Introduction
It has recently been pointed out that the mixed ’t Hooft anomaly between the CP and center symmetries sheds light on the vacuum structure of non-supersymmetric Yang-Mills theories [1 ] . In particular, for the S U ( 2 ) SU(2) S U ( 2 ) gauge group, a phase diagram of the non-supersymmetric Yang-Mills theory was read off from the CP/center mixed anomaly, as a function of the θ \theta θ -parameter. This result turned out to be in perfect agreement with the phase diagram of the super Yang-Mills (SYM) with a small SUSY-breaking mass m m m turned on. Since the non-supersymmetric Yang-Mills can be regarded as the m → ∞ m \to\infty m → ∞ limit, this suggests that the phase diagram is independent of m m m . A crucial step in this analysis is to consider the mixed gauge theory , which makes the mixed center/CP anomaly transparent. One can generalize this analysis to the S U ( N ) SU(N) S U ( N ) gauge group [1 , 2 ] . For recent applications of similar discrete anomalies to phase structures of 4d gauge theories, see [3 , 4 , 5 , 6 ] .
Given the above progress, it is important to apply the technique of CP/center mixed anomaly to other gauge groups and see how the technique of the mixed gauge theory works in a more general setup. In particular, it is interesting to study the generalization to Spin/SO Yang-Mills theory and read off the corresponding phase diagram from the center/CP mixed anomaly as a function of θ \theta θ . While such a generalization was already done for the S p i n ( 3 ) / S O ( 3 ) \mathit{Spin}(3)/SO(3) Spin ( 3 ) / S O ( 3 ) Yang-Mills theory in [1 ] , the situation for general Spin/SO gauge groups is still to be understood.
In this paper, we apply the technique of the center/CP mixed anomaly and mixed gauge theory to the S p i n ( N ) / S O ( N ) \mathit{Spin}(N)/SO(N) Spin ( N ) / S O ( N ) Yang-Mills theory for all even N N N larger than 3. We particularly identify a phase diagram consistent with the mixed anomaly by studying the mixed gauge theories constructed by compactifying these S O / S p i n SO/\mathit{Spin} S O / Spin Yang-Mills theories on S 1 S^1 S 1 and gauging a 3d one-form symmetry. We find that the D 8 D_8 D 8 symmetry that plays an important role in the S U ( 2 ) SU(2) S U ( 2 ) case is replaced by ( Z 2 ) 4 ⋊ Z 2 (\mathbb{Z}_2)^4 \rtimes\mathbb{Z}_2 ( Z 2 ) 4 ⋊ Z 2 or ( Z 4 ) 2 ⋊ Z 2 (\mathbb{Z}_4)^2 \rtimes\mathbb{Z}_2 ( Z 4 ) 2 ⋊ Z 2 , depending on the value of N N N . The S p i n ( N ) / S O ( N ) \mathit{Spin}(N)/SO(N) Spin ( N ) / S O ( N ) Yang-Mills theory for odd N N N has no center/CP anomaly, and therefore the same technique is not available.
One can regard these Yang-Mills theories as the m → ∞ m \to\infty m → ∞ limit of the softly broken S p i n / S O \mathit{Spin}/SO Spin / S O SYM with a gaugino mass m m m . It is therefore interesting to compare the above results with the phase diagrams of the softly-broken SYM for a small ∣ m ∣ |m| ∣ m ∣ , which can be read off in a perturbative analysis. We perform this comparison to find that the above results are in perfect agreement with the phase diagram of SYM for a small ∣ m ∣ |m| ∣ m ∣ , which suggests that the phase diagram is independent of m m m .
Among these S p i n / S O \mathit{Spin}/SO Spin / S O gauge groups, S p i n ( 4 ) / S O ( 4 ) \mathit{Spin}(4)/SO(4) Spin ( 4 ) / S O ( 4 ) is special. Since S p i n ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) \mathit{Spin}(4) \simeq SU(2) \times SU(2) Spin ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) , the S p i n ( 4 ) / S O ( 4 ) \mathit{Spin}(4)/SO(4) Spin ( 4 ) / S O ( 4 ) Yang-Mills theory has two independent θ \theta θ -parameters, in contrast to the other S p i n / S O \mathit{Spin}/SO Spin / S O cases. Moreover, since the Z 2 \mathbb{Z}_2 Z 2 quotient giving rise to S O ( 4 ) SO(4) S O ( 4 ) from S p i n ( 4 ) \mathit{Spin}(4) Spin ( 4 ) is associated with the diagonal sub-group of Z 2 × Z 2 ⊂ S U ( 2 ) × S U ( 2 ) \mathbb{Z}_2 \times\mathbb{Z}_2 \subset SU(2) \times SU(2) Z 2 × Z 2 ⊂ S U ( 2 ) × S U ( 2 ) , the vacuum structure of S O ( 4 ) SO(4) S O ( 4 ) Yang-Mills theory depends non-trivially on both of these two θ \theta θ -parameters. As a result, the S O ( 4 ) ± SO(4)_{\pm} S O ( 4 ) ± Yang-Mills theories are related by shifts of the θ \theta θ -parameters as
S O ( 4 ) ± θ 1 + 2 π , θ 2 = S O ( 4 ) ± θ 1 , θ 2 + 2 π = S O ( 4 ) ∓ θ 1 , θ 2 = S O ( 4 ) ∓ θ 1 + 2 π , θ 2 + 2 π . (1) SO(4)_{\pm}^{\theta_1+2\pi,\theta_2}=SO(4)_{\pm}^{\theta_1,\theta_2+2\pi}=SO(4)_{\mp}^{\theta_1,\theta_2}=SO(4)_{\mp}^{\theta_1+2\pi,\theta_2+2\pi}. \tag*{(1)} S O ( 4 ) ± θ 1 + 2 π , θ 2 = S O ( 4 ) ± θ 1 , θ 2 + 2 π = S O ( 4 ) ∓ θ 1 , θ 2 = S O ( 4 ) ∓ θ 1 + 2 π , θ 2 + 2 π . (1)
This is totally different from the S p i n ( N ) / S O ( N ) Spin(N)/SO(N) S p in ( N ) / S O ( N ) cases for N > 4 N>4 N > 4 , where S O + SO_{+} S O + and S O − SO_{-} S O − theories are not related by any shift of θ \theta θ .
Given the above special features of S p i n ( 4 ) / S O ( 4 ) Spin(4)/SO(4) S p in ( 4 ) / S O ( 4 ) , we start our analysis with the S p i n ( 4 ) / S O ( 4 ) Spin(4)/SO(4) S p in ( 4 ) / S O ( 4 ) case in § 3 , after reviewing the S U ( 2 ) SU(2) S U ( 2 ) case in § 2 . In particular, we identify a phase diagram of S p i n ( 4 ) / S O ( 4 ) ± Spin(4)/SO(4)_{\pm} S p in ( 4 ) / S O ( 4 ) ± Yang-Mills theories from their center/CP mixed anomaly, using the technique of mixed gauge theory. This result is shown in Figure 4 and Table 6 . Its consistency with the soft SUSY-breaking of S p i n ( 4 ) / S O ( 4 ) ± Spin(4)/SO(4)_{\pm} S p in ( 4 ) / S O ( 4 ) ± SYM is also discussed. We then generalize these analyses to the other S p i n ( e v e n ) / S O ( e v e n ) Spin(\mathrm{even})/SO(\mathrm{even}) S p in ( even ) / S O ( even ) gauge groups in § 4 and § 5 . A phase diagram read off from the center/CP anomaly for S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± Yang-Mills is shown in Figure 7 and Table 8 , and a similar phase diagram for S p i n ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Spin(4k+2)/SO(4k+2)_{\pm} S p in ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Yang-Mills is shown in Figure 9 and Table 10. They all turn out to be in perfect agreement with the result of the soft SUSY breaking. While the technique of center/CP anomaly is not available for S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Yang-Mills for k ≥ 2 k\ge2 k ≥ 2 , the soft SUSY breaking analysis is still possible for them, which we describe in § 6 for completeness. In § 7 , we conclude and discuss future directions. In appendix A, we summarize our notation for the charges of line operators.
Figure 4. The phase diagram of the mixed Spin (4) theory is periodic under θ k → θ k + 4 π \theta_k \to\theta_k + 4\pi θ k → θ k + 4 π . We show a single fundamental domain above. The ordinary Spin (4)/SO (4) Yang-Mills theories have the same periodicity in the presence of the background gauge field of the 1-form symmetry. The IR theories present in the four regions are summarized in Table 6 .
Table 6. The IR theories present in the four regions in Figure 4 for 4d S p i n ( 4 ) / S O ( 4 ) ± Spin(4)/SO(4)_\pm S p in ( 4 ) / S O ( 4 ) ± Yang-Mills theory. In the first line, “S P T n 1 , n 2 \mathrm{SPT}_{n_1,n_2} SPT n 1 , n 2 ” stands for the SPT phase protected by the center Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 symmetry of the S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory, and “trivial” stands for the trivial TQFT, i.e., S P T 0 , 0 \mathrm{SPT}_{0,0} SPT 0 , 0 . In the second and third lines, the SPT is regarded as the SPT phase protected by the electric Z 2 \mathbb{Z}_2 Z 2 and the magnetic Z 2 \mathbb{Z}_2 Z 2 1-form symmetries of S O ( 4 ) SO(4) S O ( 4 ) Yang-Mills theory.
Figure 7. There are two different phases of S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± theories for k ≥ 2 k \ge2 k ≥ 2 , depending on the value of θ \theta θ .
Table 8. The IR theory present in the two regions in Figure 7 for 4d S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± Yang-Mills with k ≥ 2 k\geq2 k ≥ 2 .
Figure 9. There are four different phases in the mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory at zero temperature. The periodicity of θ \theta θ is θ ∼ θ + 8 π \theta\sim\theta+ 8\pi θ ∼ θ + 8 π . This periodicity is changed to be θ ∼ θ + 4 π \theta\sim\theta+ 4\pi θ ∼ θ + 4 π when the mixed gauge theory is reduced to the S O ( 4 k + 2 ) ± SO(4k+2)_{\pm} S O ( 4 k + 2 ) ± Yang-Mills theory, as will be discussed in § 5.2 .
Review of S U ( 2 ) SU(2) S U ( 2 ) case
Here we briefly review, following [1 ] , how the phase diagram of S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory can be read off from the CP/center mixed anomaly via the technique of mixed gauge theory.
CP/center mixed anomaly
The action of the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory is
S = ∫ Tr ( − 1 4 g 2 F ∧ ∗ F + i θ 8 π 2 F ∧ F ) . (2) S=\int\operatorname{Tr}\left(-\frac{1}{4g^2}F\wedge*F+\frac{i\theta}{8\pi^2}F\wedge F\right). \tag*{(2)} S = ∫ Tr ( − 4 g 2 1 F ∧ ∗ F + 8 π 2 i θ F ∧ F ) . (2)
where F F F is the S U ( 2 ) SU(2) S U ( 2 ) gauge field strength, and g g g is the Yang-Mills coupling. In the absence of the background gauge field of the center symmetry, the parameter θ \theta θ has periodicity θ ∼ θ + 2 π \theta\sim\theta+2\pi θ ∼ θ + 2 π since the instanton number 1 8 π 2 ∫ Tr ( F ∧ F ) \frac{1}{8\pi^2}\int\operatorname{Tr}(F\wedge F) 8 π 2 1 ∫ Tr ( F ∧ F ) is always an integer. This implies that, at least classically, the theory at θ = 0 , π \theta= 0, \pi θ = 0 , π is CP-invariant. Indeed, the CP transformation
C P : θ → − θ (3) CP : \theta\to-\theta \tag*{(3)} C P : θ → − θ (3)
preserves θ = 0 \theta= 0 θ = 0 , and also preserves θ = π \theta= \pi θ = π up to the 2 π 2\pi 2 π -shift of the θ \theta θ -parameter.
The S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory has a 1-form Z 2 \mathbb{Z}_2 Z 2 global symmetry called the center symmetry. It keeps all local operators invariant but acts non-trivially on line operators such as the Wilson line. For the S U ( 2 ) SU(2) S U ( 2 ) gauge group, the only genuine line operator is the Wilson line W W W in the fundamental representation, which transforms as
W → − W (4) W \to-W \tag*{(4)} W → − W (4)
under the Z 2 \mathbb{Z}_2 Z 2 center symmetry.
One can gauge the above center symmetry by coupling the theory with a two-form Z 2 \mathbb{Z}_2 Z 2 gauge field [7 , 8 ] . This gauging adds an extra interaction term to the action (2) , which breaks the invariance under θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π [1 ] . This implies that the CP symmetry at θ = π \theta= \pi θ = π is explicitly broken when the center symmetry is gauged, signaling the presence of a CP/center mixed anomaly.
To be more specific, when the theory is coupled to a background gauge field B B B for the Z 2 \mathbb{Z}_2 Z 2 center symmetry, the 2 π 2\pi 2 π -shift of the θ \theta θ parameter induces the following change of the action
Δ S = i π 2 ∫ M P ( B ) , B ∈ Z 2 ( M , Z 2 ) , (5) \Delta S = \frac{i\pi}{2}\int_{\mathcal{M}} \mathcal{P}(B), \qquad B \in Z^2(\mathcal{M},\mathbb{Z}_2), \tag*{(5)} Δ S = 2 iπ ∫ M P ( B ) , B ∈ Z 2 ( M , Z 2 ) , (5)
where M \mathcal{M} M is the spacetime manifold, and B B B is a two-cocycle describing the background Z 2 \mathbb{Z}_2 Z 2 gauge field. The cohomology class P ( B ) \mathcal{P}(B) P ( B ) is the Pontryagin square of B B B , which is an element of H 4 ( M , Z 4 ) H^4(\mathcal{M},\mathbb{Z}_4) H 4 ( M , Z 4 ) .1 Its integral ∫ M P ( B ) \int_{\mathcal{M}}\mathcal{P}(B) ∫ M P ( B ) can be an odd or even integer when M \mathcal{M} M is an arbitrary four-manifold, but is always even when M \mathcal{M} M is a spin four-manifold. In the rest of this paper, we assume that M \mathcal{M} M is a spin manifold, and therefore ∫ M P ( B ) 2 ∈ Z \int_{\mathcal{M}}\frac{\mathcal{P}(B)}{2} \in\mathbb{Z} ∫ M 2 P ( B ) ∈ Z . In that case, we have
Δ S ∈ π i Z . (6) \Delta S \in\pi i\mathbb{Z}. \tag*{(6)} Δ S ∈ π i Z . (6)
Since (6) is not generally an integer multiple of 2 π 2\pi 2 π , we see that the invariance under θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π can be broken by coupling the theory with a two-form gauge field B B B . This is the mixed CP/center anomaly.
One can interpret (6) as a modified periodicity of the θ \theta θ -parameter. Indeed it implies that, in the presence of the background gauge field for the center symmetry, the periodicity
caption: The global symmetry of the 3d mixed S U ( 2 ) SU(2) S U ( 2 ) gauge theory. These are all 0-form symmetries.
:::
of θ \theta θ is replaced by
θ ∼ θ + 4 π . (7) \theta\sim\theta+ 4\pi. \tag*{(7)} θ ∼ θ + 4 π . (7)
This change of periodicity plays an important role in the study of the phase diagram as reviewed below.
Mixed S U ( 2 ) SU(2) S U ( 2 ) gauge theory
In the rest of this section, we review how the CP/center mixed anomaly can be used to read off the phase structure of the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory. A key ingredient is what is called the mixed gauge theory [1 ] .
To describe it, we first compactify the four-dimensional S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory on S 1 S^1 S 1 with the circumference β \beta β . The Z 2 \mathbb{Z}_2 Z 2 center symmetry in four dimensions now gives rise to a Z 2 \mathbb{Z}_2 Z 2 1-form symmetry and a Z 2 \mathbb{Z}_2 Z 2 0-form symmetry in three dimensions. When θ = n π \theta= n\pi θ = nπ for n ∈ Z n \in\mathbb{Z} n ∈ Z , the theory also has an extra Z 2 \mathbb{Z}_2 Z 2 0-form symmetry corresponding to the 4d CP symmetry. We want to identify a vacuum structure that is consistent with these global symmetries.
To that end, we consider gauging the 3d Z 2 \mathbb{Z}_2 Z 2 1-form symmetry, which gives rise to a Z 2 \mathbb{Z}_2 Z 2 0-form global symmetry.2 In other words, we convert the 3d 1-form symmetry into a 3d 0-form symmetry. Since one can always recover the former symmetry by gauging the latter back, no information is lost by this conversion. The resulting 3d theory is called the mixed S U ( 2 ) SU(2) S U ( 2 ) gauge theory . Since the global symmetries are now all 0-form symmetries, one can analyze the vacuum structure of this theory by the standard method of quantum field theory.
The global symmetry of the mixed S U ( 2 ) SU(2) S U ( 2 ) theory is shown in Table 1. When θ ≠ n π \theta\ne n\pi θ = nπ for any n ∈ Z n \in\mathbb{Z} n ∈ Z , the mixed gauge theory has only a Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 0-form symmetry. Here, one Z 2 \mathbb{Z}_2 Z 2 0-form symmetry arises from the 4d center symmetry in the S 1 S^1 S 1 -compactification. We denote it by Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) and its generator by c c c . The other Z 2 \mathbb{Z}_2 Z 2 0-form symmetry arises when gauging the 3d Z 2 \mathbb{Z}_2 Z 2 1-form symmetry. We denote it by Z 2 ( h ) \mathbb{Z}_2^{(h)} Z 2 ( h ) and its generator by h h h .
When θ = n π \theta= n\pi θ = nπ for an integer n n n , the mixed theory has an extra Z 2 \mathbb{Z}_2 Z 2 0-form symmetry arising from the 4d CP symmetry. We denote it by Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) and its generator by C P CP C P . When n n n is even, the generators c c c , h h h and C P CP C P commute with each other, and therefore the full global symmetry group is Z 2 ( c ) × Z 2 ( h ) × Z 2 ( C P ) \mathbb{Z}_2^{(c)} \times\mathbb{Z}_2^{(h)} \times\mathbb{Z}_2^{(CP)} Z 2 ( c ) × Z 2 ( h ) × Z 2 ( C P ) . However, when n n n is odd, c c c and C P CP C P do not commute. Indeed, the 4d CP/center mixed anomaly (2.4) implies the following relations among c c c , h h h and C P CP C P (at θ = n π \theta= n\pi θ = nπ for odd n n n ):
C P ⋅ c ⋅ C P = c ⋅ h , C P ⋅ h = h ⋅ C P , c ⋅ h = h ⋅ c , c 2 = h 2 = C P 2 = 1 . (8) CP \cdot c \cdot CP = c \cdot h \,, \qquad CP \cdot h = h \cdot CP \,, \qquad c \cdot h = h \cdot c \,, \qquad c^2 = h^2 = CP^2 = 1 \, . \tag*{(8)} C P ⋅ c ⋅ C P = c ⋅ h , C P ⋅ h = h ⋅ C P , c ⋅ h = h ⋅ c , c 2 = h 2 = C P 2 = 1 . (8)
This means that the full global symmetry group for odd n n n is D 8 \mathbb{D}_8 D 8 .
