Pure Super-Yang–Mills Vacua and Domain Walls

Pure four-dimensional $\mathcal N=1$ super-Yang–Mills with a simple, simply connected gauge group $G$ has $h^\vee$ supersymmetric vacuum sectors, where $h^\vee$ is the dual Coxeter number. The robust organizing data are the anomaly-reduced discrete R-symmetry, the phase of the gaugino condensate, the center one-form symmetry, and the domain-wall central charge. A mass gap, confinement, BPS saturation, and a proposed wall topological theory are logically distinct statements.

Required background. Supersymmetric Yang–Mills actions fixes the microscopic multiplet and action. R-symmetry, anomalies, and the holomorphic scale supplies the anomalous Ward identity. BPS solitons and walls supplies the wall central charge and saturation criterion.

Helpful background. Higher-form symmetry from operators and linking clarifies how the global form changes genuine lines.

A global-form-specific theory card

Let $G$ be compact, connected, simple, and simply connected, with Lie algebra $\mathfrak g$ and dual Coxeter number $h^\vee$. The only propagating multiplet is the vector multiplet $(A_\mu,\lambda_\alpha,D)$ in the adjoint. There is no matter and no tree superpotential. Define $\operatorname{Tr}W^2\equiv W^{a\alpha}W^a_\alpha$ using the basic invariant form. For $SU(N)$ this is $2\operatorname{tr}_F W^2$ when $\operatorname{tr}_F(T^aT^b)=\delta^{ab}/2$. With this convention,

$$\mathcal L\supset \int d^2\theta\,\frac{\tau}{16\pi i}\, \operatorname{Tr}W^\alpha W_\alpha +\text{h.c.}, \qquad \tau=\frac{\theta}{2\pi}+\frac{4\pi i}{g_h^2}.$$

The holomorphic scale is defined by

$$\Lambda^{3h^\vee}=\mu^{3h^\vee}e^{2\pi i\tau(\mu)}.$$

The power is invariant, while the phase of $\Lambda^3$ labels a choice of root. A change of trace or renormalization scheme rescales $\Lambda$ and the normalized condensate together.

Because all dynamical fields are adjoint, the simply connected theory has electric one-form symmetry $Z(G)$. Wilson lines with nontrivial center charge are genuine and cannot end on the gaugino. Choosing $G/\Gamma$ instead gauges part of this symmetry and introduces different magnetic lines and, in general, discrete theta data. The vacuum and wall formulas below are therefore not silently statements about every global form.

From the anomaly to $h^\vee$ sectors

Classically, gaugino number is a $U(1)_R$ symmetry with $R(\lambda)=1$. A unit instanton has $2h^\vee$ adjoint fermion zero modes. With instanton weight $e^{+i\theta}$, the Grassmann measure transforms under $\lambda\mapsto e^{i\alpha}\lambda$ by $e^{-2ih^\vee\alpha}$. Only

$$\mathbb Z_{2h^\vee}\subset U(1)_R$$

survives quantum mechanically. Fermion parity is the order-two subgroup and cannot be spontaneously broken. A nonzero bilinear $\langle\lambda\lambda\rangle$, which has R-charge two, realizes

$$\mathbb Z_{2h^\vee}\longrightarrow\mathbb Z_2$$

and gives $h^\vee$ phases. With a condensate superfield $S$ normalized on the next page, one may label them

$$\langle S\rangle_k=\Lambda^3e^{2\pi i k/h^\vee}, \qquad k=0,1,\ldots,h^\vee-1.$$

The index calculation has an important history. For $SU(N)$ and $USp(2N)$, Witten’s original small-$T^3$ calculation gives $r+1=h^\vee$, where $r$ is the rank. For $\operatorname{Spin}(N\geq7)$ and the exceptional groups, restricting the moduli space of commuting triples to its identity component gives only $r+1$. Additional disconnected components supply the missing states, and summing all components gives $h^\vee$ Witten 2000, § 3.1, pp. 12–14 and § 4.2, pp. 41–45, especially Eq. (4.14), arXiv PDF and Kac and Smilga 1999, § 1, pp. 1–4, arXiv PDF.

