$, then $\hat{A}^{(D-2-q)}$ is preserved in the theory $\mathcal{T}/A^{(q)}$. Conversely, if $A^{(q)}$ is preserved in $\mathcal{T}$, then $\hat{A}^{(D-2-q)}$ is spontaneously broken in $\mathcal{T}/A^{(q)}$. I also have a guess about how this claim generalizes to the case of partial spontaneous symmetry breaking.

Examples: An example of this claim in action is the transverse field Ising model, $$H=-J\sum_{\langle v,v'\rangle} Z_v Z_{v'} -h\sum_v X_v$$ where $X_v,Z_v$ are Pauli operators which act on qubits supported on the sites of a $(D-1)$-dimensional lattice. As the coupling $J/h$ is varied one passes through a spontaneous symmetry breaking transition of its $\mathbb{Z}_2$ global symmetry $U = \prod_v X_v$ . In $D=2$, this happens at $J/h=1$, and gauging the $\mathbb{Z}_2$ global symmetry simply swaps $h$ and $J$ (i.e. gauging recovers the Kramers-Wannier dual), so that the claim is easily seen to be true. In $D=3$, gauging the $\mathbb{Z}_2$ global symmetry results in $\mathbb{Z}_2$ lattice gauge theory, $$H/\mathbb{Z}_2 = -J\sum_e X_e -h\sum_p \prod_{e\in\partial p}Z_e$$ where the first sum is over edges of the lattice, and the second is over plaquettes. (Additionally this model should be supplemented by a Gauss's law constraint of the form $G_v = \prod_{e\in\mathrm{star}(v)} X_v = 1$ for all sites $v$.) This theory famously undergoes a confinement/deconfinement phase transition associated with the spontaneous breaking of its $\mathbb{Z}_2^{(1)}$ one-form symmetry, and it is straightforward to see that when the Ising model is spontaneously broken, $\mathbb{Z}_2$ lattice gauge theory preserves its one-form symmetry, and vice versa. A similar pattern continues to hold for all $D>3$ as well, and I believe there are continuum versions of the statements I've made above.

Another example is (3+1)D $\mathrm{SU}(N)$ Yang-Mills theory which has a $\mathbb{Z}_N^{(1)}$ one-form symmetry which is conjectured to be preserved (as measured by the area law of the fundamental Wilson loop). On the other hand, $\mathrm{SU}(N)/\mathbb{Z}_N^{(1)}$ Yang-Mills theory has a 't Hooft line charged under the emergent $\widehat{\mathbb{Z}}_N^{(1)}$ global symmetry which has a perimeter law, and so the symmetry is spontaneously broken.

My question: Does anyone know if the claim holds in general, and if so, to what extent it has been appreciated in the literature? I suspect that it is possible to prove it by arguing that an order parameter for $A^{(q)}$ in $\mathcal{T}$ maps via the gauging procedure to a disorder parameter for $\hat{A}^{(D-2-q)}$ in $\mathcal{T}/A^{(q)}$, and furthermore that their vacuum expectation values are the same (roughly because discrete gauge theories are topological and don't impact local physics). This would then prove the claim because e.g. a non-zero order parameter indicates spontaneous symmetry breaking, while a non-zero disorder parameter indicates symmetry preserving.


Your Answer

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

Browse other questions tagged or ask your own question.