Phase diagram of mixed SU(2) gauge theory
One can identify the vacuum structure of the mixed S U ( 2 ) SU(2) S U ( 2 ) gauge theory so that it correctly gives a representation of the global symmetry group shown in Table 1.
To see this, let us first consider 4d line operators wrapped on the S 1 S^1 S 1 . From the 3d view point, they are local operators with non-vanishing electro-magnetic charge ( λ e , λ m ) ∈ Z 2 (\lambda_e,\lambda_m) \in\mathbb{Z}^2 ( λ e , λ m ) ∈ Z 2 . We take the ’t Hooft line with ( λ e , λ m ) = ( 0 , 1 ) (\lambda_e,\lambda_m) = (0,1) ( λ e , λ m ) = ( 0 , 1 ) and the dyonic line with ( λ e , λ m ) = ( 1 , 1 ) (\lambda_e,\lambda_m) = (1,1) ( λ e , λ m ) = ( 1 , 1 ) as the basic line operators. We denote the ’t Hooft line wrapped on S 1 S^1 S 1 by T T T , and the dyonic line wrapped on S 1 S^1 S 1 by D D D . The action of Z 2 ( c ) × Z 2 ( h ) \mathbb{Z}_2^{(c)} \times\mathbb{Z}_2^{(h)} Z 2 ( c ) × Z 2 ( h ) on these local operators is expressed as
c : D → − D , (9) c : D \to-D \, , \tag*{(9)} c : D → − D , (9)
h : T → − T , D → − D , (10) h : T \to-T \, , \qquad D \to-D \, , \tag*{(10)} h : T → − T , D → − D , (10)
which can be read off from the fact that Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) and Z 2 ( h ) \mathbb{Z}_2^{(h)} Z 2 ( h ) are respectively related to the electric and magnetic 1-form symmetries of the 4d theory. The action of Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) (which is present only at θ = n π \theta= n\pi θ = nπ for n ∈ Z n \in\mathbb{Z} n ∈ Z ) is more non-trivial. When θ = n π \theta= n\pi θ = nπ for even n n n , it just flips the sign of the electric charge, and therefore preserves ( T , D ) (T,D) ( T , D ) .3 However, when θ = n π \theta= n\pi θ = nπ for odd n n n , the CP also involves θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π . Since this induces a Witten effect [12 ] , the action of Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) is now expressed as
C P : T ↔ D (when θ = n π for odd n ) . (11) CP : T \leftrightarrow D \qquad\text{(when } \theta= n\pi\text{ for odd } n\text{)} \, . \tag*{(11)} C P : T ↔ D (when θ = nπ for odd n ) . (11)
A vacuum structure consistent with the above actions of c c c , h h h and C P CP C P was identified in [1 ] . In Figure 1 and Figure 2 , we reproduce the vacuum structure and phase diagram for temperature lower than the deconfining temperature.4 There are two vacua for generic values of θ \theta θ , but four vacua when θ = n π \theta=n\pi θ = nπ for odd n n n .
Figure 1. Region 1 is { θ ∈ R ∣ − π + 4 π n < θ < π + 4 π n \{\theta\in\mathbb{R}\mid-\pi+4\pi n<\theta<\pi+4\pi n { θ ∈ R ∣ − π + 4 π n < θ < π + 4 π n for n ∈ Z } n\in\mathbb{Z}\} n ∈ Z } and region 2 is { θ ∈ R ∣ π + 4 π n < θ < 3 π + 4 π n \{\theta\in\mathbb{R}\mid\pi+4\pi n<\theta<3\pi+4\pi n { θ ∈ R ∣ π + 4 π n < θ < 3 π + 4 π n for n ∈ Z } n\in\mathbb{Z}\} n ∈ Z } . There is a phase transition at θ = π + 2 π n \theta=\pi+2\pi n θ = π + 2 π n for n ∈ Z n\in\mathbb{Z} n ∈ Z .
Figure 2. The vacuum structure of the 3d mixed S U ( 2 ) SU(2) S U ( 2 ) gauge theory below the deconfining temperature. Regions 1 and 2 are specified in Figure 1 . Each dot stands for a single vacuum, and the axes represent the VEVs of T T T and D D D .
Note that the above is not the unique possible phase diagram consistent with the center/CP anomaly. In particular, it is based on an assumption that the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills is gapped for all values of θ \theta θ . For an alternative scenario without this assumption, see Sec. 7 of [1 ] . We here follow [1 ] and assume that (at zero temperature) the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory is gapped for all values of θ \theta θ and has a trivial non-degenerate vacuum at θ = 0 \theta=0 θ = 0 . Then the above phase diagram is the unique one consistent with the global symmetry.
Phase diagram of S U ( 2 ) SU(2) S U ( 2 ) and S O ( 3 ) ± SO(3)_{\pm} S O ( 3 ) ± Yang-Mills theories
If the above phase diagram of the mixed gauge theory is correct, one can now read off the phase diagram of the S U ( 2 ) / S O ( 3 ) ± SU(2)/SO(3)_{\pm} S U ( 2 ) / S O ( 3 ) ± Yang-Mills theories by discrete gaugings. Indeed, these Yang-Mills theories are obtained by the following Z 2 \mathbb{Z}_2 Z 2 -gaugings from the mixed S U ( 2 ) SU(2) S U ( 2 ) theory:
S U ( 2 ) = ( mixed S U ( 2 ) ) / Z 2 ( h ) , (12) SU(2) = \left(\text{mixed } SU(2)\right) / \mathbb{Z}_2^{(h)} \, , \tag*{(12)} S U ( 2 ) = ( mixed S U ( 2 ) ) / Z 2 ( h ) , (12)
S O ( 3 ) + = ( mixed S U ( 2 ) ) / Z 2 ( c ) , (13) SO(3)_+ = \left(\text{mixed } SU(2)\right) / \mathbb{Z}_2^{(c)} \, , \tag*{(13)} S O ( 3 ) + = ( mixed S U ( 2 ) ) / Z 2 ( c ) , (13)
S O ( 3 ) − = ( mixed S U ( 2 ) ) / Z 2 ( c ⋅ h ) , (14) SO(3)_- = \left(\text{mixed } SU(2)\right) / \mathbb{Z}_2^{(c\cdot h)} \, , \tag*{(14)} S O ( 3 ) − = ( mixed S U ( 2 ) ) / Z 2 ( c ⋅ h ) , (14)
where a quotient means a discrete gauging. Here, the relation (2.11) follows from the definition of the mixed gauge theory. The next one (2.12) follows from the fact that S O ( 3 ) SO(3) S O ( 3 ) gauge theory is obtained from S U ( 2 ) SU(2) S U ( 2 ) gauge theory by gauging its Z 2 \mathbb{Z}_2 Z 2 center symmetry. From the 3d viewpoint, this means to gauge the 3d 0-form Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) as well as the 3d 1-form Z 2 \mathbb{Z}_2 Z 2 symmetry of the compactified theory. Since the latter was already gauged in constructing the mixed gauge theory, we only need to gauge Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) of the mixed theory to obtain an S O ( 3 ) SO(3) S O ( 3 ) Yang-Mills theory. The difference between S O ( 3 ) + SO(3)_+ S O ( 3 ) + and S O ( 3 ) − SO(3)_- S O ( 3 ) − is the spectrum of genuine line operators, which is translated into the difference between (2.12) and (2.13).
Let us now look at the S U ( 2 ) SU(2) S U ( 2 ) theory and analyze its phase diagram using (2.11). Note that, since Z 2 ( h ) \mathbb{Z}_2^{(h)} Z 2 ( h ) acts on T T T and D D D as in Eq. (10) , neither T T T nor D D D is invariant under Z 2 ( h ) \mathbb{Z}_2^{(h)} Z 2 ( h ) . The gauge-invariant operator is instead the Wilson line
W = T D . (15) W = TD \, . \tag*{(15)} W = T D . (15)
The vacuum structure is read off from Figure 2 by identifying two vacua related by Z 2 ( h ) \mathbb{Z}_2^{(h)} Z 2 ( h ) as a single vacuum. We see that only a single vacuum with ⟨ W ⟩ = 0 \langle W\rangle= 0 ⟨ W ⟩ = 0 exists both in regions 1 and 2 of Figure 1 , which is in the confining phase in region 1, and in an oblique confining phase in region 2. These two phases are distinguished only by the Z 2 \mathbb{Z}_2 Z 2 center symmetry. Note that, when θ / π \theta/\pi θ / π is an odd integer, these two vacua exist simultaneously. When a background gauge field of the center symmetry is turned on, these two vacua can be distinguished and therefore the periodicity of θ \theta θ is θ ∼ θ + 4 π \theta\sim\theta+ 4\pi θ ∼ θ + 4 π . When the background gauge field is turned off, one cannot distinguish the two vacua in regions 1 and 2, and therefore the periodicity is θ ∼ θ + 2 π \theta\sim\theta+ 2\pi θ ∼ θ + 2 π . This result is summarized in the first line of Table 2 .
Table 2. The IR TQFT for various vacua of 4d S U ( 2 ) / S O ( 3 ) ± SU(2)/SO(3)_\pm S U ( 2 ) / S O ( 3 ) ± Yang-Mills theory
One can similarly read off the phase diagrams of the S O ( 3 ) + SO(3)_+ S O ( 3 ) + Yang-Mills theory using (2.12). Since Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) acts on T T T and D D D as in (2.8), the gauge-invariant operator is now T T T instead of W W W . The vacua of S O ( 3 ) + SO(3)_+ S O ( 3 ) + Yang-Mills theory is obtained by taking the Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) -quotient of the vacua shown in Figure 2 . We see that there are two vacua characterized by ⟨ T ⟩ ≠ 0 \langle T\rangle\ne0 ⟨ T ⟩ = 0 in region 1, but there is only one vacuum with ⟨ T ⟩ = 0 \langle T\rangle= 0 ⟨ T ⟩ = 0 in region 2 (Figure 3 ). Note that, in region 1, the Z 2 ( c ) \mathbb{Z}_2^{(c)} Z 2 ( c ) gauge symmetry does not couple to anything, and therefore there is an unbroken Z 2 \mathbb{Z}_2 Z 2 gauge theory in the infrared.5 In contrast, in region 2, the Z 2 \mathbb{Z}_2 Z 2 gauge symmetry is broken and therefore the single vacuum is in the trivially gapped phase. These results are summarized in the second line of Table 2 .
Figure 3. The vacuum structure of the S O ( 3 ) + SO(3)_+ S O ( 3 ) + Yang-Mills theory compactified on S 1 S^1 S 1 . The two vacua in region 1 corresponds to two different holonomies of the unbroken Z 2 \mathbb{Z}_2 Z 2 gauge symmetry along the S 1 S^1 S 1 . When going back to the original 4d theory, they are mapped to a single vacuum supporting the Z 2 \mathbb{Z}_2 Z 2 TQFT.
Finally, the phase digram of the S O ( 3 ) − SO(3)_- S O ( 3 ) − Yang-Mills theory is obtained exactly in the same way. The result is precisely such that
S O ( 3 ) ± θ + 2 π = S O ( 3 ) ∓ θ (16) SO(3)_{\pm}^{\theta+2\pi}=SO(3)_{\mp}^{\theta} \tag*{(16)} S O ( 3 ) ± θ + 2 π = S O ( 3 ) ∓ θ (16)
as expected. Namely, the roles of regions 1 and 2 are exchanged compared to S O ( 3 ) + SO(3)_+ S O ( 3 ) + , as shown in the third line of Table 2 .
Spin(4)/SO(4) Yang-Mills theory
In this section, we apply the technique of the mixed center/CP anomaly to the S p i n ( 4 ) / S O ( 4 ) ± Spin(4)/SO(4)_{\pm} S p in ( 4 ) / S O ( 4 ) ± Yang-Mills theories and identify their phase diagrams. Since S p i n ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) Spin(4) \simeq SU(2) \times SU(2) S p in ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) , the relevant center/CP anomaly can be read off from that of the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory.
We first consider the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) gauge theory and identify its phase diagram, from which we read off the phase diagram of the S p i n ( 4 ) / S O ( 4 ) ± Spin(4)/SO(4)_{\pm} S p in ( 4 ) / S O ( 4 ) ± Yang-Mills theories by discrete gaugings. We also show that the resulting phase diagrams are in perfect agreement with what are suggested by the softly broken N = 1 \mathcal{N}=1 N = 1 SYM.
Spin(4) gauge theory
We begin by considering S p i n ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) Spin(4) \simeq SU(2) \times SU(2) S p in ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) Yang-Mills theory on R 4 \mathbb{R}^{4} R 4 . Its action is
S = ∫ ∑ k = 1 2 Tr ( − 1 4 g k F ( k ) ∧ ∗ F ( k ) + i θ k 8 π 2 F ( k ) ∧ F ( k ) ) , (17) S = \int\sum_{k=1}^{2} \operatorname{Tr}\left(-\frac{1}{4g_k}F^{(k)} \wedge\ast F^{(k)} + \frac{i\theta_k}{8\pi^2}F^{(k)} \wedge F^{(k)}\right), \tag*{(17)} S = ∫ k = 1 ∑ 2 Tr ( − 4 g k 1 F ( k ) ∧ ∗ F ( k ) + 8 π 2 i θ k F ( k ) ∧ F ( k ) ) , (17)
where F ( 1 ) F^{(1)} F ( 1 ) and F ( 2 ) F^{(2)} F ( 2 ) are the field strengths of two S U ( 2 ) SU(2) S U ( 2 ) gauge fields, and g 1 g_1 g 1 and g 2 g_2 g 2 are Yang-Mills couplings. The parameters θ k \theta_k θ k (k = 1 , 2 k = 1, 2 k = 1 , 2 ) are periodic θ k ∼ θ k + 2 π \theta_k \sim\theta_k + 2\pi θ k ∼ θ k + 2 π unless a background gauge field of the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry is turned on. The CP transformation flips the signs of these θ \theta θ -parameters simultaneously as
C P : θ k → − θ k ( k = 1 , 2 ) , (18) CP : \theta_k \to-\theta_k \quad(k = 1, 2), \tag*{(18)} C P : θ k → − θ k ( k = 1 , 2 ) , (18)
and therefore the CP is a symmetry only at ( θ 1 , θ 2 ) = ( n 1 π , n 2 π ) (\theta_1,\theta_2) = (n_1\pi,n_2\pi) ( θ 1 , θ 2 ) = ( n 1 π , n 2 π ) for n 1 , n 2 ∈ N n_1,n_2 \in\mathbb{N} n 1 , n 2 ∈ N .
When ( θ 1 , θ 2 ) = ( n 1 π , n 2 π ) (\theta_1,\theta_2) = (n_1\pi,n_2\pi) ( θ 1 , θ 2 ) = ( n 1 π , n 2 π ) for n k ∈ N n_k \in\mathbb{N} n k ∈ N and at least one of n k n_k n k is odd, the theory has a mixed center/CP anomaly. Specifically, in the presence of the background gauge field ( B 1 , B 2 ) (B_1,B_2) ( B 1 , B 2 ) of the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry, shifting the θ \theta θ -parameters as θ k → θ k + 2 π n k \theta_k \to\theta_k + 2\pi n_k θ k → θ k + 2 π n k induces an extra term
Δ S = i π 2 ∑ k = 1 2 n k ∫ M P ( B k ) (19) \Delta S = \frac{i\pi}{2}\sum_{k=1}^{2} n_k \int_{\mathcal{M}} \mathcal{P}(B_k) \tag*{(19)} Δ S = 2 iπ k = 1 ∑ 2 n k ∫ M P ( B k ) (19)
in the action. When M \mathcal{M} M is a spin manifold, the integrals ∫ M P ( B k ) \int_{\mathcal{M}} \mathcal{P}(B_k) ∫ M P ( B k ) are even integers and therefore (3.3) is expressed as
Δ S = i π ∑ k = 1 2 ℓ k (20) \Delta S = i\pi\sum_{k=1}^{2} \ell_k \tag*{(20)} Δ S = iπ k = 1 ∑ 2 ℓ k (20)
for some ℓ k ∈ Z \ell_k \in\mathbb{Z} ℓ k ∈ Z . If n 1 n_1 n 1 or n 2 n_2 n 2 is odd, the sum ∑ k = 1 2 ℓ k \sum_{k=1}^{2}\ell_k ∑ k = 1 2 ℓ k can be odd depending on the values of B k B_k B k . In that case, the shift θ k → θ k + 2 π n k \theta_k \to\theta_k + 2\pi n_k θ k → θ k + 2 π n k does not preserve the theory. This means that the CP at ( n 1 π , n 2 π ) (n_1\pi,n_2\pi) ( n 1 π , n 2 π ) is broken in the presence of non-vanishing B k B_k B k , signaling the the mixed center/CP anomaly at these values of θ k \theta_k θ k . This also implies that, when generic background gauge fields ( B 1 , B 2 ) (B_1,B_2) ( B 1 , B 2 ) are turned on, the periodicity of θ k \theta_k θ k is modified to be
θ k ∼ θ k + 4 π ( k = 1 , 2 ) . (21) \theta_k \sim\theta_k + 4\pi\quad(k=1,2). \tag*{(21)} θ k ∼ θ k + 4 π ( k = 1 , 2 ) . (21)
3d mixed Spin(4) theory and its phase diagram
When compactifying the 4d S p i n ( 4 ) Spin(4) S p in ( 4 ) theory on S 1 S^1 S 1 , the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry splits into a 0-form and a 1-form Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 symmetry in three dimensions. By gauging the 3d 1-form Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 symmetry, we obtain the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) gauge theory.
The mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory is a 3d theory with at least a 0-form ( Z 2 ) 4 (\mathbb{Z}_2)^4 ( Z 2 ) 4 symmetry. We denote by Z 2 ( c 1 ) × Z 2 ( c 2 ) \mathbb{Z}_2^{(c_1)} \times\mathbb{Z}_2^{(c_2)} Z 2 ( c 1 ) × Z 2 ( c 2 ) the 3d 0-form symmetry arising from the 4d center symmetry upon compactification, and by Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( h 1 ) × Z 2 ( h 2 ) the 0-form symmetry obtained by gauging the 3d 1-form symmetry. All these discrete symmetries act on the space of 3d local operators arising from 4d line operators wrapping on S 1 S^1 S 1 . We denote ’t Hooft lines wrapped on S 1 S^1 S 1 by T 1 T_1 T 1 and T 2 T_2 T 2 where the sub-script represents under which of the two S U ( 2 ) SU(2) S U ( 2 ) factors in S p i n ( 4 ) ≃ S U ( 2 ) 1 × S U ( 2 ) 2 Spin(4) \simeq SU(2)_1 \times SU(2)_2 S p in ( 4 ) ≃ S U ( 2 ) 1 × S U ( 2 ) 2 they are charged magnetically. Similarly, we denote by D k D_k D k dyonic lines wrapped on S 1 S^1 S 1 . Then the action of Z 2 ( c 1 ) × Z 2 ( h 1 ) × Z 2 ( c 2 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(c_1)} \times\mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(c_2)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( c 1 ) × Z 2 ( h 1 ) × Z 2 ( c 2 ) × Z 2 ( h 2 ) on them is expressed as
c k : D k → − D k , h k : T k → − T k , D k → − D k , (22) c_k:D_k \to-D_k,\qquad h_k:T_k \to-T_k,\ D_k \to-D_k, \tag*{(22)} c k : D k → − D k , h k : T k → − T k , D k → − D k , (22)
where c 1 c_1 c 1 , c 2 c_2 c 2 , h 1 h_1 h 1 and h 2 h_2 h 2 are the generators of Z 2 ( c 1 ) \mathbb{Z}_2^{(c_1)} Z 2 ( c 1 ) , Z 2 ( c 2 ) \mathbb{Z}_2^{(c_2)} Z 2 ( c 2 ) , Z 2 ( h 1 ) \mathbb{Z}_2^{(h_1)} Z 2 ( h 1 ) and Z 2 ( h 2 ) \mathbb{Z}_2^{(h_2)} Z 2 ( h 2 ) , respectively.