Interpreting this compactified index as $h^\vee$ massive vacua on $\mathbb R^3$ assumes that no vacuum escapes to infinity and that the infrared theory has the standard gap. The anomaly identifies the allowed symmetry-breaking phases, while the complete index independently counts supersymmetric ground states under these hypotheses. Witten’s original calculation remains the historical starting point Witten 1982, §§ 4–5, pp. 277–299; a concise relation to the discrete symmetry appears in Intriligator and Seiberg 1996, § 1.2, pp. 3–4, arXiv PDF.

For reference, $h^\vee=N$ for $SU(N)$, $N+1$ for $USp(2N)$, and $N-2$ for the Lie algebra $\mathfrak{so}(N)$ with $N\geq5$. The corresponding simply connected orthogonal group is $\operatorname{Spin}(N)$; low-rank cases obey familiar isomorphisms, and $\operatorname{Spin}(4)$ is not simple. For $G_2,F_4,E_6,E_7,E_8$, the values are $4,9,12,18,30$. A Lie-algebra label alone does not specify the global form.

What is exact and what is dynamical

Four levels of conclusion should not be merged.

1. Index and anomaly constraints. The complete index counts $h^\vee$ supersymmetric ground states under its compactification hypotheses. The anomalous discrete symmetry fixes the allowed condensate phases, but not a nonzero magnitude.
2. Holomorphic condensate data. Holomorphic decoupling and controlled compactified calculations establish a nonzero condensate in a fixed operator-and-scale convention. Once established, its $h^\vee$ branches and their $\theta$ monodromy are protected chiral data.
3. Nonholomorphic dynamics. A mass gap, confinement, unbroken center realization, and the spectrum are additional dynamical statements. Compactified semiclassics and other evidence support the standard picture, but a holomorphic superpotential does not compute the four-dimensional mass spectrum.
4. Global-form refinement. When $Z(G)$ is nontrivial, coupling its one-form symmetry to a two-form background can distinguish vacuum-dependent topological responses. Gauging a center subgroup can split sectors or add topological degeneracy; the answer depends on the quotient and discrete theta angle. Centerless $G_2,F_4$, and $E_8$ still have $h^\vee$ chiral vacua, so the number of vacua is not the order of the center.

For a visual synthesis of condensate branches, exact superpotentials, and their limits, see the quantum-moduli, superpotential, and confinement flow.

Thus the equality “$h^\vee$ vacua” is cleanest for the simply connected theory on $\mathbb R^{3,1}$. It is not a license to erase line operators, spacetime topology, or discrete gauge sectors.

Domain walls and the central charge

Distinct discrete vacua define wall sectors. Take a planar wall normal to $x^3$, approaching vacuum $k$ at $x^3\to-\infty$ and vacuum $\ell$ at $x^3\to+\infty$. The four-dimensional $\mathcal N=1$ algebra admits a two-form central extension, and for the conventional normalization of the low-energy superpotential the tension obeys

$$T_{k\to\ell}\geq2\left|\Delta W\right|, \qquad \Delta W=W_\ell-W_k.$$

A wall that preserves half of the supercharges saturates the inequality. If the branch superpotential is normalized as $W_k=h^\vee\Lambda^3e^{2\pi i k/h^\vee}$, a BPS wall would have

$$T_{k\to\ell}^{\rm BPS} =4h^\vee|\Lambda^3| \left|\sin\frac{\pi(\ell-k)}{h^\vee}\right|.$$

The sign and phase of $\Delta W$ select which linear combination of supercharges is preserved. Exchanging the two endpoints sends $\Delta W\mapsto-\Delta W$ and shifts its phase by $\pi$; a CPT-conjugate wall additionally complex-conjugates the vacuum data. Dimensional analysis checks the result: a wall tension has mass dimension three.

The algebra fixes the bound, not the existence or multiplicity of BPS solutions. Nor does the Veneziano–Yankielowicz superpotential determine a reliable wall profile, because its Kähler potential and extra massive degrees of freedom are not fixed. Modern proposals for the wall infrared TQFT pass nontrivial partition-function and anomaly tests, but they remain dynamical proposals rather than consequences of the central charge alone; this status is explicit in Delmastro and Gomis 2021, abstract and § 1, arXiv v2.

