Three-Dimensional N=4 Mirror Symmetry

Three-dimensional $\mathcal N=4$ mirror symmetry identifies distinct ultraviolet gauge theories whose infrared fixed point is the same after exchanging the two R-symmetry factors. It sends Higgs-branch operators to Coulomb-branch monopoles, ordinary flavor currents to topological currents, and real masses to FI parameters. The exchange is powerful precisely because geometry that is quantum on one side is a classical hyperkähler quotient on the other.

Required background. We use monopole operators and quantum Coulomb branches, the construction of hyperkähler quotients, and the standards for duality claims, dictionaries, and evidence.

What mirror symmetry exchanges

An $\mathcal N=4$ superconformal fixed point has algebra $OSp(4|4)$ and R symmetry $SU(2)_H\times SU(2)_C$. Higgs-branch scalar primaries transform under $SU(2)_H$; Coulomb-branch scalar primaries, including monopoles, transform under $SU(2)_C$. A mirror pair $A\leftrightarrow B$ obeys

$$SU(2)_H^A\longleftrightarrow SU(2)_C^B, \qquad SU(2)_C^A\longleftrightarrow SU(2)_H^B.$$

Consequently,

$$\mathcal M_H(A)\simeq\mathcal M_C(B), \qquad \mathcal M_C(A)\simeq\mathcal M_H(B),$$

as hyperkähler cones at the conformal point, including their operator-ring and symmetry actions. This is stronger than equality of dimensions. A mere dimension match cannot distinguish different singularities or global symmetry enhancements.

Masses are background vector-multiplet scalars for Higgs-branch flavor symmetries, while FI parameters determine background scalars for Coulomb/topological symmetries. With the chapter convention $-\zeta D$ and $m_J=-2\pi\zeta$, mirror symmetry therefore demands

$$m_A\longleftrightarrow m_{J,B}=-2\pi\zeta_B, \qquad m_{J,A}=-2\pi\zeta_A\longleftrightarrow m_B.$$

The triplet nature of each deformation is visible only in full $\mathcal N=4$ notation. Writing one real component chooses an $\mathcal N=2$ subalgebra.

The Abelian A-series mirror pair

For an integer $N\ge1$, the standard family makes every part of the dictionary explicit Intriligator and Seiberg 1996, §3. It is useful to write complete $\mathcal N=2$ theory cards, because the neutral adjoint chirals and their superpotentials impose the complex moment-map relations.

Theory A: SQED with $N$ hypermultiplets. The gauge group is $U(1)$. Each hypermultiplet is a pair $(Q_i,\widetilde Q_i)$ of gauge charges $(+1,-1)$, and the vector multiplet contains a neutral chiral $\varphi$. At the conformal origin,

$$W_A=\varphi\sum_{i=1}^{N}Q_i\widetilde Q_i, \qquad m_i=\zeta_A=0.$$

Theory B: the gauge-fixed affine $A_{N-1}$ quiver. Its gauge group is

$$\frac{U(1)^N}{U(1)_{\rm diag}}\simeq U(1)^{N-1}.$$

In a linear gauge-fixed basis it has $N$ hypermultiplets $(X_i,Y_i)$ with charge matrix

$$S_i{}^a=\delta_{i,a}-\delta_{i,a+1}, \qquad a=1,\ldots,N-1,\qquad i=1,\ldots,N,$$

for $X_i$, and the opposite charges for $Y_i$. Thus the first and last hypers are end fundamentals and the intermediate ones are bifundamentals. With neutral chirals $\Phi_a$ in the vector multiplets,

$$W_B=\sum_{a=1}^{N-1}\Phi_a \left(X_aY_a-X_{a+1}Y_{a+1}\right).$$

Removing the diagonal vector is part of the theory definition, not a later simplification: retaining it would add a free vector multiplet and an extra topological current.

Their quaternionic dimensions already cross:

$$\begin{array}{c|cc} &\dim_{\mathbb H}\mathcal M_H&\dim_{\mathbb H}\mathcal M_C\\ \hline A&N-1&1\\ B&1&N-1 \end{array}$$

For Theory A, the Higgs branch is the hyperkähler quotient $\mathbb H^N///U(1)$. At a generic FI triplet it is the smooth resolution $T^*\mathbb{CP}^{N-1}$; for $N\ge2$, its zero-FI limit is the closure of the minimal nilpotent orbit of $\mathfrak{sl}_N$. Its Coulomb branch is quantum corrected to the $A_{N-1}$ singularity $\mathbb C^2/\mathbb Z_N$. Theory B realizes these geometries in the opposite order: its Higgs branch is $\mathbb C^2/\mathbb Z_N$, while its quantum Coulomb branch reproduces the Higgs cone of Theory A.