The full global symmetry of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory depends on the values of θ 1 \theta_1 θ 1 and θ 2 \theta_2 θ 2 . When θ k / π ∉ Z \theta_k/\pi\notin\mathbb{Z} θ k / π ∈ / Z for some k k k , the 4d classical Lagrangian is not CP-invariant, and therefore the above ( Z 2 ) 4 (\mathbb{Z}_2)^4 ( Z 2 ) 4 symmetry is the full global symmetry of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory. When θ k / π ∈ Z \theta_k/\pi\in\mathbb{Z} θ k / π ∈ Z for both k = 1 k=1 k = 1 and k = 2 k=2 k = 2 , the mixed theory has an additional Z 2 \mathbb{Z}_2 Z 2 symmetry corresponding to the 4d CP symmetry. We denote its generator by C P CP C P . When θ k / π \theta_k/\pi θ k / π is an odd integer, the C P CP C P exchanges T k T_k T k and D k D_k D k , i.e.,
T k ↔ D k . (23) T_k \leftrightarrow D_k. \tag*{(23)} T k ↔ D k . (23)
In the presence of Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) , the full global symmetry is a Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) -extension of Z 2 ( c 1 ) × Z 2 ( c 2 ) × Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(c_1)} \times\mathbb{Z}_2^{(c_2)} \times\mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( c 1 ) × Z 2 ( c 2 ) × Z 2 ( h 1 ) × Z 2 ( h 2 ) , as shown in Table 3 . In particular, when both θ 1 / π \theta_1/\pi θ 1 / π and θ 2 / π \theta_2/\pi θ 2 / π are odd integers, the full global symmetry is ( Z 2 ) 4 ⋊ Z 2 (\mathbb{Z}_2)^4 \rtimes\mathbb{Z}_2 ( Z 2 ) 4 ⋊ Z 2 such that
Table 3. The global symmetry of the 3d mixed S U ( 2 ) × S U ( 2 ) SU(2) \times SU(2) S U ( 2 ) × S U ( 2 ) theory. These are all 0-form symmetries.
C P ⋅ c k ⋅ C P = c k ⋅ h k , C P ⋅ h k = h k ⋅ C P , c k ⋅ h ℓ = h ℓ ⋅ c k , c k 2 = h k 2 = C P 2 = 1. (24) CP \cdot c_k \cdot CP = c_k \cdot h_k,\qquad CP \cdot h_k = h_k \cdot CP,\qquad c_k \cdot h_\ell= h_\ell\cdot c_k,\qquad c_k^2 = h_k^2 = CP^2 = 1. \tag*{(24)} C P ⋅ c k ⋅ C P = c k ⋅ h k , C P ⋅ h k = h k ⋅ C P , c k ⋅ h ℓ = h ℓ ⋅ c k , c k 2 = h k 2 = C P 2 = 1. (24)
Here, the first two equations show that the semi-direct product is associated with the most non-trivial homomorphism from Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) to Aut ( ( Z 2 ) 4 ) \operatorname{Aut}((\mathbb{Z}_2)^4) Aut (( Z 2 ) 4 ) .6
As in the case of the S U ( 2 ) SU(2) S U ( 2 ) Yang-Mills theory reviewed in § 2 , one can find a phase diagram of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory that is consistent with the global symmetry groups. We assume that at zero temperature the (ordinary) S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory is gapped for all values of θ k \theta_k θ k and has a trivial non-degenerate vacuum at θ k = 0 \theta_k = 0 θ k = 0 , which is the same assumption as in most sections of [1 ] . Then the phase diagram of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory at zero temperature is described as in Figure 4 and Table 4 .
Table 4. The VEVs of T k T_k T k and D k D_k D k in various regions in Figure 4 , where a k a_k a k and b k b_k b k are non-vanishing values. The signs are not correlated. There are four vacua in each region.
The four phases of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory shown in Figure 4 are characterized by the VEVs of T k T_k T k and D k D_k D k , as described in Table 4 . For instance, in region 1, we have ⟨ D k ⟩ = 0 \langle D_k \rangle= 0 ⟨ D k ⟩ = 0 and ⟨ T k ⟩ ≠ 0 \langle T_k \rangle\ne0 ⟨ T k ⟩ = 0 for both k = 1 , 2 k = 1, 2 k = 1 , 2 . Depending on the sign of ⟨ T 1 ⟩ \langle T_1 \rangle ⟨ T 1 ⟩ and ⟨ T 2 ⟩ \langle T_2 \rangle ⟨ T 2 ⟩ , there are four different vacua in region 1. This will turn out to be necessary and sufficient for the ordinary S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory to have a unique gapped confining (or oblique confining) vacuum for generic values of θ k \theta_k θ k .
The above four phases are also characterized by which sub-group of Z 2 ( c 1 ) × Z 2 ( h 1 ) × Z 2 ( c 2 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(c_1)} \times\mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(c_2)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( c 1 ) × Z 2 ( h 1 ) × Z 2 ( c 2 ) × Z 2 ( h 2 ) is unbroken, as shown in Table 5 . From this, we see that the phase transitions of the mixed Spin (4) theory (that occur when crossing dashed lines in Figure 4 ) are all conventional phase transitions rather than SPT phase transitions.
Table 5. The unbroken global symmetry of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) gauge theory in various regions in Figure 4 . Here, Z 2 ( g ) \mathbb{Z}_2^{(g)} Z 2 ( g ) stands for a Z 2 \mathbb{Z}_2 Z 2 generated by g g g .
Along each dashed line in Figure 4 , two regions are attached to each other. When the θ \theta θ -parameters are on such a line, the vacua present in these two regions exist simultaneously, leading to eight vacua. The most non-trivial situation occurs where two dashed lines intersect, i.e., when both θ 1 / π \theta_1/\pi θ 1 / π and θ 2 / π \theta_2/\pi θ 2 / π are odd integers. In this case, the mixed Spin (4) theory has sixteen vacua such that
( T 1 , D 1 ; T 2 , D 2 ) = ( ± a 1 , 0 ; ± a 2 , 0 ) , ( 0 , ± a 1 ; 0 , ± a 2 ) , ( ± a 1 , 0 ; 0 , ± a 2 ) , ( 0 , ± a 1 ; ± a 2 , 0 ) (25) (T_1,D_1;T_2,D_2) = (\pm a_1,0;\pm a_2,0),\ (0,\pm a_1;0,\pm a_2),\ (\pm a_1,0;0,\pm a_2),\ (0,\pm a_1;\pm a_2,0) \tag*{(25)} ( T 1 , D 1 ; T 2 , D 2 ) = ( ± a 1 , 0 ; ± a 2 , 0 ) , ( 0 , ± a 1 ; 0 , ± a 2 ) , ( ± a 1 , 0 ; 0 , ± a 2 ) , ( 0 , ± a 1 ; ± a 2 , 0 ) (25)
where a 1 a_1 a 1 and a 2 a_2 a 2 are a positive real number determined by the temperature. Note that these sixteen vacua belong to two orbits of ( Z 2 ) 4 ⋊ Z 2 (\mathbb{Z}_2)^4 \rtimes\mathbb{Z}_2 ( Z 2 ) 4 ⋊ Z 2 .
Phase diagram of Spin(4) Yang-Mills theory
The S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory is obtained from the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory by gauging its Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( h 1 ) × Z 2 ( h 2 ) symmetry, i.e.,
S p i n ( 4 ) = mixed S p i n ( 4 ) Z 2 ( h 1 ) × Z 2 ( h 2 ) . (26) Spin(4) = \frac{\text{mixed }Spin(4)}{\mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)}} . \tag*{(26)} S p in ( 4 ) = Z 2 ( h 1 ) × Z 2 ( h 2 ) mixed S p in ( 4 ) . (26)
The action of Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( h 1 ) × Z 2 ( h 2 ) on T k T_k T k and D k D_k D k are shown in (22) . At generic points in regions 1 ∼ \sim ∼ 4, gauging Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( h 1 ) × Z 2 ( h 2 ) reduces four vacua into a single vacuum. Along the dashed line in Figure 4 , eight vacua of the mixed gauge theory are reduced to two vacua. This can be seen by analyzing how Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)} Z 2 ( h 1 ) × Z 2 ( h 2 ) acts on these eight vacua. Finally, at intersections of two dashed lines, the sixteen vacua of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory reduce to four vacua in the (ordinary) S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory.
The order parameters of the S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory are the Wilson lines
W k ≡ T k D k (27) W_k \equiv T_kD_k \tag*{(27)} W k ≡ T k D k (27)
for k = 1 , 2 k = 1, 2 k = 1 , 2 . The above various vacua all have ⟨ W k ⟩ = 0 \langle W_k\rangle= 0 ⟨ W k ⟩ = 0 for k = 1 , 2 k = 1, 2 k = 1 , 2 , and therefore they are in the confining or an oblique confining phase. In the convention that region 1 corresponds to the confining phase (trivial phase), the other three regions correspond to oblique confining phases, i.e., SPT phases protected by the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry [14 ] . This result is summarized in the first line of Table 6 . Thus, the conventional phase transitions of the mixed Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) theory are now all mapped to SPT phase transitions of the (ordinary) Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) Yang-Mills theory.
Phase diagram of SO(4) Yang-Mills theory
Let us now turn to the SO ( 4 ) ± \operatorname{SO}(4)_{\pm} SO ( 4 ) ± Yang-Mills theory. These theories are obtained from the mixed Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) theory by gauging the Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) or Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2\cdot h_1)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) symmetry, i.e.,$7 $
SO ( 4 ) + = ( mixed Spin ( 4 ) ) / ( Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) ) , (28) \operatorname{SO}(4)_+ = \left(\text{mixed }\operatorname{Spin}(4)\right) \big/ \left(\mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)}\right), \tag*{(28)} SO ( 4 ) + = ( mixed Spin ( 4 ) ) / ( Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) ) , (28)
SO ( 4 ) − = ( mixed Spin ( 4 ) ) / ( Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) ) . (29) \operatorname{SO}(4)_- = \left(\text{mixed }\operatorname{Spin}(4)\right) \big/ \left(\mathbb{Z}_2^{(c_1\cdot c_2\cdot h_1)} \times\mathbb{Z}_2^{(h_1\cdot h_2)}\right). \tag*{(29)} SO ( 4 ) − = ( mixed Spin ( 4 ) ) / ( Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) ) . (29)
From the viewpoint of the Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) Yang-Mills theory, these discrete gauging correspond to gauging the diagonal Z 2 \mathbb{Z}_2 Z 2 sub-group of the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry. As a result, the SO ( 4 ) ± \operatorname{SO}(4)_{\pm} SO ( 4 ) ± Yang-Mills theory has an electric 1-form Z 2 \mathbb{Z}_2 Z 2 symmetry and a magnetic 1-form Z 2 \mathbb{Z}_2 Z 2 symmetry.
SO(4)+
Let us first study the phase diagram of the SO ( 4 ) + \operatorname{SO}(4)_+ SO ( 4 ) + theory. Since T k T_k T k and D k D_k D k transform under c 1 ⋅ c 2 c_1 \cdot c_2 c 1 ⋅ c 2 and h 1 ⋅ h 2 h_1 \cdot h_2 h 1 ⋅ h 2 as
c 1 ⋅ c 2 : D 1 → − D 1 , D 2 → − D 2 , (30) c_1 \cdot c_2:\quad D_1 \to-D_1,\quad D_2 \to-D_2, \tag*{(30)} c 1 ⋅ c 2 : D 1 → − D 1 , D 2 → − D 2 , (30)
h 1 ⋅ h 2 : T 1 → − T 1 , T 2 → − T 2 , D 1 → − D 1 , D 2 → − D 2 , (31) h_1 \cdot h_2:\quad T_1 \to-T_1,\quad T_2 \to-T_2,\quad D_1 \to-D_1,\quad D_2 \to-D_2, \tag*{(31)} h 1 ⋅ h 2 : T 1 → − T 1 , T 2 → − T 2 , D 1 → − D 1 , D 2 → − D 2 , (31)
gauging Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) removes T k T_k T k and D k D_k D k from the spectrum, leaving the gauge-invariant combinations
T 1 T 2 , D 1 D 2 . (32) T_1T_2,\quad D_1D_2. \tag*{(32)} T 1 T 2 , D 1 D 2 . (32)
The VEVs of these two gauge-invariant operators are the order parameters of the SO ( 4 ) + \operatorname{SO}(4)_+ SO ( 4 ) + Yang-Mills theory. In the case ⟨ T 1 T 2 ⟩ ≠ 0 \langle T_1T_2\rangle\ne0 ⟨ T 1 T 2 ⟩ = 0 the magnetic 1-form Z 2 \mathbb{Z}_2 Z 2 is broken, and in the case ⟨ D 1 D 2 ⟩ ≠ 0 \langle D_1D_2\rangle\ne0 ⟨ D 1 D 2 ⟩ = 0 the electric and magnetic Z 2 \mathbb{Z}_2 Z 2 are both broken.
When the θ \theta θ -parameters are in region 3 or 4 of Figure 4 , the Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) freely acts on the four vacua of the mixed Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) theory, and therefore gauging it leads to a single vacuum of the SO ( 4 ) + \operatorname{SO}(4)_+ SO ( 4 ) + theory, as shown in the rightmost picture of Figure 5 . These are in the trivial and SPT phases protected by the 1-form symmetry. On the other hand, when the θ \theta θ -parameters are in region 1 or 2, there exists a Z 2 \mathbb{Z}_2 Z 2 sub-group of Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) that trivially acts on the four vacua. In this case, gauging Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) gives rise to two degenerate vacua accompanied with an unbroken Z 2 \mathbb{Z}_2 Z 2 TQFT (the left and middle pictures of Figure 5 ). These results are summarized in the second line of Table 6.
Figure 5. The vacuum structure of S O ( 4 ) + SO(4)_+ S O ( 4 ) + Yang-Mills theory compactified on S 1 S^1 S 1 for generic values of the θ \theta θ -parameters. Two vacua in each of regions 1 and 2 are associated with different Z 2 \mathbb{Z}_2 Z 2 holonomies along S 1 S^1 S 1 . When going back to four dimensions, they are mapped to a single vacuum that supports a Z 2 \mathbb{Z}_2 Z 2 TQFT.
When the θ \theta θ -parameters are on a dashed line in Figure 4 , the S O ( 4 ) + SO(4)_+ S O ( 4 ) + Yang-Mills theory has three degenerate vacua as shown in the left two pictures of Figure 6 ; two of the three are accompanied by an unbroken Z 2 \mathbb{Z}_2 Z 2 TQFT and the remaining one is in the confining/oblique confining phase. When the θ \theta θ -parameters are at an intersection of dashed lines, the S O ( 4 ) + SO(4)_+ S O ( 4 ) + Yang-Mills theory has six vacua as shown in the rightmost picture of Figure 6 , where two vacua are located at ⟨ T 1 T 2 ⟩ = ⟨ D 1 D 2 ⟩ = 0 \langle T_1T_2\rangle= \langle D_1D_2\rangle= 0 ⟨ T 1 T 2 ⟩ = ⟨ D 1 D 2 ⟩ = 0 .
Figure 6. The vacuum structure of S O ( 4 ) + SO(4)_+ S O ( 4 ) + theory compactified on S 1 S^1 S 1 when the θ \theta θ -parameters are at boundaries of different regions. There is a single vacuum at the origin ⟨ T 1 T 2 ⟩ = ⟨ D 1 D 2 ⟩ = 0 \langle T_1T_2\rangle= \langle D_1D_2\rangle= 0 ⟨ T 1 T 2 ⟩ = ⟨ D 1 D 2 ⟩ = 0 in the left two pictures, while there are two degenerate vacua at the origin in the rightmost picture. Every pair of vacua connected by T 1 T 2 → − T 1 T 2 T_1T_2 \to-T_1T_2 T 1 T 2 → − T 1 T 2 and D 1 D 2 → − D 1 D 2 D_1D_2 \to-D_1D_2 D 1 D 2 → − D 1 D 2 is associated with different Z 2 \mathbb{Z}_2 Z 2 holonomies along the S 1 S^1 S 1 , and therefore corresponds to a single vacuum in the original 4d theory.
S O ( 4 ) − SO(4)_- S O ( 4 ) −
The phase diagram of the S O ( 4 ) − SO(4)_- S O ( 4 ) − theory is obtained similarly. We now gauge the Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_2^{(c_1\cdot c_2\cdot h_1)} \times\mathbb{Z}_2^{(h_1\cdot h_2)} Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) symmetry of the mixed S p i n ( 4 ) Spin(4) S p in ( 4 ) theory, whose action is characterized by
c 1 ⋅ c 2 ⋅ h 1 : T 1 → − T 1 , D 2 → − D 2 , (33) \begin{aligned}
c_1 \cdot c_2 \cdot h_1 &: \quad T_1 \to-T_1,\quad D_2 \to-D_2 \, , \tag*{(33)}
\end{aligned} c 1 ⋅ c 2 ⋅ h 1 : T 1 → − T 1 , D 2 → − D 2 , (33)
h 1 ⋅ h 2 : T 1 → − T 1 , T 2 → − T 2 , D 1 → − D 1 , D 2 → − D 2 . (34) \begin{aligned}
h_1 \cdot h_2 &: \quad T_1 \to-T_1,\quad T_2 \to-T_2,\quad D_1 \to-D_1,\quad D_2 \to-D_2 \, . \tag*{(34)}
\end{aligned} h 1 ⋅ h 2 : T 1 → − T 1 , T 2 → − T 2 , D 1 → − D 1 , D 2 → − D 2 . (34)
Since this can be simply obtained from (30) and (31) by T 1 ↔ D 1 T_1 \leftrightarrow D_1 T 1 ↔ D 1 , the phase diagram of the S O ( 4 ) − SO(4)_- S O ( 4 ) − Yang-Mills theory is simply obtained by exchanging the roles of T 1 T_1 T 1 and D 1 D_1 D 1 in our discussion of S O ( 4 ) + SO(4)_+ S O ( 4 ) + . In particular, the gauge-invariant order parameters are now the VEVs of
T 1 D 2 , D 1 T 2 . (35) T_1D_2 \,,\qquad D_1T_2 \, . \tag*{(35)} T 1 D 2 , D 1 T 2 . (35)
The resulting phase structure is summarized in the third line of Table 6 . We see that these results are consistent with the relation (1) .