$SU(N)$ as the canonical example

For simply connected $SU(N)$,

$$\mathbb Z_{2N}^{R}\to\mathbb Z_2, \qquad k\in\mathbb Z_N, \qquad \Gamma^{(1)}=\mathbb Z_N.$$

The $k$ labels may be viewed as branches related by a $2\pi$ shift of $\theta$: increasing $\theta$ by $2\pi$ cyclically permutes them. The theory as a whole is $2\pi$-periodic even though an individual branch is not. A wall between $k$ and $k+n$ is an $n$-wall; $n$ and $N-n$ are exchanged by reversing orientation. The proposed worldvolume topological data depend on $n$, on the bulk global form, and on which one-form background is turned on.

For $PSU(N)=SU(N)/\mathbb Z_N$, fundamental Wilson lines are absent and magnetic lines become genuine. On a four-manifold with nonzero obstruction to lifting the bundle to $SU(N)$, or with suitable line-defect or boundary data, allowed bundles can have fractional instanton number. Theta periodicity and discrete theta labels then differ. On ordinary $S^4$ or $\mathbb R^4$ without such data, this obstruction vanishes. One must redo the vacuum-sector analysis rather than recycling the simply connected line table.

Common pitfalls

The index is not the condensate. A nonzero index prevents complete supersymmetry breaking under its hypotheses. It does not calculate $\langle\lambda\lambda\rangle$ or prove a mass gap.

The BPS bound is not a wall solution. The algebra supplies $T\geq2|\Delta W|$. Saturation, multiplicity, binding, and the wall worldvolume theory require dynamical input.

The Lie algebra is not the theory. $SU(N)$ and $PSU(N)$ share $\mathfrak{su}(N)$ and local fields but differ in genuine lines, one-form symmetry, bundle sectors, and theta data.

Exercises

1. Assuming a BPS wall exists, start from $W_k=N\Lambda^3e^{2\pi ik/N}$ and derive its central-charge tension between vacua separated by $n$ units.

Solution

Use

$$e^{2\pi i(k+n)/N}-e^{2\pi ik/N} =2i\,e^{\pi i(2k+n)/N}\sin\frac{\pi n}{N}.$$

Its magnitude is $2|\sin(\pi n/N)|$. Multiplication by $2$ from the BPS bound and by $N|\Lambda^3|$ gives

$$T_n^{\rm BPS}=4N|\Lambda^3|\left|\sin\frac{\pi n}{N}\right|.$$

2. Explain why adjoint matter leaves a center one-form symmetry but fundamental matter does not.

Solution

The center acts trivially on the adjoint, so no local adjoint field can terminate a Wilson line with nonzero center charge. A fundamental field has unit center charge and can sit at such an endpoint, screening the line. Hence pure SYM retains $Z(G)$ as an electric one-form symmetry for simply connected $G$, whereas SQCD with fundamentals does not.

References

• Delmastro, Diego, and Jaume Gomis. “Domain Walls in 4d $\mathcal N=1$ Supersymmetric Yang–Mills.” Journal of High Energy Physics 2021, no. 3 (2021): 259. DOI. Open PDF, arXiv v2.
• Intriligator, Kenneth, and Nathan Seiberg. “Lectures on Supersymmetric Gauge Theories and Electric-Magnetic Duality.” Nuclear Physics B - Proceedings Supplements 45BC (1996): 1–28. doi:10.1016/0920-5632(95)00626-5. Open PDF.
• Kac, Victor G., and Alexei V. Smilga. “Vacuum Structure in Supersymmetric Yang–Mills Theories with Any Gauge Group.” Nuclear Physics B 571 (2000): 515–554. DOI. Open PDF.
• Witten, Edward. “Constraints on Supersymmetry Breaking.” Nuclear Physics B 202 (1982): 253–316. doi:10.1016/0550-3213(82)90071-2.
• Witten, Edward. “Supersymmetric Index in Four-Dimensional Gauge Theories.” Advances in Theoretical and Mathematical Physics 5, no. 5 (2001): 841–907. doi:10.4310/ATMP.2001.v5.n5.a1. Open PDF.