The coordinate-ring square

The branch exchange can be checked generator by generator. For Theory A, let $V_\pm$ be the flux-$\pm1$ monopoles. The exact Coulomb ring in one complex structure is

$$\mathbb C[\mathcal M_C(A)] =\frac{\mathbb C[\varphi,V_+,V_-]} {(V_+V_- - \varphi^N)}.$$

The monopole dimension formula gives $\Delta(V_\pm)=N/2$, while $\Delta(\varphi)=1$, so the relation is homogeneous Cremonesi, Hanany, and Zaffaroni 2014, §3.1.

On the Higgs branch of Theory B, the $\Phi_a$ F-terms imply

$$X_1Y_1=X_2Y_2=\cdots=X_NY_N\equiv Z.$$

The remaining complexified gauge quotient is generated by

$$X=\prod_{i=1}^{N}X_i, \qquad Y=\prod_{i=1}^{N}Y_i, \qquad Z=X_iY_i,$$

and these invariants obey $XY=Z^N$. Hence the protected map can be normalized as

$$\varphi\longleftrightarrow Z, \qquad V_+\longleftrightarrow X, \qquad V_-\longleftrightarrow Y.$$

The opposite half of the square is equally structural. Theory A has meson matrix $M_i{}^j=Q_i\widetilde Q^j$ with $\operatorname{Tr}M=0$ and $\operatorname{rank}M\le1$; equivalently $M^2=0$. These Higgs-branch moment maps correspond to Theory-B Coulomb monopoles and vector scalars organized into the enhanced $PSU(N)$ current multiplet. Thus mirror symmetry matches coordinate rings and symmetry actions, not merely the two quaternionic dimensions.

Parameters, currents, and probes

Choose masses $m_i^A$ for Theory A modulo their common, gauge-redundant shift, and simple-root FI coordinates $\zeta_a^B$ for $a=1,\ldots,N-1$. One convenient orientation convention is

$$-2\pi\zeta_a^B=m_a^A-m_{a+1}^A.$$

The normalized topological mass $-2\pi\zeta_A$ maps to the common flavor mass $m_B$ of Theory B:

$$m_B=-2\pi\zeta_A.$$

Overall signs can be reversed by changing the topological-current or quiver-orientation convention; a complete calculation fixes them by matching a chosen vortex charge. What is invariant is the integral linear map between the flavor and topological charge lattices.

The manifest $PSU(N)$ Higgs-branch flavor symmetry of Theory A maps to an infrared enhancement of the topological $U(1)^{N-1}$ symmetry of Theory B. Conversely, the $U(1)_J$ topological symmetry of Theory A maps to the ordinary flavor symmetry acting on the end-to-end Higgs coordinate of Theory B.

At the level of protected operators:

• mesons $Q_i\widetilde Q^j$ and their moment-map relations in Theory A map to monopoles of the quiver with magnetic charges in the $A_{N-1}$ root/weight lattice;
• the minimal monopoles $V_+$ and $V_-$ of Theory A map to the long Higgs operators $X=\prod_iX_i$ and $Y=\prod_iY_i$;
• the vector scalar completing the Theory-A Coulomb triplet maps to the diagonal Higgs moment map in Theory B.

The flux, flavor, and R charges on each line of this dictionary must agree. The zero-mode construction supplies a direct monopole check; in particular, it demonstrates that the relevant monopoles occupy the short representations predicted by mirror symmetry Borokhov, Kapustin, and Wu 2002, §§3–5.

Mirror symmetry also exchanges Wilson-type probes with vortex-type probes. This slogan requires care: the allowed charges and whether a line is genuine depend on the global gauge group and on which one-form symmetry has been gauged. A local-operator mirror pair does not by itself determine an equivalence of every possible line category.

The N=1 check

For one hypermultiplet, Theory A is $\mathcal N=4$ $U(1)$ SQED with one flavor and Theory B has no gauge node: $(X_1,Y_1)$ is a free twisted hypermultiplet. The relation $V_+V_-=\varphi$ eliminates $\varphi$, so

$$\mathbb C[\mathcal M_C(A)]=\mathbb C[V_+,V_-]\simeq\mathbb C^2.$$

The monopoles $V_\pm$ have dimension $1/2$ and map directly to the two free complex scalars; $\varphi=V_+V_-$ is their composite moment-map operator, not a third independent free scalar. This is a particularly sharp operator-level realization of mirror symmetry, not a statement that the ultraviolet gauge field was free Borokhov, Kapustin, and Wu 2002, §4.2.