Comparison to soft SUSY breaking
In this sub-section, we study the phase diagrams of supersymmetric Spin ( 4 ) / S O ( 4 ) ± \operatorname{Spin}(4)/SO(4)_\pm Spin ( 4 ) / S O ( 4 ) ± Yang-Mills theories with a small gaugino mass m m m turned on in the small ∣ m ∣ |m| ∣ m ∣ approximation. Since the non-supersymmetric Yang-Mills is regarded as the m → ∞ m \to\infty m → ∞ limit of this theory, the phase diagram in this approximation does not necessarily coincide with the diagrams we obtained above. However, we will find the two diagrams in perfect agreement with each other, which suggests that the phase diagram is independent of m m m . A similar analysis for the S U ( N ) SU(N) S U ( N ) Yang-Mills theory was performed in [1 ] .
N = 1 \mathcal{N}=1 N = 1 Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) SYM
Let us start with the N = 1 \mathcal{N}=1 N = 1 Spin ( 4 ) \operatorname{Spin}(4) Spin ( 4 ) SYM, which has four supersymmetric vacua labeled by two parameters ϵ 1 \epsilon_1 ϵ 1 and ϵ 2 \epsilon_2 ϵ 2 taking values in { + 1 , − 1 } \{+1,-1\} { + 1 , − 1 } [15 , 16 ] . These four vacua are confining and oblique confining vacua, protected by the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry. We take the convention that the vacuum for ϵ 1 = ϵ 2 = 1 \epsilon_1=\epsilon_2=1 ϵ 1 = ϵ 2 = 1 corresponds to the confining phase (i.e., the trivial phase), and the other vacua correspond to oblique confining phases.
When we introduce a soft SUSY breaking mass term, the degeneracy of the four vacua is resolved. To see this, let us introduce a small gaugino mass term
δ L = m ( Tr ( η α η α ) + Tr ( ξ α ξ α ) ) + c.c. , (36) \delta\mathcal{L}=m\left(\operatorname{Tr}(\eta_\alpha\eta^\alpha)+\operatorname{Tr}(\xi_\alpha\xi^\alpha)\right)+\text{c.c.}, \tag*{(36)} δ L = m ( Tr ( η α η α ) + Tr ( ξ α ξ α ) ) + c.c. , (36)
where ( η α , ξ α ) (\eta_\alpha,\xi_\alpha) ( η α , ξ α ) are gaugino fields in the adjoint representation of Spin ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) \operatorname{Spin}(4)\simeq SU(2)\times SU(2) Spin ( 4 ) ≃ S U ( 2 ) × S U ( 2 ) . The gaugino condensation is characterized by
⟨ Tr ( η α η α ) ⟩ = ϵ 1 κ 1 Λ 1 3 , ⟨ Tr ( ξ α ξ α ) ⟩ = ϵ 2 κ 2 Λ 2 3 , (37) \left\langle\operatorname{Tr}(\eta_\alpha\eta^\alpha)\right\rangle=\epsilon_1\kappa_1\Lambda_1^3,\qquad\left\langle\operatorname{Tr}(\xi_\alpha\xi^\alpha)\right\rangle=\epsilon_2\kappa_2\Lambda_2^3, \tag*{(37)} ⟨ Tr ( η α η α ) ⟩ = ϵ 1 κ 1 Λ 1 3 , ⟨ Tr ( ξ α ξ α ) ⟩ = ϵ 2 κ 2 Λ 2 3 , (37)
where κ 1 \kappa_1 κ 1 and κ 2 \kappa_2 κ 2 are non-vanishing complex constants, and Λ 1 \Lambda_1 Λ 1 and Λ 2 \Lambda_2 Λ 2 are two dynamical scales so that Λ k 6 = ∣ Λ k ∣ 6 e i θ k \Lambda_k^6=\lvert\Lambda_k\rvert^6e^{i\theta_k} Λ k 6 = ∣ Λ k ∣ 6 e i θ k . We take the origins of θ k \theta_k θ k so that m κ 1 m\kappa_1 m κ 1 and m κ 2 m\kappa_2 m κ 2 are positive real numbers.
When ∣ m ∣ \lvert m\rvert ∣ m ∣ is sufficiently small, the effective potential is evaluated as V = − m ( ⟨ Tr ( η α η α ) ⟩ + ⟨ Tr ( ξ α ξ α ) ⟩ ) + c.c. V=-m(\langle\operatorname{Tr}(\eta_\alpha\eta^\alpha)\rangle+\langle\operatorname{Tr}(\xi_\alpha\xi^\alpha)\rangle)+\text{c.c.} V = − m (⟨ Tr ( η α η α )⟩ + ⟨ Tr ( ξ α ξ α )⟩) + c.c. Using (37) and Λ k 6 = ∣ Λ k ∣ 6 e i θ k \Lambda_k^6=\lvert\Lambda_k\rvert^6e^{i\theta_k} Λ k 6 = ∣ Λ k ∣ 6 e i θ k , one can rewrite it as
V = − 2 ∑ k = 1 2 Re ( ϵ k ∣ m κ k Λ k 3 ∣ e i θ k / 2 ) = − 2 ∑ k = 1 2 ϵ k ∣ m κ k Λ k 3 ∣ cos θ k 2 , (38) V=-2\sum_{k=1}^{2}\operatorname{Re}\left(\epsilon_k\lvert m\kappa_k\Lambda_k^3\rvert e^{i\theta_k/2}\right)=-2\sum_{k=1}^{2}\epsilon_k\lvert m\kappa_k\Lambda_k^3\rvert\cos\frac{\theta_k}{2}, \tag*{(38)} V = − 2 k = 1 ∑ 2 Re ( ϵ k ∣ m κ k Λ k 3 ∣ e i θ k /2 ) = − 2 k = 1 ∑ 2 ϵ k ∣ m κ k Λ k 3 ∣ cos 2 θ k , (38)
which is periodic under θ k → θ k + 4 π \theta_k\to\theta_k+4\pi θ k → θ k + 4 π for each of k = 1 , 2 k=1,2 k = 1 , 2 . This effective potential implies that four vacua corresponding to ( ϵ 1 , ϵ 2 ) = ( + 1 , + 1 ) (\epsilon_1,\epsilon_2)=(+1,+1) ( ϵ 1 , ϵ 2 ) = ( + 1 , + 1 ) , ( + 1 , − 1 ) (+1,-1) ( + 1 , − 1 ) , ( − 1 , + 1 ) (-1,+1) ( − 1 , + 1 ) and ( − 1 , − 1 ) (-1,-1) ( − 1 , − 1 ) now have different vacuum energies. For generic values of the θ \theta θ -parameters, there is only one true vacuum with the minimum energy.8
There are special values of the θ \theta θ -parameters which give rise to degenerate vacua. For instance, when θ 1 / π \theta_1/\pi θ 1 / π is an odd integer but θ 2 / π \theta_2/\pi θ 2 / π is not, the effective potential (38) reduces to
V = − 2 ϵ 2 ∣ m κ 2 Λ 2 3 ∣ cos θ 2 2 , (39) V=-2\epsilon_2\lvert m\kappa_2\Lambda_2^3\rvert\cos\frac{\theta_2}{2}, \tag*{(39)} V = − 2 ϵ 2 ∣ m κ 2 Λ 2 3 ∣ cos 2 θ 2 , (39)
which implies two degenerate vacua corresponding to ϵ 1 = ± 1 \epsilon_1=\pm1 ϵ 1 = ± 1 .9 The same situation occurs when θ 2 / π \theta_2/\pi θ 2 / π is an odd integer but θ 1 / π \theta_1/\pi θ 1 / π is not. The most non-trivial situation occurs when both θ 1 / π \theta_1/\pi θ 1 / π and θ 2 / π \theta_2/\pi θ 2 / π are odd integers, in which case the effective potential vanishes:
V = 0. (40) V=0. \tag*{(40)} V = 0. (40)
Then the four vacua corresponding to ( ϵ 1 , ϵ 2 ) = ( 1 , 1 ) , ( 1 , − 1 ) , ( − 1 , 1 ) (\epsilon_1,\epsilon_2)=(1,1),(1,-1),(-1,1) ( ϵ 1 , ϵ 2 ) = ( 1 , 1 ) , ( 1 , − 1 ) , ( − 1 , 1 ) and ( − 1 , − 1 ) (-1,-1) ( − 1 , − 1 ) are all degenerate even after a small gaugino mass m m m is turned on.
The above results in the small ∣ m ∣ |m| ∣ m ∣ approximation are in perfect agreement with the phase diagram of the Spin(4) Yang-Mills theory that we have identified using the center/CP anomaly in § 3.3 .
N = 1 \mathcal{N}=1 N = 1 S O ( 4 ) ± SO(4)_{\pm} S O ( 4 ) ± SYM
We now move on to the soft SUSY breaking of the S O ( 4 ) ± SO(4)_{\pm} S O ( 4 ) ± SYM. In this case, the supersymmetric theory still has four vacua corresponding to ϵ 1 , ϵ 2 = ± 1 \epsilon_1,\epsilon_2=\pm1 ϵ 1 , ϵ 2 = ± 1 , but the IR degrees of freedom present in these vacua are different from the Spin(4) case. Indeed, for the S O ( 4 ) − SO(4)_{-} S O ( 4 ) − SYM, the vacua for ϵ 1 = ϵ 2 = ± 1 \epsilon_1=\epsilon_2=\pm1 ϵ 1 = ϵ 2 = ± 1 are confining and oblique confining phases, while those for ϵ 1 = − ϵ 2 = ± 1 \epsilon_1=-\epsilon_2=\pm1 ϵ 1 = − ϵ 2 = ± 1 support a Z 2 \mathbb{Z}_2 Z 2 TQFT. For the S O ( 4 ) + SO(4)_{+} S O ( 4 ) + SYM, the roles of ϵ 1 = ϵ 2 \epsilon_1=\epsilon_2 ϵ 1 = ϵ 2 and ϵ 1 = − ϵ 2 \epsilon_1=-\epsilon_2 ϵ 1 = − ϵ 2 are exchanged.
This means that, when θ 1 , θ 2 ∈ ( π , 3 π ) \theta_1,\theta_2\in(\pi,3\pi) θ 1 , θ 2 ∈ ( π , 3 π ) or θ 1 , θ 2 ∈ [ 0 , π ) ∪ ( 3 π , 4 π ) \theta_1,\theta_2\in[0,\pi)\cup(3\pi,4\pi) θ 1 , θ 2 ∈ [ 0 , π ) ∪ ( 3 π , 4 π ) , the S O ( 4 ) − SO(4)_{-} S O ( 4 ) − theory is in the confining or oblique confining phase. On the other hand, when one of θ 1 \theta_1 θ 1 and θ 2 \theta_2 θ 2 is in ( π , 3 π ) (\pi,3\pi) ( π , 3 π ) and the other is in [ 0 , π ) ∪ ( 3 π , 4 π ) [0,\pi)\cup(3\pi,4\pi) [ 0 , π ) ∪ ( 3 π , 4 π ) , the IR degrees of freedom in the S O ( 4 ) − SO(4)_{-} S O ( 4 ) − Yang-Mills theory is a Z 2 \mathbb{Z}_2 Z 2 TQFT. The result for S O ( 4 ) + SO(4)_{+} S O ( 4 ) + is obtained similarly, which will turn out to be consistent with the relation S O ( 4 ) + θ 1 , θ 2 = S O ( 4 ) − θ 1 + 2 π , θ 2 SO(4)_{+}^{\theta_1,\theta_2}=SO(4)_{-}^{\theta_1+2\pi,\theta_2} S O ( 4 ) + θ 1 , θ 2 = S O ( 4 ) − θ 1 + 2 π , θ 2 .
One finds that the above results for a small gaugino mass m m m are precisely identical to the phase diagrams of S O ( 4 ) ± SO(4)_{\pm} S O ( 4 ) ± Yang-Mills theories obtained via the center/CP anomaly in § 3.4 .
Spin(4k)/SO(4k) Yang-Mills theory
In this section, we generalize our discussions in § 3 to Spin(4k)/SO(4k) Yang-Mills theories for k > 1 k>1 k > 1 . An important difference from *Spin(4)/SO(4)_{\pm}$ is that there is only a single θ \theta θ -parameter. Furthermore, the center/CP anomaly is slightly different between odd and even k k k .
Center/CP anomaly and mixed gauge theory
Center/CP anomaly
When k k k is odd, the mixed center/CP anomaly for the Spin(4k) Yang-Mills theory is almost identical to that of the Spin(4) Yang-Mills [17 , 16 , 18 , 4 ] . Namely, when the θ \theta θ -parameter is shifted by 2 π 2\pi 2 π , the action receives
Δ S = i π 2 ∫ M ( P ( B 1 ) + P ( B 2 ) ) , (41) \Delta S=\frac{\mathrm{i}\pi}{2}\int_{\mathcal{M}}\left(\mathcal{P}(B_1)+\mathcal{P}(B_2)\right), \tag*{(41)} Δ S = 2 i π ∫ M ( P ( B 1 ) + P ( B 2 ) ) , (41)
where ( B 1 , B 2 ) ∈ Z 2 ( M , Z 2 × Z 2 ) (B_1,B_2) \in Z^2(\mathcal{M},\mathbb{Z}_2 \times\mathbb{Z}_2) ( B 1 , B 2 ) ∈ Z 2 ( M , Z 2 × Z 2 ) stands for the background gauge fields for the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry. Since this is precisely the same as the S p i n ( 4 ) / S O ( 4 ) ± \mathit{Spin}(4)/\mathit{SO}(4)_{\pm} Spin ( 4 ) / SO ( 4 ) ± case with the constraint
θ 1 = θ 2 = θ , (42) \theta_1 = \theta_2 = \theta\,, \tag*{(42)} θ 1 = θ 2 = θ , (42)
it is straightforward to read off the phase diagrams of the S p i n ( 4 k ) / S O ( 4 k ) ± \mathit{Spin}(4k)/\mathit{SO}(4k)_{\pm} Spin ( 4 k ) / SO ( 4 k ) ± Yang-Mills theory for odd k k k from the results of § 3 . In particular, the periodicity of θ \theta θ for generic values of ( B 1 , B 2 ) (B_1,B_2) ( B 1 , B 2 ) is θ ∼ θ + 4 π \theta\sim\theta+ 4\pi θ ∼ θ + 4 π .
When k k k is even, the above anomaly (41) is modified to be [17 , 16 , 18 , 4 ]
Δ S = π i ∫ M B 1 ∪ B 2 . (43) \Delta S = \pi i \int_{\mathcal{M}} B_1 \cup B_2 \,. \tag*{(43)} Δ S = π i ∫ M B 1 ∪ B 2 . (43)
Since ∫ M B 1 ∪ B 2 ∈ Z 2 \int_{\mathcal{M}} B_1 \cup B_2 \in\mathbb{Z}_2 ∫ M B 1 ∪ B 2 ∈ Z 2 , we find again that the periodicity under a general background field is θ ∼ θ + 4 π \theta\sim\theta+ 4\pi θ ∼ θ + 4 π .
Below, we consider the mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory separately for odd and even k k k .
Mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory for odd k k k
Let us first consider the case of odd k k k , and study the mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory. Namely, we compactify the S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) Yang-Mills on S 1 S^1 S 1 and gauge the 3d 1-form Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 symmetry. The global symmetry of this theory can be read off from the diagonal elements of Table 3 .
The 4d center/CP anomaly (41) implies that the periodicity of the θ \theta θ -parameter for the mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory is θ ∼ θ + 4 π \theta\sim\theta+4\pi θ ∼ θ + 4 π . From the global symmetry group, one heuristically finds a vacuum structure shown in Figure 7 and Table 7 , where T 1 , T 2 , D 1 T_1,T_2,D_1 T 1 , T 2 , D 1 and D 2 D_2 D 2 are line operators labeled respectively by ( 0 , 0 ; 1 , 0 ) (0,0;1,0) ( 0 , 0 ; 1 , 0 ) , ( 0 , 0 ; 0 , 1 ) (0,0;0,1) ( 0 , 0 ; 0 , 1 ) , ( 1 , 0 ; 1 , 0 ) (1,0;1,0) ( 1 , 0 ; 1 , 0 ) , ( 0 , 1 ; 0 , 1 ) ∈ Z 2 (0,1;0,1) \in\mathbb{Z}_2 ( 0 , 1 ; 0 , 1 ) ∈ Z 2 (See appendix § 10 for our notation of charges of line operators). Note that this vacuum structure is consistent with the fact that the Witten effect ( T 1 , T 2 ) → ( D 1 , D 2 ) (T_1,T_2) \to(D_1,D_2) ( T 1 , T 2 ) → ( D 1 , D 2 ) under θ → θ + 2 π \theta\to\theta+2\pi θ → θ + 2 π . When the value of θ \theta θ is generic, the mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory has four vacua, while it has eight vacua when θ / π \theta/\pi θ / π is an odd integer. All vacua now belong to a single orbit of ( Z 2 ) 4 ⋊ Z 2 (\mathbb{Z}_2)^4 \rtimes\mathbb{Z}_2 ( Z 2 ) 4 ⋊ Z 2 in contrast to the S p i n ( 4 ) \mathit{Spin}(4) Spin ( 4 ) case.
Table 7. The VEVs of T i T_i T i and D i D_i D i in various regions of Fig. 7, where a i a_i a i and b i b_i b i are non-vanishing values. The signs are not correlated. There are four vacua in each region. The T 1 T_1 T 1 and T 2 T_2 T 2 are associated with the lines labeled by ( 0 , 0 ; 1 , 0 ) (0,0;1,0) ( 0 , 0 ; 1 , 0 ) , ( 0 , 0 ; 0 , 1 ) ∈ Z 4 2 (0,0;0,1) \in\mathbb{Z}_4^2 ( 0 , 0 ; 0 , 1 ) ∈ Z 4 2 , respectively. The D 1 D_1 D 1 and D 2 D_2 D 2 are respectively associated with ( 1 , 0 ; 1 , 0 ) (1,0;1,0) ( 1 , 0 ; 1 , 0 ) , ( 0 , 1 ; 0 , 1 ) (0,1;0,1) ( 0 , 1 ; 0 , 1 ) for odd k k k , and with ( 0 , 1 ; 1 , 0 ) (0,1;1,0) ( 0 , 1 ; 1 , 0 ) , ( 1 , 0 ; 0 , 1 ) (1,0;0,1) ( 1 , 0 ; 0 , 1 ) for even k k k .
Mixed S p i n ( 4 k ) \mathit{Spin}(4k) Spin ( 4 k ) theory for even k k k
When k k k is even, the mixed center/CP anomaly is different from the S p i n ( 4 ) \mathit{Spin}(4) Spin ( 4 ) case, and therefore the corresponding mixed gauge theory has a different global symmetry.
Indeed, replacing (41) with (43) implies that the relations (24) for the 0-form global symmetry at θ = π \theta=\pi θ = π are now replaced with
C P ⋅ c 1 ⋅ C P = c 1 ⋅ h 2 , C P ⋅ c 2 ⋅ C P = c 2 ⋅ h 1 , C P ⋅ h k = h k ⋅ C P , (44) CP \cdot c_1 \cdot CP = c_1 \cdot h_2 \,, \qquad CP \cdot c_2 \cdot CP = c_2 \cdot h_1 \,, \qquad CP \cdot h_k = h_k \cdot CP \,, \tag*{(44)} C P ⋅ c 1 ⋅ C P = c 1 ⋅ h 2 , C P ⋅ c 2 ⋅ C P = c 2 ⋅ h 1 , C P ⋅ h k = h k ⋅ C P , (44)
c k ⋅ h ℓ = h ℓ ⋅ c k , c k 2 = h k 2 = C P 2 = 1. (45) c_k \cdot h_\ell= h_\ell\cdot c_k,\qquad c_k^2 = h_k^2 = CP^2 = 1 . \tag*{(45)} c k ⋅ h ℓ = h ℓ ⋅ c k , c k 2 = h k 2 = C P 2 = 1. (45)
This shows that the global symmetry at θ = π \theta= \pi θ = π is still expressed as ( Z 2 ) 4 ⋊ Z 2 (\mathbb{Z}_2)^4 \rtimes\mathbb{Z}_2 ( Z 2 ) 4 ⋊ Z 2 , but the relevant semi-direct product is different from the one for odd k k k .
Note that, due to (43) , the Witten effect now implies ( T 1 , T 2 ) ⟶ ( T 1 W 2 , T 2 W 1 ) (T_1,T_2) \longrightarrow(T_1W_2,T_2W_1) ( T 1 , T 2 ) ⟶ ( T 1 W 2 , T 2 W 1 ) under θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π , where W 1 W_1 W 1 and W 2 W_2 W 2 are Wilson lines labeled respectively by ( 1 , 0 ; 0 , 0 ) ∈ Z 4 2 (1,0;0,0) \in\mathbb{Z}_4^2 ( 1 , 0 ; 0 , 0 ) ∈ Z 4 2 and ( 0 , 1 ; 0 , 0 ) ∈ Z 4 2 (0,1;0,0) \in\mathbb{Z}_4^2 ( 0 , 1 ; 0 , 0 ) ∈ Z 4 2 . This implies that the vacuum structure shown in Table 7 is still valid for even k k k if D 1 D_1 D 1 and D 2 D_2 D 2 are understood as D 1 = T 1 W 2 D_1 = T_1W_2 D 1 = T 1 W 2 and D 2 = T 2 W 1 D_2 = T_2W_1 D 2 = T 2 W 1 .
Phase diagrams of Spin(4k)/SO(4k) theories
In the rest of this section, we use the notation that ( D 1 , D 2 ) = ( T 1 W 1 , T 2 W 2 ) (D_1,D_2) = (T_1W_1,T_2W_2) ( D 1 , D 2 ) = ( T 1 W 1 , T 2 W 2 ) for odd k k k and ( D 1 , D 2 ) = ( T 1 W 2 , T 2 W 1 ) (D_1,D_2) = (T_1W_2,T_2W_1) ( D 1 , D 2 ) = ( T 1 W 2 , T 2 W 1 ) for even k k k , so that we can treat odd and even k k k in the same way. Below, we read off the phase diagrams of the S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± theories from the above results for the mixed S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) theory, via the relations
S p i n ( 4 k ) = ( mixed S p i n ( 4 k ) ) / ( Z 2 ( h 1 ) × Z 2 ( h 2 ) ) . (46) Spin(4k) = \bigl(\text{mixed } Spin(4k)\bigr) \big/ \left(\mathbb{Z}_2^{(h_1)} \times\mathbb{Z}_2^{(h_2)}\right) . \tag*{(46)} S p in ( 4 k ) = ( mixed S p in ( 4 k ) ) / ( Z 2 ( h 1 ) × Z 2 ( h 2 ) ) . (46)
S O ( 4 k ) + = ( mixed S p i n ( 4 k ) ) / ( Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) ) (47) SO(4k)_{+}=\left(\text{mixed }Spin(4k)\right)/\left(\mathbb{Z}_{2}^{(c_{1}\cdot c_{2})}\times\mathbb{Z}_{2}^{(h_{1}\cdot h_{2})}\right) \tag*{(47)} S O ( 4 k ) + = ( mixed S p in ( 4 k ) ) / ( Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) ) (47)
S O ( 4 k ) − = ( mixed S p i n ( 4 k ) ) / ( Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) ) (48) SO(4k)_{-}=\left(\text{mixed }Spin(4k)\right)/\left(\mathbb{Z}_{2}^{(c_{1}\cdot c_{2}\cdot h_{1})}\times\mathbb{Z}_{2}^{(h_{1}\cdot h_{2})}\right) \tag*{(48)} S O ( 4 k ) − = ( mixed S p in ( 4 k ) ) / ( Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) ) (48)
Let us start with the S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) Yang-Mills, which is obtained by gauging the Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_{2}^{(h_{1})}\times\mathbb{Z}_{2}^{(h_{2})} Z 2 ( h 1 ) × Z 2 ( h 2 ) symmetry of the mixed S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) theory. Here, Z 2 ( h 1 ) × Z 2 ( h 2 ) \mathbb{Z}_{2}^{(h_{1})}\times\mathbb{Z}_{2}^{(h_{2})} Z 2 ( h 1 ) × Z 2 ( h 2 ) is generated by h 1 h_{1} h 1 and h 2 h_{2} h 2 , whose actions on T k T_{k} T k and D k D_{k} D k are shown in (22) . We see that both in regions 1 and 2 there exists a single vacuum corresponding to ⟨ W k ⟩ = 0 \langle W_{k}\rangle=0 ⟨ W k ⟩ = 0 . In the convention that region 1 corresponds to the trivial phase, region 2 corresponds to an SPT phase protected by the center symmetry of S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) . These results are summarized in the first line of Table 8 .
Let us next consider the S O ( 4 k ) + SO(4k)_{+} S O ( 4 k ) + theory using (47) . Since Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_{2}^{(c_{1}\cdot c_{2})}\times\mathbb{Z}_{2}^{(h_{1}\cdot h_{2})} Z 2 ( c 1 ⋅ c 2 ) × Z 2 ( h 1 ⋅ h 2 ) acts on T k T_{k} T k and D k D_{k} D k as in (30) and (31) , we see that the resulting vacuum structure is as shown in Figure 8 . For generic values of θ \theta θ , there are two vacua accompanied with an unbroken Z 2 \mathbb{Z}_{2} Z 2 gauge theory. When θ / π \theta/\pi θ / π is an odd integer, the S O ( 4 k ) + SO(4k)_{+} S O ( 4 k ) + theory has four degenerate vacua. These results are summarized in the second line of Table 8 .
Figure 8. The vacuum structure of S O ( 4 k ) + SO(4k)_{+} S O ( 4 k ) + Yang-Mills theory compactified on S 1 S^{1} S 1 for k ≥ 1 k \ge1 k ≥ 1 . Each vacuum is accompanied with a Z 2 \mathbb{Z}_{2} Z 2 TQFT. When going back to four dimensions, every pair of vacua connected by T 1 T 2 → − T 1 T 2 T_{1}T_{2} \to-T_{1}T_{2} T 1 T 2 → − T 1 T 2 and D 1 D 2 → − D 1 D 2 D_{1}D_{2} \to-D_{1}D_{2} D 1 D 2 → − D 1 D 2 is mapped to a single vacuum.
One can analyze the S O ( 4 k ) − SO(4k)_{-} S O ( 4 k ) − theory in the same way. The Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) \mathbb{Z}_{2}^{(c_{1}\cdot c_{2}\cdot h_{1})}\times\mathbb{Z}_{2}^{(h_{1}\cdot h_{2})} Z 2 ( c 1 ⋅ c 2 ⋅ h 1 ) × Z 2 ( h 1 ⋅ h 2 ) acts on T k T_{k} T k and D k D_{k} D k as in (33) and (34) , leaving only one vacuum for generic values of θ \theta θ . This single vacuum is characterized by ⟨ T 1 D 2 ⟩ = ⟨ D 1 T 2 ⟩ = 0 \langle T_{1}D_{2}\rangle=\langle D_{1}T_{2}\rangle=0 ⟨ T 1 D 2 ⟩ = ⟨ D 1 T 2 ⟩ = 0 . When the vacuum for region 1 is called the trivial vacuum, region 2 is in the non-trivial SPT phase protected by the 1-form symmetry. When θ / π \theta/\pi θ / π is an odd integer, the S O ( 4 k ) − SO(4k)_{-} S O ( 4 k ) − theory has two degenerate vacua. These results are shown in the third line of Table 8 .
Note that, in contrast to the case of S O ( 4 ) ± SO(4)_{\pm} S O ( 4 ) ± , the IR physics of the S O ( 4 k ) + SO(4k)_{+} S O ( 4 k ) + theory and that of the S O ( 4 k ) − SO(4k)_{-} S O ( 4 k ) − theory are completely different. This is possible since for k > 1 k>1 k > 1 one has
S O ( 4 k ) ± θ + 2 π = S O ( 4 k ) ± θ (49) SO(4k)_{\pm}^{\theta+2\pi}=SO(4k)_{\pm}^{\theta} \tag*{(49)} S O ( 4 k ) ± θ + 2 π = S O ( 4 k ) ± θ (49)
and therefore these two theories are not related by any shift of the θ \theta θ -parameter.
Comparison to soft SUSY breaking
Here we compare the above phase diagrams with those of supersymmetric S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± Yang-Mills theories with a small gaugino mass turned on. We will see that they are in perfect agreement with each other.
Let us start with the N = 1 \mathcal{N}=1 N = 1 S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) SYM theory. For k ≥ 2 k \ge2 k ≥ 2 , the theory has ( 4 k − 2 ) (4k-2) ( 4 k − 2 ) SUSY vacua [15 , 16 ] . In terms of the gaugino bilinear operator λ α λ α \lambda_{\alpha}\lambda^{\alpha} λ α λ α , these ( 4 k − 2 ) (4k-2) ( 4 k − 2 ) vacua are characterized by
⟨ Tr ( λ α λ α ) ⟩ = κ Λ 3 e − 2 π i 4 k − 2 n , n = 0 , 1 , 2 , ⋯ , 4 k − 3. (50) \langle\operatorname{Tr}(\lambda_{\alpha}\lambda^{\alpha})\rangle= \kappa\Lambda^{3}e^{-\frac{2\pi i}{4k-2}n}, \qquad n=0,1,2,\cdots,4k-3 . \tag*{(50)} ⟨ Tr ( λ α λ α )⟩ = κ Λ 3 e − 4 k − 2 2 π i n , n = 0 , 1 , 2 , ⋯ , 4 k − 3. (50)
where κ \kappa κ is a constant, and Λ \Lambda Λ is the dynamical scale such that Λ 3 ( 4 k − 2 ) = ∣ Λ 3 ( 4 k − 2 ) ∣ e i θ \Lambda^{3(4k-2)} = \lvert\Lambda^{3(4k-2)}\rvert e^{i\theta} Λ 3 ( 4 k − 2 ) = ∣ Λ 3 ( 4 k − 2 ) ∣ e i θ . Note that shifting θ \theta θ by 2 π 2\pi 2 π shifts n n n by one. Since the center of the gauge group is ( Z 2 ) 2 (\mathbb{Z}_{2})^{2} ( Z 2 ) 2 , line operators are characterized by ( z 1 , z 2 ; w 1 , w 2 ) ∈ ( Z 2 ) 4 (z_{1},z_{2};w_{1},w_{2}) \in(\mathbb{Z}_{2})^{4} ( z 1 , z 2 ; w 1 , w 2 ) ∈ ( Z 2 ) 4 , where ( z 1 , z 2 ) (z_{1},z_{2}) ( z 1 , z 2 ) corresponds to the electric charges of the line (up to root lattice of the gauge group), while ( w 1 , w 2 ) (w_{1},w_{2}) ( w 1 , w 2 ) corresponds to the magnetic charges.10 The S p i n ( 4 k ) Spin(4k) S p in ( 4 k ) SYM contains ( 1 , 0 ; 0 , 0 ) (1,0;0,0) ( 1 , 0 ; 0 , 0 ) , ( 0 , 1 ; 0 , 0 ) (0,1;0,0) ( 0 , 1 ; 0 , 0 ) and ( 1 , 1 ; 0 , 0 ) (1,1;0,0) ( 1 , 1 ; 0 , 0 ) as genuine line operators. At vacua for even n n n , purely magnetic monopoles are condensed, while at the other vacua dyons are condensed.11 From this we see that, at any vacuum, all the genuine line operators have an area law, since every genuine line operator has a non-vanishing Dirac pairing with at least one condensed particle.12 Therefore, all the ( 4 k − 2 ) (4k-2) ( 4 k − 2 ) vacua are in the confining or an oblique confining phase. In the convention that the vacua for even n n n are in the confining phase, those for odd n n n are in an oblique confining phase. This oblique confining phase is an SPT phase protected by the Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 center symmetry.
Turning on a soft SUSY breaking term δ L = m Tr ( λ α λ α ) + c . c . \delta\mathcal{L}=m\operatorname{Tr}(\lambda^\alpha\lambda^\alpha)+\mathrm{c.c.} δ L = m Tr ( λ α λ α ) + c.c. , we obtain the effective potential13
V = − 2 ∣ m κ Λ 3 ∣ cos ( θ − 2 π n 4 k − 2 ) , (51) V=-2|m\kappa\Lambda^3|\cos\left(\frac{\theta-2\pi n}{4k-2}\right), \tag*{(51)} V = − 2∣ mκ Λ 3 ∣ cos ( 4 k − 2 θ − 2 π n ) , (51)
which resolves the above ( 4 k − 2 ) (4k-2) ( 4 k − 2 ) degeneracy. Specifically, the vacuum for n = ℓ n=\ell n = ℓ has the minimum energy when ( 2 ℓ − 1 ) π < θ < ( 2 ℓ + 1 ) π (2\ell-1)\pi<\theta<(2\ell+1)\pi ( 2 ℓ − 1 ) π < θ < ( 2 ℓ + 1 ) π . When θ / π \theta/\pi θ / π is an odd integer, two vacua are degenerate. Although there are ( 4 k − 2 ) (4k-2) ( 4 k − 2 ) different values of n n n , one can only distinguish vacua for even n n n from those for odd n n n . Thus, we observe only two different phases in the infrared. The vacua for even n n n are in the trivial phase, and those for odd n n n are in an SPT phase protected by the center Z 2 × Z 2 \mathbb{Z}_2\times\mathbb{Z}_2 Z 2 × Z 2 symmetry. This result is in perfect agreement with the first line of Table 8 read off from the center/CP anomaly of non-supersymmetric Spin ( 4 k ) \operatorname{Spin}(4k) Spin ( 4 k ) Yang-Mills theory.
The generalization of the above discussion to SO ( 4 k ) ± \operatorname{SO}(4k)_{\pm} SO ( 4 k ) ± theories is straightforward. The only difference is that each vacuum is associated with a different TQFT due to different spectra of line operators. Let us first consider the SO ( 4 k ) + \operatorname{SO}(4k)_+ SO ( 4 k ) + Yang-Mills theory, in which genuine line operators have charge ( z 1 , z 2 ; w 1 , w 2 ) = ( 1 , 1 ; 0 , 0 ) , ( 0 , 0 ; 1 , 1 ) (z_1,z_2;w_1,w_2)=(1,1;0,0),(0,0;1,1) ( z 1 , z 2 ; w 1 , w 2 ) = ( 1 , 1 ; 0 , 0 ) , ( 0 , 0 ; 1 , 1 ) and ( 1 , 1 ; 1 , 1 ) (1,1;1,1) ( 1 , 1 ; 1 , 1 ) [16 ] . Note that, while the line for ( 1 , 1 ; 0 , 0 ) (1,1;0,0) ( 1 , 1 ; 0 , 0 ) has an area law, either ( 0 , 0 ; 1 , 1 ) (0,0;1,1) ( 0 , 0 ; 1 , 1 ) or ( 1 , 1 ; 1 , 1 ) (1,1;1,1) ( 1 , 1 ; 1 , 1 ) has a perimeter law since its charge is proportional to that of the condensed particles up to a shift by the root lattice. This implies that the IR TQFT is a Z 2 \mathbb{Z}_2 Z 2 TQFT at all vacua. Note that the Z 2 \mathbb{Z}_2 Z 2 TQFT for − π < θ < π -\pi<\theta<\pi − π < θ < π and that for π < θ < 3 π \pi<\theta<3\pi π < θ < 3 π are different, which simultaneously exist when θ / π \theta/\pi θ / π is an odd integer. This result is in perfect agreement with the second line of Table 8 . The case of SO ( 4 k ) − \operatorname{SO}(4k)_- SO ( 4 k ) − can be analyzed in the same way, whose genuine line operators are those of charge ( 1 , 1 ; 0 , 0 ) , ( 1 , 0 ; 1 , 1 ) (1,1;0,0),(1,0;1,1) ( 1 , 1 ; 0 , 0 ) , ( 1 , 0 ; 1 , 1 ) and ( 0 , 1 ; 1 , 1 ) (0,1;1,1) ( 0 , 1 ; 1 , 1 ) . Since all these lines always have an area law, every vacuum is in the trivial phase or in the SPT phase protected by the 1-form symmetry. This result is identical to the third line of Table 8 .
Spin(4k+2)/SO(4k+2) Yang-Mills theory
In this section, we study the phase diagrams of the Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± \operatorname{Spin}(4k+2)/\operatorname{SO}(4k+2)_{\pm} Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± Yang-Mills theories for a positive integer k k k . Since the center of Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) is Z 4 \mathbb{Z}_4 Z 4 instead of Z 2 × Z 2 \mathbb{Z}_2\times\mathbb{Z}_2 Z 2 × Z 2 , we will find essentially different phase diagrams from what we have obtained in the previous sections.
In Sec. 5.1, we will study it using the mixed center/CP anomaly of the theory. In Sec. 5.2, we will show that the result of Sec. 5.1 is in agreement with the result suggested by the soft SUSY breaking of N = 1 \mathcal{N}=1 N = 1 SYM.
Phase digram from the center/CP anomaly
The mixed center/CP anomaly for Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory is expressed as [16 , 17 , 18 , 4 ]
Δ S = ± i π 4 ∫ M P ( B ) , (52) \Delta S = \pm\frac{i\pi}{4}\int_{\mathcal{M}} \mathcal{P}(B), \tag*{(52)} Δ S = ± 4 iπ ∫ M P ( B ) , (52)
where Δ S \Delta S Δ S is the change of the action under the shift θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π of the θ \theta θ -parameter, and B ∈ Z 2 ( M , Z 4 ) B \in Z^2(\mathcal{M},\mathbb{Z}_4) B ∈ Z 2 ( M , Z 4 ) is the background gauge field for the center Z 4 \mathbb{Z}_4 Z 4 symmetry of the Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) Yang-Mills theory. The sign on the right-hand side depends on whether k k k is even or odd. The above expression for Δ S \Delta S Δ S implies that the periodicity is
θ ∼ θ + 8 π , (53) \theta\sim\theta+ 8\pi, \tag*{(53)} θ ∼ θ + 8 π , (53)
in the presence of generic B ≠ 0 B \ne0 B = 0 .
For later use, let us briefly discuss the line operators of the Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) Yang-Mills theory. Since the center of Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) is Z 4 \mathbb{Z}_4 Z 4 , 4d line operators are labeled by ( z ; w ) ∈ Z 4 × Z 4 (z;w) \in\mathbb{Z}_4 \times\mathbb{Z}_4 ( z ; w ) ∈ Z 4 × Z 4 (See appendix A for our notation of the charges of line operators). One can take the one corresponding to ( z ; w ) = ( 0 ; 1 ) (z;w)=(0;1) ( z ; w ) = ( 0 ; 1 ) and the one corresponding to ( z ; w ) = ( 1 ; 0 ) (z;w)=(1;0) ( z ; w ) = ( 1 ; 0 ) as primitive line operators. In the Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory, the ( 0 ; 1 ) (0;1) ( 0 ; 1 ) line is not a genuine line operator (and therefore attached to a surface), but will play an important role in the study of the mixed gauge theory below.
Mixed Spin(4k + 2) gauge theory and its phase diagram
Compactifying the 4d Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) Yang-Mills on S 1 S^1 S 1 , we obtain a 1-form Z 4 \mathbb{Z}_4 Z 4 symmetry and a 0-form Z 4 \mathbb{Z}_4 Z 4 symmetry in three dimensions. We call this 0-form Z 4 \mathbb{Z}_4 Z 4 symmetry Z 4 ( c ) \mathbb{Z}_4^{(c)} Z 4 ( c ) and denote its generator by c c c . We also gauge the 3d 1-form Z 4 \mathbb{Z}_4 Z 4 symmetry to obtain another 0-form Z 4 \mathbb{Z}_4 Z 4 symmetry, which we call Z 4 ( h ) \mathbb{Z}_4^{(h)} Z 4 ( h ) and whose generator we denote by h h h . The resulting 3d theory with the 0-form Z 4 ( c ) × Z 4 ( h ) \mathbb{Z}_4^{(c)} \times\mathbb{Z}_4^{(h)} Z 4 ( c ) × Z 4 ( h ) symmetry is called the mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory.
For generic values of θ \theta θ , the 4d CP symmetry is explicitly broken by the θ \theta θ -term, and therefore the 3d mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory has only Z 4 ( c ) × Z 4 ( h ) \mathbb{Z}_4^{(c)} \times\mathbb{Z}_4^{(h)} Z 4 ( c ) × Z 4 ( h ) as a global symmetry. However, when θ / π \theta/\pi θ / π is an integer, the 4d CP symmetry gives rise to an extra Z 2 \mathbb{Z}_2 Z 2 0-form symmetry of the mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory, which we denote by Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) and whose generator we denote by C P CP C P . The full global symmetry group is an extension of Z 4 ( c ) × Z 4 ( h ) \mathbb{Z}_4^{(c)} \times\mathbb{Z}_4^{(h)} Z 4 ( c ) × Z 4 ( h ) by Z 2 ( C P ) \mathbb{Z}_2^{(CP)} Z 2 ( C P ) .
Specifically, when θ = ℓ π + 4 π n \theta= \ell\pi+ 4\pi n θ = ℓ π + 4 π n for ℓ = 0 , 1 , 2 \ell= 0,1,2 ℓ = 0 , 1 , 2 or 3 3 3 and n ∈ Z n \in\mathbb{Z} n ∈ Z , the C P CP C P involves a shift θ → θ + 2 π ℓ \theta\to\theta+ 2\pi\ell θ → θ + 2 π ℓ . In this case, the 4d mixed anomaly (52) implies the following relations:
C P ⋅ c ⋅ C P = c − 1 ⋅ h ± ℓ , (54) CP \cdot c \cdot CP = c^{-1} \cdot h^{\pm\ell}, \tag*{(54)} C P ⋅ c ⋅ C P = c − 1 ⋅ h ± ℓ , (54)
C P ⋅ h = h ⋅ C P , (55) CP \cdot h = h \cdot CP, \tag*{(55)} C P ⋅ h = h ⋅ C P , (55)
c ⋅ h = h ⋅ c , (56) c \cdot h = h \cdot c, \tag*{(56)} c ⋅ h = h ⋅ c , (56)
c 4 = h 4 = C P 2 = 1 , (57) c^4 = h^4 = CP^2 = 1, \tag*{(57)} c 4 = h 4 = C P 2 = 1 , (57)
where the power of h h h on the right-hand side of (54) is + ℓ +\ell + ℓ for odd k k k and − ℓ -\ell − ℓ for even k k k . Note that these relations are slightly different from (8) for S U ( 2 ) SU(2) S U ( 2 ) gauge group.14 Indeed, this global symmetry group is identified as ( Z 4 ( c ) × Z 4 ( h ) ) ⋊ φ ℓ Z 2 ( C P ) (\mathbb{Z}_4^{(c)} \times\mathbb{Z}_4^{(h)}) \rtimes_{\varphi_\ell} \mathbb{Z}_2^{(CP)} ( Z 4 ( c ) × Z 4 ( h ) ) ⋊ φ ℓ Z 2 ( C P ) associated with a group homomorphism
φ ℓ : Z 2 → Aut ( Z 4 ) 2 (58) \varphi_\ell:\mathbb{Z}_2 \to\operatorname{Aut}(\mathbb{Z}_4)^2 \tag*{(58)} φ ℓ : Z 2 → Aut ( Z 4 ) 2 (58)
specified by (54) and (55) . The full global symmetry of the mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory is then identified as shown in Table 9 .
Table 9. The global symmetry of the 3d mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory for a positive integer k k k . These are all 0-form symmetries. The φ ℓ \varphi_\ell φ ℓ is a group homomorphism φ ℓ : Z 2 → Aut ( Z 4 ) 2 \varphi_\ell:\mathbb{Z}_2 \to\operatorname{Aut}(\mathbb{Z}_4)^2 φ ℓ : Z 2 → Aut ( Z 4 ) 2 specified by (54) and (55) , where C P CP C P is the generator of Z 2 \mathbb{Z}_2 Z 2 and h h h and c c c are generators of ( Z 4 ) 2 (\mathbb{Z}_4)^2 ( Z 4 ) 2 .
We now identify a vacuum structure of the mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) gauge theory for various values of θ \theta θ . For that, we first need to find the order parameters of the phases. Recall that our primitive line operators in four dimensions are the one corresponding to ( z ; w ) = ( 0 ; 1 ) (z;w) = (0;1) ( z ; w ) = ( 0 ; 1 ) and the one corresponding to ( z ; w ) = ( 1 ; 0 ) (z;w) = (1;0) ( z ; w ) = ( 1 ; 0 ) . We denote the former line operator wrapping on S 1 S^1 S 1 by T T T , and the latter wrapping on S 1 S^1 S 1 by W W W . These are local operators in the 3d mixed gauge theory, and their VEVs characterize its phases. Under c c c and h h h , these operators transform as
c : T → T , W → ± i W , (59) c:T \to T,\qquad W \to\pm iW, \tag*{(59)} c : T → T , W → ± iW , (59)
h : T → ± i T , W → W , (60) h : T \to\pm iT,\qquad W \to W, \tag*{(60)} h : T → ± i T , W → W , (60)
where the signs in front of i W iW iW and i T iT i T are both taken to be plus for odd k k k and minus for even k k k . Similarly, under the C P CP C P , they transform as
C P : T → W ± ℓ T , W → W − 1 , (61) CP : T \to W^{\pm\ell}T,\qquad W \to W^{-1}, \tag*{(61)} C P : T → W ± ℓ T , W → W − 1 , (61)
when θ = ℓ π + 4 π n \theta= \ell\pi+ 4\pi n θ = ℓ π + 4 π n for ℓ = 0 , 1 , 2 , 3 \ell= 0,1,2,3 ℓ = 0 , 1 , 2 , 3 and n ∈ Z n \in\mathbb{Z} n ∈ Z . The transformation rule of W W W under the C P CP C P can be understood by noting that the C P CP C P flips the sign of the electric charge. Note that the above actions of c c c , h h h and C P CP C P correctly realize a representation of (5.3)–(5.6).
Under the same assumption as before, we now identify a vacuum structure of the mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory that is consistent with the global symmetry group shown in Table 9 . The result is shown in in Figure 9 and Figure 10 . For generic values of θ \theta θ , the mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory has four vacua, while it has eight vacua when θ / π \theta/\pi θ / π is an odd integer. We see that these vacua correctly give representations of the global symmetry groups generated by (59) , (5.9) and (5.10). Furthermore, they are also consistent with the Witten effect
Figure 10. The vacuum structure of mixed S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) theory for a positive integer k k k for each region in Figure 9 . In all these regions, the VEV of W W W (corresponding to the Wilson line in the spinor representation of S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) ) vanishes, while a magnetically charged operator has a non-vanishing VEV.
T → W ± 1 T (62) T \to W^{\pm1}T \tag*{(62)} T → W ± 1 T (62)
under θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π . These are strong evidence for our vacuum structure and phase diagram of the mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory.
Phase diagram of Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± \operatorname{Spin}(4k+2)/\operatorname{SO}(4k+2)_{\pm} Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± Yang-Mills theories
We now read off the phase diagrams of the Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± \operatorname{Spin}(4k+2)/\operatorname{SO}(4k+2)_{\pm} Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± Yang-Mills theories from that of the mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory identified above. These Yang-Mills theories are obtained from the mixed Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) theory by the following discrete gaugings:
Spin ( 4 k + 2 ) = ( mixed Spin ( 4 k + 2 ) ) / Z 4 ( h ) (63) \operatorname{Spin}(4k+2)=\left(\text{mixed }\operatorname{Spin}(4k+2)\right)/\mathbb{Z}_{4}^{(h)} \tag*{(63)} Spin ( 4 k + 2 ) = ( mixed Spin ( 4 k + 2 ) ) / Z 4 ( h ) (63)
SO ( 4 k + 2 ) + = ( mixed Spin ( 4 k + 2 ) ) / ( Z 2 ( h 2 ) × Z 2 ( c 2 ) ) (64) \operatorname{SO}(4k+2)_{+}=\left(\text{mixed }\operatorname{Spin}(4k+2)\right)/\left(\mathbb{Z}_{2}^{(h^{2})}\times\mathbb{Z}_{2}^{(c^{2})}\right) \tag*{(64)} SO ( 4 k + 2 ) + = ( mixed Spin ( 4 k + 2 ) ) / ( Z 2 ( h 2 ) × Z 2 ( c 2 ) ) (64)
SO ( 4 k + 2 ) − = ( mixed Spin ( 4 k + 2 ) ) / Z 4 ( c 2 h ) (65) \operatorname{SO}(4k+2)_{-}=\left(\text{mixed }\operatorname{Spin}(4k+2)\right)/\mathbb{Z}_{4}^{(c^{2}h)} \tag*{(65)} SO ( 4 k + 2 ) − = ( mixed Spin ( 4 k + 2 ) ) / Z 4 ( c 2 h ) (65)
where ( Z 2 ) ( g ) (\mathbb{Z}_{2})^{(g)} ( Z 2 ) ( g ) is a Z 2 \mathbb{Z}_{2} Z 2 -subgroup generated by g g g . Therefore, the vacuum structures of the Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± \operatorname{Spin}(4k+2)/\operatorname{SO}(4k+2)_{\pm} Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± Yang-Mills theories are obtained by taking the corresponding discrete quotients of the vacuum structure shown in Figure 10 .
Let us start with the Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) Yang-Mills theory. Since h h h acts on T T T and W W W as in (5.9), the genuine line operators of the Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) Yang-Mills are W W W , W 2 W^{2} W 2 and W 3 W^{3} W 3 . In any of regions 1 – 4 of Figure 9 , the discrete gauging (5.12) leaves only a single vacuum with ⟨ W ⟩ = 0 \langle W\rangle=0 ⟨ W ⟩ = 0 . These are confining and oblique confining vacua. In the convention that region 1 is in the confining phase, the other regions are in SPT phases protected by the Z 4 \mathbb{Z}_{4} Z 4 center symmetry of Spin ( 4 k + 2 ) \operatorname{Spin}(4k+2) Spin ( 4 k + 2 ) . This result is summarized in the first line of Table 10. When crossing a boundary of the regions, one observes an SPT phase transition. At the phase boundary, the CP is spontaneously broken. The periodicity of θ \theta θ is θ ∼ θ + 8 π \theta\sim\theta+8\pi θ ∼ θ + 8 π in the presence of a generic background gauge field of the Z 4 \mathbb{Z}_{4} Z 4 center symmetry.
Let us next consider the SO ( 4 k + 2 ) + \operatorname{SO}(4k+2)_{+} SO ( 4 k + 2 ) + Yang-Mills theory, which is obtained by gauging the Z 2 ( h 2 ) × Z 2 ( c 2 ) \mathbb{Z}_{2}^{(h^{2})}\times\mathbb{Z}_{2}^{(c^{2})} Z 2 ( h 2 ) × Z 2 ( c 2 ) symmetry of the mixed gauge theory. Since h 2 h^{2} h 2 and c 2 c^{2} c 2 act on T T T and W W W as
h 2 : T → − T , W → W , c 2 : T → T , W → − W (66) h^{2}: \ T\to-T,\quad W\to W,\qquad c^{2}: \ T\to T,\quad W\to-W \tag*{(66)} h 2 : T → − T , W → W , c 2 : T → T , W → − W (66)
this gauging maps the vacuum structure of Figure 10 to the one shown in Figure 11 . There, the gauge-invariant operators are taken to be T 2 T^{2} T 2 , W 2 W^{2} W 2 and W 2 T 2 W^{2}T^{2} W 2 T 2 , corresponding to the genuine line operators of the SO ( 4 k + 2 ) + \operatorname{SO}(4k+2)_{+} SO ( 4 k + 2 ) + Yang-Mills theory. Every vacuum shown in Figure 11 is
Figure 11. The vacuum structure of S O ( 4 k + 2 ) + SO(4k+2)_+ S O ( 4 k + 2 ) + Yang-Mills theory accompanied by a Z 2 \mathbb{Z}_2 Z 2 TQFT.15
caption: The IR TQFT present in the four regions in Figure 9 for 4d S p i n ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Spin(4k+2)/SO(4k+2)_\pm S p in ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Yang-Mills theory. The SPT phases for S p i n ( 4 k + 2 ) Spin(4k+2) S p in ( 4 k + 2 ) are protected by the electric Z 4 \mathbb{Z}_4 Z 4 1-form symmetry (i.e., the center symmetry), while those for S O ( 4 k + 2 ) − SO(4k+2)_- S O ( 4 k + 2 ) − are protected by electro-magnetic Z 4 \mathbb{Z}_4 Z 4 1-form symmetry.
:::
This result is summarized in the second line of Table 10. There is a residual Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 0-form global symmetry generated by c c c and h h h , which is a 3d 0-form symmetry arising from the 4d electric and magnetic Z 2 \mathbb{Z}_2 Z 2 1-form symmetries of the S O ( 4 k + 2 ) + SO(4k+2)_+ S O ( 4 k + 2 ) + Yang-Mills theory.
Finally, we consider the S O ( 4 k + 2 ) − SO(4k+2)_- S O ( 4 k + 2 ) − Yang-Mills theory obtained by gauging the Z 4 ( c 2 h ) \mathbb{Z}_4^{(c^2h)} Z 4 ( c 2 h ) symmetry of the mixed gauge theory. Noting that c 2 h c^2h c 2 h acts on T T T and W W W as
c 2 h : T → ± i T , W → − W , (67) c^2h:\quad T\to\pm iT,\quad W\to-W, \tag*{(67)} c 2 h : T → ± i T , W → − W , (67)
we find that the gauge-invariant operators are now W 2 W^2 W 2 , W T 2 WT^2 W T 2 and W 3 T 2 W^3T^2 W 3 T 2 . One can check that they indeed correspond to the genuine line operators of the S O ( 4 k + 2 ) − SO(4k+2)_- S O ( 4 k + 2 ) − Yang-Mills theory. We see that in each of regions 1–4 there exists only one vacuum with ⟨ W 2 ⟩ = ⟨ W T 2 ⟩ = ⟨ W 3 T 2 ⟩ = 0 \langle W^2\rangle=\langle WT^2\rangle=\langle W^3T^2\rangle=0 ⟨ W 2 ⟩ = ⟨ W T 2 ⟩ = ⟨ W 3 T 2 ⟩ = 0 . The residual 0-form global symmetry is now Z 4 \mathbb{Z}_4 Z 4 generated by
c : T → T , W → ± i W , (68) c:\quad T\to T,\quad W\to\pm iW, \tag*{(68)} c : T → T , W → ± iW , (68)
which is a 3d 0-form symmetry arising from the 4d Z 4 \mathbb{Z}_4 Z 4 1-form symmetry of S O ( 4 k + 2 ) − SO(4k+2)_- S O ( 4 k + 2 ) − Yang-Mills.16 The vacua in regions 1–4 are SPT phases protected by Z 4 ( c ) \mathbb{Z}_4^{(c)} Z 4 ( c ) . This result is summarized in the third line of Table 10.
Comparison to soft SUSY breaking
Here we show that the above phase diagrams read off from the center/CP anomaly of the non-supersymmetric theories are in perfect agreement with those of the N = 1 \mathcal{N}=1 N = 1 S p i n ( 4 k + 2 ) / S O ( 4 k + 2 ) ± \mathit{Spin}(4k+2)/\mathrm{SO}(4k+2)_{\pm} Spin ( 4 k + 2 ) / SO ( 4 k + 2 ) ± SYM with a small gaugino mass turned on.
Let us start with the S p i n ( 4 k + 2 ) \mathit{Spin}(4k+2) Spin ( 4 k + 2 ) gauge group. The N = 1 \mathcal{N}=1 N = 1 S p i n ( 4 k + 2 ) \mathit{Spin}(4k+2) Spin ( 4 k + 2 ) SYM has 4 k 4k 4 k SUSY vacua [15 , 16 ] , labeled by the
⟨ ( Tr λ α λ α ) ⟩ = κ Λ 3 e − 2 π i 4 k n , n = 0 , 1 , 2 , … , 4 k − 1 , (69) \left\langle\left(\operatorname{Tr}\lambda_{\alpha}\lambda^{\alpha}\right)\right\rangle=\kappa\Lambda^{3}e^{-\frac{2\pi i}{4k}n},\qquad n=0,1,2,\ldots,4k-1, \tag*{(69)} ⟨ ( Tr λ α λ α ) ⟩ = κ Λ 3 e − 4 k 2 π i n , n = 0 , 1 , 2 , … , 4 k − 1 , (69)
where Tr ( λ α λ α ) \operatorname{Tr}(\lambda_{\alpha}\lambda^{\alpha}) Tr ( λ α λ α ) is the gaugino bilinear operator, and κ \kappa κ is a constant. The dynamical scale Λ \Lambda Λ is related to the θ \theta θ -angle by Λ 12 k = ∣ Λ 12 k ∣ e i θ \Lambda^{12k}=|\Lambda^{12k}|e^{i\theta} Λ 12 k = ∣ Λ 12 k ∣ e i θ . The line operators are labeled by ( z ; w ) ∈ Z 4 × Z 4 (z;w)\in\mathbb{Z}_4\times\mathbb{Z}_4 ( z ; w ) ∈ Z 4 × Z 4 , and those with ( z ; w ) = ( 1 , 0 ) , ( 2 , 0 ) (z;w)=(1,0),(2,0) ( z ; w ) = ( 1 , 0 ) , ( 2 , 0 ) and ( 3 , 0 ) (3,0) ( 3 , 0 ) are genuine line operators of the S p i n ( 4 k + 2 ) \mathit{Spin}(4k+2) Spin ( 4 k + 2 ) SYM. At the n n n -th vacuum, the condensed particle has a vanishing Dirac pairing with the line operator of ( z ; w ) = ( ± n , 1 ) (z;w)=(\pm n,1) ( z ; w ) = ( ± n , 1 ) , where the sign depends on whether k k k is even or odd. One can distinguish different values of n n n only modulo 4 4 4 , corresponding to the SPT phases protected by the Z 4 \mathbb{Z}_4 Z 4 center symmetry of S p i n ( 4 k + 2 ) \mathit{Spin}(4k+2) Spin ( 4 k + 2 ) . When turning on the soft SUSY breaking mass term δ L = m Tr ( λ α λ α ) + c . c . \delta\mathcal{L}=m\operatorname{Tr}(\lambda_{\alpha}\lambda^{\alpha})+\mathrm{c.c.} δ L = m Tr ( λ α λ α ) + c.c. , we find the effective potential
V = − 2 ∣ m κ Λ 3 ∣ cos ( θ − 2 π n 4 k ) . V=-2|m\kappa\Lambda^{3}|\cos\left(\frac{\theta-2\pi n}{4k}\right). V = − 2∣ mκ Λ 3 ∣ cos ( 4 k θ − 2 π n ) .
Let us next consider the N = 1 \mathcal{N}=1 N = 1 S O ( 4 k + 2 ) + \mathrm{SO}(4k+2)_+ SO ( 4 k + 2 ) + SYM. While the number of supersymmetric vacua is the same as above, the IR theory existing at each vacuum is different. The reason for this is that the genuine line operators of the S O ( 4 k + 2 ) + \mathrm{SO}(4k+2)_+ SO ( 4 k + 2 ) + theory are those with ( z ; w ) = ( 2 ; 0 ) , ( 0 ; 2 ) (z;w)=(2;0),(0;2) ( z ; w ) = ( 2 ; 0 ) , ( 0 ; 2 ) and ( 2 ; 2 ) (2;2) ( 2 ; 2 ) . At the n n n -th vacuum, the line of ( z ; w ) = ( 0 ; 2 ) (z;w)=(0;2) ( z ; w ) = ( 0 ; 2 ) has a perimeter law if n n n is even, and the one with ( z ; w ) = ( 2 ; 2 ) (z;w)=(2;2) ( z ; w ) = ( 2 ; 2 ) has a perimeter law if n n n is odd, with the other lines having an area law. This means that at every vacuum the IR theory is a Z 2 \mathbb{Z}_2 Z 2 TQFT. When the soft SUSY breaking term is added, only one vacuum has the minimum energy for generic values of θ \theta θ . This result coincides with the second line of Table 10.
Finally, we consider the N = 1 \mathcal{N}=1 N = 1 S O ( 4 k + 2 ) − SO(4k+2)_{-} S O ( 4 k + 2 ) − SYM. The genuine lines are now those with ( z ; w ) = ( 2 , 0 ) , ( 1 , 2 ) (z;w)=(2,0),(1,2) ( z ; w ) = ( 2 , 0 ) , ( 1 , 2 ) and ( 3 , 2 ) (3,2) ( 3 , 2 ) . Since the charge lattice of these lines is generated by ( 1 , 2 ) (1,2) ( 1 , 2 ) , the electric and magnetic 1-form symmetries are now combined into Z 4 \mathbb{Z}_4 Z 4 . Since every genuine line is in an area law, these phases are the trivial and SPT phases protected by this Z 4 \mathbb{Z}_4 Z 4 . When the soft SUSY breaking mass is turned on, only a single vacuum is selected as the true vacuum, depending on the value of θ \theta θ . This result is in perfect agreement with the third line of Table 10.
Spin(2k+1)/SO(2k+1) Yang-Mills theory
In the case of S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Yang-Mills theory for k ≥ 2 k\geq2 k ≥ 2 , there is no mixed center/CP anomaly [16 , 17 , 18 , 4 ] on spin manifolds. Therefore, one cannot apply the technique of mixed gauge theory to these theories. Below, we simply provide the phase diagram of the soft broken S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± SYM for completeness. Its comparison to the phase diagram of non-supersymmetric S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Yang-Mills theories is left for future work.
We start with the N = 1 \mathcal{N}=1 N = 1 S p i n ( 2 k + 1 ) Spin(2k+1) S p in ( 2 k + 1 ) SYM for a positive integer k ≥ 2 k\geq2 k ≥ 2 . Line operators are labeled by ( z ; w ) ∈ Z 2 × Z 2 (z;w)\in\mathbb{Z}_2\times\mathbb{Z}_2 ( z ; w ) ∈ Z 2 × Z 2 , and the genuine line operator of the S p i n ( 2 k + 1 ) Spin(2k+1) S p in ( 2 k + 1 ) Yang-Mills theory is the Wilson line W 4 d W_{4\mathrm d} W 4 d in the spinor representation corresponding to ( z ; w ) = ( 1 ; 0 ) (z;w)=(1;0) ( z ; w ) = ( 1 ; 0 ) . The theory has ( 2 k − 1 ) (2k-1) ( 2 k − 1 ) SUSY vacua [15 , 16 ] characterized by
⟨ Tr ( λ α λ α ) ⟩ = κ Λ 3 e − 2 π i 2 k − 1 n , n = 0 , 1 , 2 , ⋯ , 2 k − 2 , (70) \langle\operatorname{Tr}(\lambda_\alpha\lambda^\alpha)\rangle=\kappa\Lambda^3 e^{-\frac{2\pi i}{2k-1}n},\qquad n=0,1,2,\cdots,2k-2, \tag*{(70)} ⟨ Tr ( λ α λ α )⟩ = κ Λ 3 e − 2 k − 1 2 π i n , n = 0 , 1 , 2 , ⋯ , 2 k − 2 , (70)
where κ \kappa κ is a constant. At the n n n -th vacuum, the electro-magnetic charge of the condensed particles is proportional to that of the line operator of ( z ; w ) = ( 2 n , 1 ) = ( 0 , 1 ) (z;w)=(2n,1)=(0,1) ( z ; w ) = ( 2 n , 1 ) = ( 0 , 1 ) up to a shift by the root lattice. Therefore the genuine line operator always has an area law, meaning that every vacuum is in the confining phase. A big difference from the S p i n ( even ) / S O ( even ) ± Spin(\text{even})/SO(\text{even})_{\pm} S p in ( even ) / S O ( even ) ± theories is that, even though the theory has a non-trivial center symmetry, one cannot distinguish vacua with different values of n n n by it. Indeed, the center Z 2 \mathbb{Z}_2 Z 2 symmetry maps the Wilson line in the spinor representation W 4 d W_{4\mathrm d} W 4 d as
W 4 d → − W 4 d , (71) W_{4\mathrm d}\to-W_{4\mathrm d}, \tag*{(71)} W 4 d → − W 4 d , (71)
but preserves the lines with ( z ; w ) = ( 2 n ; 1 ) (z;w)=(2n;1) ( z ; w ) = ( 2 n ; 1 ) invariant. Therefore, the above ( 2 k − 1 ) (2k-1) ( 2 k − 1 ) vacua are all in the trivially gapped phase, i.e., the confining phase.
When turning on the soft SUSY breaking term δ L = m Tr ( λ α λ α ) + c . c . \delta\mathcal{L}=m\operatorname{Tr}(\lambda_\alpha\lambda^\alpha)+\mathrm{c.c.} δ L = m Tr ( λ α λ α ) + c.c. , one finds the effective potential V = − 2 ∣ m κ Λ 3 ∣ cos ( θ − 2 π n 2 k − 1 ) V=-2|m\kappa\Lambda^3|\cos\left(\frac{\theta-2\pi n}{2k-1}\right) V = − 2∣ mκ Λ 3 ∣ cos ( 2 k − 1 θ − 2 π n ) at the n n n -th vacuum, which means a single true
caption: Table 11: The phase structure of the N = 1 \mathcal{N}=1 N = 1 S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± SYM with a small gaugino mass m m m turned on. Since there is no center/CP mixed anomaly, the two-fold degeneracy at θ = π \theta=\pi θ = π modulo 2 π 2\pi 2 π might be resolved when ∣ m ∣ \lvert m\rvert ∣ m ∣ is large.
:::
vacuum for generic values of θ \theta θ . Since all the ( 2 k − 1 ) (2k-1) ( 2 k − 1 ) vacua are in the trivially gapped phase, we find that the vacuum structure has the periodicity θ ∼ θ + 2 π \theta\sim\theta+2\pi θ ∼ θ + 2 π . This result is summarized in the first line of Table 11.
Let us now move on to the N = 1 \mathcal{N}=1 N = 1 S O ( 2 k + 1 ) + SO(2k+1)_{+} S O ( 2 k + 1 ) + SYM. Its genuine line operator is the one with ( z ; w ) = ( 0 ; 1 ) (z;w)=(0;1) ( z ; w ) = ( 0 ; 1 ) , which has a perimeter law at all vacua since whose electro-magnetic charge has a vanishing pairing with the charges of the condensed particles. Therefore, at every vacuum, we find a Z 2 \mathbb{Z}_2 Z 2 TQFT in the infrared. Thus we find the phase diagram of the softly broken S O ( 2 k + 1 ) + SO(2k+1)_{+} S O ( 2 k + 1 ) + SYM as in the second line of Table 11.
In the case of S O ( 2 k + 1 ) − SO(2k+1)_{-} S O ( 2 k + 1 ) − SYM, the genuine line operator is the one for ( z ; w ) = ( 1 , 1 ) (z;w)=(1,1) ( z ; w ) = ( 1 , 1 ) . Since this line has a non-vanishing paring with the electro-magnetic charge of condensed particles at every vacuum, all vacua of this SYM are in the confining phase.18 This implies the phase diagram shown in the third line of Table 11.
For all S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± SYM, an interesting situation occurs when θ / π \theta/\pi θ / π is an odd integer. In this case, the S p i n ( 2 k + 1 ) / S O ( 2 k + 1 ) ± Spin(2k+1)/SO(2k+1)_{\pm} S p in ( 2 k + 1 ) / S O ( 2 k + 1 ) ± SYM with a small gaugino mass has two degenerate vacua. In contrast to the S p i n ( e v e n ) / S O ( e v e n ) Spin(\mathrm{even})/SO(\mathrm{even}) S p in ( even ) / S O ( even ) cases, there is no mixed center/CP anomaly on spin manifolds that prevents this degeneracy from being resolved. Note that, since the phase diagram is periodic under θ → θ + 2 π \theta\to\theta+2\pi θ → θ + 2 π , the absence of phase transition at θ = ( 2 n + 1 ) π \theta=(2n+1)\pi θ = ( 2 n + 1 ) π for n ∈ Z n\in\mathbb{Z} n ∈ Z does not immediately lead to a contradiction. A possible way to study this degeneracy is to consider the mixed gauge theory on non-spin manifolds, on which there is a mixed anomaly between θ → θ + 2 π \theta\to\theta+2\pi θ → θ + 2 π and the center symmetry [18 ] . We leave a careful study of this subtlety for future work.
Summary and discussions
In this paper, we have studied the low-temperature phase diagrams of S p i n ( N ) / S O ( N ) ± Spin(N)/SO(N)_{\pm} S p in ( N ) / S O ( N ) ± Yang-Mills theories as functions of the θ \theta θ -parameter. In particular, for every even N ≥ 4 N \ge4 N ≥ 4 , we have identified a phase diagram consistent with the center/CP mixed anomaly via the technique of mixed gauge theory. For that, we have made an assumption that the S p i n ( N ) Spin(N) S p in ( N ) Yang-Mills theory is gapped for all values of the θ \theta θ parameter and has a trivial non-degenerate vacuum at θ = 0 \theta= 0 θ = 0 , as in most sections of [1 ] .
Among other cases, the N = 4 N = 4 N = 4 case is special in the sense that the theory has two independent θ \theta θ -parameters. Therefore we have started our analysis with the N = 4 N = 4 N = 4 case in Sec. 3. The resulting phase diagram is shown in Figure 4 and Table 6 . We have particularly argued that, for the S p i n ( 4 ) Spin(4) S p in ( 4 ) Yang-Mills theory, there are four degenerate vacua at intersections of dashed lines in Figure 4 .
As discussed in Sec. 4, its generalization to the S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± for k ∈ N k \in\mathbb{N} k ∈ N is straightforward, except that now the theory has only a single θ \theta θ -parameter and the center/CP anomaly for even k k k is slightly different from that for odd k k k . The phase diagrams of S p i n ( 4 k ) / S O ( 4 k ) ± Spin(4k)/SO(4k)_{\pm} S p in ( 4 k ) / S O ( 4 k ) ± Yang-Mills theories are shown in Figure 7 and Table 8 .
The S p i n ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Spin(4k + 2)/SO(4k + 2)_{\pm} S p in ( 4 k + 2 ) / S O ( 4 k + 2 ) ± Yang-Mills theories for k ∈ N k \in\mathbb{N} k ∈ N are essentially different from the above examples, since the center symmetry is now Z 4 \mathbb{Z}_4 Z 4 instead of Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 . As a result, in the S p i n ( 4 k + 2 ) Spin(4k + 2) S p in ( 4 k + 2 ) case, the phase diagram we have identified has periodicity θ ∼ θ + 8 π \theta\sim\theta+ 8\pi θ ∼ θ + 8 π , as shown in Figure 9 and Table 10.
For all the above gauge groups, we have also analyzed the phase diagrams of SYM with a small gaugino mass m m m turned on. The resulting phase diagrams turn out to be in perfect agreement with those of the non-supersymmetric Yang-Mills theories read off from the center/CP mixed anomaly. This strongly suggests that the phase diagram is independent of m m m . The small-∣ m ∣ |m| ∣ m ∣ analysis for the S p i n ( o d d ) / S O ( o d d ) ± Spin(\mathrm{odd})/SO(\mathrm{odd})_{\pm} S p in ( odd ) / S O ( odd ) ± Yang-Mills theories is also given in Sec. 6 for completeness.
The absence of a center/CP mixed anomaly for S p i n ( o d d ) Spin(\mathrm{odd}) S p in ( odd ) might suggest that the degeneracy of two vacua at θ = π \theta= \pi θ = π in the corresponding softly broken SYM can be resolved by quantum corrections. However, on non-spin manifolds, there is a mixed anomaly between the center symmetry and θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π for S p i n ( o d d ) Spin(\mathrm{odd}) S p in ( odd ) [18 ] . It would therefore be interesting to study these theories on non-spin manifolds and see this point further.
Another future direction is to test the phase diagrams obtained in this paper through numerical simulations. In the conventional Monte Carlo simulation the presence of the θ \theta θ -term gives rise to a sign problem, but there are approaches to overcome it [19 ] . It would also be interesting to apply the tensor network and quantum computing in the Hamiltonian formalism as they are free from the sign problem. Indeed, various numerical studies in the Hamiltonian formalism have been conducted on the two-dimensional Schwinger model with a non-vanishing θ \theta θ -parameter [20 ] .
Declaration of generative AI and AI-assisted technologies in the manuscript preparation process
During the preparation of this work, the authors used ChatGPT (OpenAI) for language editing and to help identify typos and potential inconsistencies in the manuscript. After using this tool, the authors critically evaluated the suggestions and independently verified the calculations and arguments underlying any revisions. The authors take full responsibility for the content of the article.
Acknowledgements
We are grateful to E. Itou, A. Matsumoto, K. Maruyoshi and S. Moriyama for illuminating discussions. The authors’ research is partially supported by JSPS KAKENHI Grant Number JP21H04993. In addition, T. N’s research is partially supported by JSPS KAKENHI Grant Numbers JP18K13547, JP23K03393, JP23K03394 and JP26H01997.
The spectrum of genuine line operators
In this appendix, we review the spectrum of genuine line operators in a four-dimensional Yang-Mills theory of gauge group G G G associated with a finite-dimensional real Lie algebra g \mathfrak{g} g , following [16 ] . We assume that G G G is connected, which means that there is a sub-group H H H of the center C C C of the simply-connected Lie group G ~ \widetilde{G} G integrating g \mathfrak{g} g such that
G = G ~ / H . (72) G = \widetilde{G}/H . \tag*{(72)} G = G / H . (72)
A genuine line operator is generally labeled by
( λ e , λ m ) ∈ Λ w ( g ) × Λ w ( g ∨ ) , (73) (\lambda_e,\lambda_m) \in\Lambda_w(\mathfrak{g}) \times\Lambda_w(\mathfrak{g}^{\vee}), \tag*{(73)} ( λ e , λ m ) ∈ Λ w ( g ) × Λ w ( g ∨ ) , (73)
where Λ w ( g ) \Lambda_w(\mathfrak{g}) Λ w ( g ) is the weight lattice of g \mathfrak{g} g , and g ∨ \mathfrak{g}^{\vee} g ∨ is the Langlands-dual of g \mathfrak{g} g . When g \mathfrak{g} g is a simply-laced simple Lie algebra, these two lattices are identical.
The spectrum of genuine line operators in the Yang-Mills theory depends not only on g \mathfrak{g} g but G G G . In particular, purely-electric genuine lines (i.e., Wilson lines) are those labeled by ( λ e , 0 ) (\lambda_e,0) ( λ e , 0 ) invariant under the action of H H H . This means that, if the gauge group is G ~ \widetilde{G} G then the line with ( λ e , 0 ) (\lambda_e,0) ( λ e , 0 ) for every λ e ∈ Λ w ( g ) \lambda_e \in\Lambda_w(\mathfrak{g}) λ e ∈ Λ w ( g ) is a genuine Wilson line, but if G ≄ G ~ G \not\simeq\widetilde{G} G ≃ G not all representations of g \mathfrak{g} g give genuine Wilson lines. The complete set L L L of genuine line operators, which also includes lines with non-vanishing magnetic charges, is then determined so that, for every pair L 1 , L 2 ∈ L \mathcal{L}_1,\mathcal{L}_2 \in L L 1 , L 2 ∈ L , their Dirac pairing ⟨ L 1 , L 2 ⟩ \langle\mathcal{L}_1,\mathcal{L}_2\rangle ⟨ L 1 , L 2 ⟩ vanishes.
Note that if L L L contains a line with ( λ e ( 1 ) , λ m ( 1 ) ) (\lambda_e^{(1)},\lambda_m^{(1)}) ( λ e ( 1 ) , λ m ( 1 ) ) and another line with ( λ e ( 2 ) , λ m ( 2 ) ) (\lambda_e^{(2)},\lambda_m^{(2)}) ( λ e ( 2 ) , λ m ( 2 ) ) then L L L also contains a line with ( λ e ( 1 ) + λ e ( 2 ) , λ m ( 1 ) + λ m ( 2 ) ) (\lambda_e^{(1)}+\lambda_e^{(2)},\lambda_m^{(1)}+\lambda_m^{(2)}) ( λ e ( 1 ) + λ e ( 2 ) , λ m ( 1 ) + λ m ( 2 ) ) . In addition, the charge conjugation implies that the presence of a line with ( λ e , λ m ) (\lambda_e,\lambda_m) ( λ e , λ m ) gives rise to another line with ( − λ e , − λ m ) (-\lambda_e,-\lambda_m) ( − λ e , − λ m ) . We also see that the Wilson in the adjoint representation of g \mathfrak{g} g and the ’t Hooft line in the adjoint representation of g ∨ \mathfrak{g}^{\vee} g ∨ always exist. Combining all these facts, we see that the spectrum of genuine line operators is essentially characterized by a subset of ( Λ w ( g ) / Λ r ( g ) ) × ( Λ w ( g ∨ ) / Λ r ( g ∨ ) ) ≃ C × C (\Lambda_w(\mathfrak{g})/\Lambda_r(\mathfrak{g}))\times(\Lambda_w(\mathfrak{g}^{\vee})/\Lambda_r(\mathfrak{g}^{\vee}))\simeq C\times C ( Λ w ( g ) / Λ r ( g )) × ( Λ w ( g ∨ ) / Λ r ( g ∨ )) ≃ C × C , where Λ r ( g ) \Lambda_r(\mathfrak{g}) Λ r ( g ) is the root lattice of g \mathfrak{g} g and we used the fact that the simply-connected Lie group integrating g ∨ \mathfrak{g}^{\vee} g ∨ has the same center as G ~ \widetilde{G} G . We therefore label genuine line operators in the theory by
( z e , z m ) ∈ C × C . (74) (z_e,z_m)\in C\times C . \tag*{(74)} ( z e , z m ) ∈ C × C . (74)
In the rest of this appendix, we focus on the cases of G = S p i n ( N ) G=\mathit{Spin}(N) G = Spin ( N ) and G = S O ( N ) G=\mathit{SO}(N) G = SO ( N ) . Then the center C C C is Z 2 \mathbb{Z}_2 Z 2 when N N N is odd, Z 2 × Z 2 \mathbb{Z}_2\times\mathbb{Z}_2 Z 2 × Z 2 when N / 2 N/2 N /2 is an even integer, and Z 4 \mathbb{Z}_4 Z 4 when N / 2 N/2 N /2 is an odd integer. In the case of G = S p i n ( N ) G=\mathit{Spin}(N) G = Spin ( N ) , genuine lines are those of ( z e , 0 ) (z_e,0) ( z e , 0 ) for all z e ∈ C z_e\in C z e ∈ C . In the case of G = S O ( N ) G=\mathit{SO}(N) G = SO ( N ) , the spectrum of genuine lines is identified by noting that
S O ( N ) = S p i n ( N ) / Z 2 , (75) \mathit{SO}(N)=\mathit{Spin}(N)/\mathbb{Z}_2 , \tag*{(75)} SO ( N ) = Spin ( N ) / Z 2 , (75)
where Z 2 \mathbb{Z}_2 Z 2 is a particular sub-group of C C C . At the level of the charges of genuine line operators, this quotient has two effects:
The allowed values of z e z_e z e are restricted to those that are invariant under Z 2 \mathbb{Z}_2 Z 2 .
Extra lines with non-vanishing z m z_m z m can be included in the spectrum.
There are two different options for the set of “extra lines” with non-vanishing z m z_m z m , corresponding to S O ( N ) + \mathit{SO}(N)_+ SO ( N ) + and S O ( N ) − \mathit{SO}(N)_- SO ( N ) − Yang-Mills theories.
Spin(2k+1)/SO(2k+1) case
Let us describe the cases of G = S p i n ( 2 k + 1 ) G = Spin(2k+1) G = S p in ( 2 k + 1 ) and G = S O ( 2 k + 1 ) G = SO(2k+1) G = S O ( 2 k + 1 ) for a positive integer k k k in more detail. In this case, since C = Z 2 C = \mathbb{Z}_2 C = Z 2 , genuine line operators are labeled by
( z e , z m ) ∈ Z 2 × Z 2 . (76) (z_e,z_m) \in\mathbb{Z}_2 \times\mathbb{Z}_2. \tag*{(76)} ( z e , z m ) ∈ Z 2 × Z 2 . (76)
For a line with ( z e ( 1 ) , z m ( 1 ) ) (z_e^{(1)},z_m^{(1)}) ( z e ( 1 ) , z m ( 1 ) ) and another line with ( z e ( 2 ) , z m ( 2 ) ) (z_e^{(2)},z_m^{(2)}) ( z e ( 2 ) , z m ( 2 ) ) , the vanishing of their Dirac pairing is equivalent to
z e ( 1 ) z m ( 2 ) − z m ( 1 ) z e ( 2 ) = 0 m o d 2. (77) z_e^{(1)}z_m^{(2)} - z_m^{(1)}z_e^{(2)} = 0 \mod2. \tag*{(77)} z e ( 1 ) z m ( 2 ) − z m ( 1 ) z e ( 2 ) = 0 mod 2. (77)
When G = S p i n ( 2 k + 1 ) G = Spin(2k+1) G = S p in ( 2 k + 1 ) , the only non-trivial genuine line is the one for ( z e , z m ) = ( 1 , 0 ) (z_e,z_m) = (1,0) ( z e , z m ) = ( 1 , 0 ) , i.e., the Wilson line in the fundamental representation of S p i n ( 2 k + 1 ) Spin(2k+1) S p in ( 2 k + 1 ) . When G = S O ( 2 k + 1 ) ≃ S p i n ( 2 k + 1 ) / Z 2 G = SO(2k+1) \simeq Spin(2k+1)/\mathbb{Z}_2 G = S O ( 2 k + 1 ) ≃ S p in ( 2 k + 1 ) / Z 2 , this line is no longer genuine since it is not invariant under Z 2 \mathbb{Z}_2 Z 2 . Therefore, there is no purely-electric non-trivial genuine line operator in the S O ( 2 k + 1 ) SO(2k+1) S O ( 2 k + 1 ) Yang-Mills theory. Instead, one can include an extra line with non-vanishing z m z_m z m . We call the theory the S O ( 2 k + 1 ) + SO(2k+1)_+ S O ( 2 k + 1 ) + Yang-Mills theory when the line with ( z e , z m ) = ( 0 , 1 ) (z_e,z_m) = (0,1) ( z e , z m ) = ( 0 , 1 ) is included in the spectrum, and the S O ( 2 k + 1 ) − SO(2k+1)_- S O ( 2 k + 1 ) − Yang-Mills theory when the one with ( z e , z m ) = ( 1 , 1 ) (z_e,z_m) = (1,1) ( z e , z m ) = ( 1 , 1 ) is included. Note that one cannot include further extra lines without breaking the condition (A.6).
The above discussion tells us about the allowed electro-magnetic charges ( λ e , λ m ) (\lambda_e,\lambda_m) ( λ e , λ m ) of genuine line operators. For example, when G = S p i n ( 3 ) ≃ S U ( 2 ) G = Spin(3) \simeq SU(2) G = S p in ( 3 ) ≃ S U ( 2 ) , the allowed charges are shown in the left picture of Figure 12 . We see from this picture that the S p i n ( 3 ) Spin(3) S p in ( 3 ) Yang-Mills theory has an electric Z 2 \mathbb{Z}_2 Z 2 1-form symmetry whose generator assigns the phase factor e π i λ e = e π i z e e^{\pi i\lambda_e} = e^{\pi i z_e} e π i λ e = e π i z e to the line operator associated with ( λ e , λ m ) (\lambda_e,\lambda_m) ( λ e , λ m ) . When G = S O ( 3 ) + G = SO(3)_+ G = S O ( 3 ) + , the allowed values of ( λ e , λ m ) (\lambda_e,\lambda_m) ( λ e , λ m ) are now as shown in the middle picture of Figure 12 . The 1-form symmetry is now a magnetic Z 2 \mathbb{Z}_2 Z 2 symmetry whose generator assigns e π i λ m = e π i z m e^{\pi i\lambda_m} = e^{\pi i z_m} e π i λ m = e π i z m to line operators. The spectrum of genuine lines of S O ( 3 ) − SO(3)_- S O ( 3 ) − Yang-Mills theory is shown in the right picture of Figure 12 . We see that there is now a Z 2 \mathbb{Z}_2 Z 2 1-form symmetry whose generator assigns e π i λ e = e π i z e e^{\pi i\lambda_e} = e^{\pi i z_e} e π i λ e = e π i z e to the lines.19 We summarize these lines that transform non-trivially under the 1-form symmetry in Table 12 .
Figure 12. The electro-magnetic charges of genuine line operators in S p i n ( 3 ) ≃ S U ( 2 ) Spin(3) \simeq SU(2) S p in ( 3 ) ≃ S U ( 2 ) and S O ( 3 ) ± SO(3)_{\pm} S O ( 3 ) ± Yang-Mills theories.
Table 12. The list of non-trivial genuine lines and the 1-form symmetry in the case of g = s o ( 2 k + 1 ) \mathfrak{g}=\mathfrak{so}(2k+1) g = so ( 2 k + 1 ) (k ≥ 2 k \ge2 k ≥ 2 )
Spin(N)/SO(N) for even N
Here, we briefly describe the genuine line operators in S p i n ( N ) Spin(N) S p in ( N ) and S O ( N ) ± SO(N)_\pm S O ( N ) ± Yang-Mills theories for even N N N , generalizing the above discussions.
We start with the case of N = 4 k + 2 N=4k+2 N = 4 k + 2 for a positive integer k k k , in which case C = Z 4 C=\mathbb{Z}_4 C = Z 4 and therefore line operators are classified by ( z e , z m ) ∈ Z 4 × Z 4 (z_e,z_m)\in\mathbb{Z}_4\times\mathbb{Z}_4 ( z e , z m ) ∈ Z 4 × Z 4 . The vanishing condition of the Dirac pairing is now
z e ( 1 ) z m ( 2 ) − z m ( 1 ) z e ( 2 ) = 0 ( m o d 4 ) . (78) z_e^{(1)}z_m^{(2)}-z_m^{(1)}z_e^{(2)}=0 \pmod{4}. \tag*{(78)} z e ( 1 ) z m ( 2 ) − z m ( 1 ) z e ( 2 ) = 0 ( mod 4 ) . (78)
When G = S p i n ( 4 k + 2 ) G=Spin(4k+2) G = S p in ( 4 k + 2 ) , non-trivial genuine lines are those of ( z e , z m ) = ( 1 , 0 ) , ( 2 , 0 ) (z_e,z_m)=(1,0),(2,0) ( z e , z m ) = ( 1 , 0 ) , ( 2 , 0 ) and ( 3 , 0 ) (3,0) ( 3 , 0 ) . There is a 1-form Z 4 \mathbb{Z}_4 Z 4 symmetry whose generator assigns e π i z e / 2 e^{\pi i z_e/2} e π i z e /2 to the lines. When G = S O ( 4 k + 2 ) + G=SO(4k+2)_+ G = S O ( 4 k + 2 ) + , non-trivial genuine lines are those of ( z e , z m ) = ( 2 , 0 ) , ( 0 , 2 ) (z_e,z_m)=(2,0),(0,2) ( z e , z m ) = ( 2 , 0 ) , ( 0 , 2 ) and ( 2 , 2 ) (2,2) ( 2 , 2 ) . The 1-form symmetry is now Z 2 × Z 2 \mathbb{Z}_2\times\mathbb{Z}_2 Z 2 × Z 2 whose generators assign e π i z e / 2 e^{\pi i z_e/2} e π i z e /2 and e π i z m / 2 e^{\pi i z_m/2} e π i z m /2 to the lines. When G = S O ( 4 k + 2 ) − G=SO(4k+2)_- G = S O ( 4 k + 2 ) − , non-trivial genuine lines are those of ( z e , z m ) = ( 2 , 0 ) , ( 1 , 2 ) (z_e,z_m)=(2,0),(1,2) ( z e , z m ) = ( 2 , 0 ) , ( 1 , 2 ) and ( 3 , 2 ) (3,2) ( 3 , 2 ) . The 1-form symmetry is Z 4 \mathbb{Z}_4 Z 4 whose generator assigns e π i ( z m − z e ) / 2 e^{\pi i(z_m-z_e)/2} e π i ( z m − z e ) /2 to the line associated with ( z e , z m ) (z_e,z_m) ( z e , z m ) .20 These results are summarized in Table 13 .
Table 13. The spectrum of non-trivial genuine line operators for g = s o ( 4 k + 2 ) \mathfrak{g}=\mathfrak{so}(4k+2) g = so ( 4 k + 2 ) ( k ≥ 1 ) (k \ge1) ( k ≥ 1 ) and the corresponding 1-form symmetry.
In the case of N = 4 k N=4k N = 4 k for a positive integer k k k , we have C = Z 2 S × Z 2 C C=\mathbb{Z}_2^{S}\times\mathbb{Z}_2^{C} C = Z 2 S × Z 2 C and therefore line operators are characterized by
( z e , S , z e , C ; z m , S , z m , C ) ∈ ( Z 2 ) 4 . (79) (z_{e,S},z_{e,C};z_{m,S},z_{m,C}) \in(\mathbb{Z}_2)^4 . \tag*{(79)} ( z e , S , z e , C ; z m , S , z m , C ) ∈ ( Z 2 ) 4 . (79)
The vanishing condition for the Dirac pairing is
z e , S ( 1 ) z m , S ( 2 ) − z m , S ( 1 ) z e , S ( 2 ) + z e , C ( 1 ) z m , C ( 2 ) − z m , C ( 1 ) z e , C ( 2 ) = 0 ( m o d 2 ) (80) z_{e,S}^{(1)}z_{m,S}^{(2)}-z_{m,S}^{(1)}z_{e,S}^{(2)}+z_{e,C}^{(1)}z_{m,C}^{(2)}-z_{m,C}^{(1)}z_{e,C}^{(2)}=0 \pmod{2} \tag*{(80)} z e , S ( 1 ) z m , S ( 2 ) − z m , S ( 1 ) z e , S ( 2 ) + z e , C ( 1 ) z m , C ( 2 ) − z m , C ( 1 ) z e , C ( 2 ) = 0 ( mod 2 ) (80)
for any integer k k k . Note that this notation is slightly different from the one used in [16 ] when k k k is even. As a result, the Witten effect under θ → θ + 2 π \theta\to\theta+ 2\pi θ → θ + 2 π is written as
( z e , S , z e , C ; z m , S , z m , C ) → ( z e , S + z m , S , z e , C + z m , C ; z m , S , z m , C ) (81) (z_{e,S},z_{e,C};z_{m,S},z_{m,C}) \to(z_{e,S}+z_{m,S},z_{e,C}+z_{m,C};z_{m,S},z_{m,C}) \tag*{(81)} ( z e , S , z e , C ; z m , S , z m , C ) → ( z e , S + z m , S , z e , C + z m , C ; z m , S , z m , C ) (81)
when k k k is odd, while it is expressed as
( z e , S , z e , C ; z m , S , z m , C ) → ( z e , S + z m , C , z e , C + z m , S ; z m , S , z m , C ) (82) (z_{e,S},z_{e,C};z_{m,S},z_{m,C}) \to(z_{e,S}+z_{m,C},z_{e,C}+z_{m,S};z_{m,S},z_{m,C}) \tag*{(82)} ( z e , S , z e , C ; z m , S , z m , C ) → ( z e , S + z m , C , z e , C + z m , S ; z m , S , z m , C ) (82)
when k k k is even.
When G = S p i n ( 4 k ) G=Spin(4k) G = S p in ( 4 k ) , non-trivial genuine lines are those for ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 0 ; 0 , 0 ) , ( 0 , 1 ; 0 , 0 ) (z_{e,S},z_{e,C};z_{m,S},z_{m,C})=(1,0;0,0),(0,1;0,0) ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 0 ; 0 , 0 ) , ( 0 , 1 ; 0 , 0 ) and ( 1 , 1 ; 0 , 0 ) (1,1;0,0) ( 1 , 1 ; 0 , 0 ) . The 1-form symmetry is Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 whose generators assign e π i z e , S e^{\pi i z_{e,S}} e π i z e , S and e π i z e , C e^{\pi i z_{e,C}} e π i z e , C to the lines. When G = S O ( 4 k ) + G=SO(4k)_+ G = S O ( 4 k ) + , genuine lines are those for ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 1 ; 0 , 0 ) , ( 0 , 0 ; 1 , 1 ) (z_{e,S},z_{e,C};z_{m,S},z_{m,C})=(1,1;0,0),(0,0;1,1) ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 1 ; 0 , 0 ) , ( 0 , 0 ; 1 , 1 ) and ( 1 , 1 ; 1 , 1 ) (1,1;1,1) ( 1 , 1 ; 1 , 1 ) . The 1-form symmetry in this case is Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 whose generators assign e π i z e , S e^{\pi i z_{e,S}} e π i z e , S and e π i z m , S e^{\pi i z_{m,S}} e π i z m , S to the lines. Finally, when G = S O ( 4 k ) − G = SO(4k)_- G = S O ( 4 k ) − , genuine lines are those for ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 1 ; 0 , 0 ) (z_{e,S}, z_{e,C}; z_{m,S}, z_{m,C}) = (1,1;0,0) ( z e , S , z e , C ; z m , S , z m , C ) = ( 1 , 1 ; 0 , 0 ) , ( 1 , 0 ; 1 , 1 ) (1,0;1,1) ( 1 , 0 ; 1 , 1 ) and ( 0 , 1 ; 1 , 1 ) (0,1;1,1) ( 0 , 1 ; 1 , 1 ) . The 1-form symmetry is again Z 2 × Z 2 \mathbb{Z}_2 \times\mathbb{Z}_2 Z 2 × Z 2 whose generators assign e π i z e , S e^{\pi i z_{e,S}} e π i z e , S and e π i z e , C e^{\pi i z_{e,C}} e π i z e , C to the lines. These results are summarized in Table 14 .
Table 14. The spectrum of genuine line operators for g = s o ( 4 k ) \mathfrak{g}=\mathfrak{so}(4k) g = so ( 4 k ) ( k ≥ 1 ) (k \ge1) ( k ≥ 1 ) , and the corresponding 1-form symmetry.