Checks that probe different structures

No single protected observable supplies the entire global dictionary. Useful independent tests include:

1. Branches and rings: compare Hilbert series, singular loci, symmetry actions, and deformed resolutions—not only dimensions.
2. Mass–FI response: generic triplet deformations should select corresponding vacua and exchange particle and vortex central charges.
3. Monopoles: zero-mode charges and chiral-ring relations must equal the mirror Higgs-operator data.
4. Sphere partition functions: localized matrix integrals are related by Fourier-transform identities in Abelian examples Kapustin, Willett, and Yaakov 2010, §§3–4.
5. Lines and boundaries: after global forms are fixed, genuine line charges and boundary anomalies must map.

The Yang–Mills couplings need not map as parameters of the infrared fixed point: they are dimensionful and irrelevant there. Some constructions introduce BF-coupled ultraviolet completions in which a stronger finite-scale transform can be formulated, but that is extra structure, not part of the minimal infrared claim.

For a compact comparison with the regulator bookkeeping required by less-supersymmetric descendants, see the chapter’s monopole–contact–duality map.

Limitations and supersymmetry breaking

The nonrenormalization of the Higgs branch and shortening of monopole operators rely on $\mathcal N=4$. An arbitrary $\mathcal N=2$ superpotential, Chern–Simons term, or unequal real mass can split multiplets, lift branches, and reduce the dictionary. A controlled $\mathcal N=2$ descendant must recompute parity contact terms and monopole charges rather than inherit them by name.

Similarly, a soft deformation that breaks supersymmetry does not preserve the equality of scalar and fermion masses. Following the resulting phase diagram can motivate particle–vortex or bosonization dualities, but the critical equivalence then contains a new dynamical assumption.

Common pitfalls

Equating mirror symmetry with an equality of classical moduli spaces. The Coulomb geometry is generally quantum corrected. Mirror symmetry identifies the quantum Coulomb branch with the protected Higgs branch of the other theory.

Writing masses and FI parameters as unstructured lists. They live in integral charge lattices, with redundancies and Weyl actions. The mirror map is an integral linear map after a basis and current normalization are fixed.

Ignoring decoupled Abelian factors. A diagonal $U(1)$ with no charged matter changes topological currents and partition functions. Remove or retain it explicitly on both sides.

Exercises

1. For Theory A with $N$ hypers, derive $\dim_{\mathbb H}\mathcal M_H=N-1$ by hyperkähler quotient and $\dim_{\mathbb H}\mathcal M_C=1$ by gauge rank. Repeat for Theory B.

Solution

Theory A starts with quaternionic dimension $N$ and quotienting by $U(1)$ removes one, giving $N-1$; its rank is one. Theory B has $N$ hypers and gauge rank $N-1$, so its Higgs quotient has dimension $N-(N-1)=1$, while its Coulomb branch has quaternionic dimension $N-1$. The dimensions are exchanged.

2. For $N=2$, impose the complex moment-map equation $Q_1\widetilde Q_1+Q_2\widetilde Q_2=0$ and quotient by the complexified $U(1)$. Show that the invariant coordinates can obey an $A_1$ relation $xy=z^2$ up to a harmless sign redefinition.

Solution

Take $x=Q_1\widetilde Q_2$, $y=Q_2\widetilde Q_1$, and $z=Q_1\widetilde Q_1=-Q_2\widetilde Q_2$. Then $xy=(Q_1\widetilde Q_1)(Q_2\widetilde Q_2)=-z^2$. Redefining one of $x,y$ by a minus sign gives $xy=z^2$, the coordinate ring of $\mathbb C^2/\mathbb Z_2$.

3. Using $S_i{}^a=\delta_{i,a}-\delta_{i,a+1}$, show that $X=\prod_iX_i$ and $Y=\prod_iY_i$ are gauge invariant in Theory B and derive $XY=Z^N$.

Solution

For each gauge factor $a$, the charge of $X$ is $\sum_iS_i{}^a=1-1=0$; $Y$ has the opposite and therefore also vanishing charge. The $\Phi_a$ F-terms set every $X_iY_i$ equal to the same invariant $Z$. Consequently

$$XY=\prod_{i=1}^{N}(X_iY_i)=Z^N.$$

This is the same presentation as $V_+V_-=\varphi^N$ on the Theory-A Coulomb branch, so the map $V_+\leftrightarrow X$, $V_-\leftrightarrow Y$, and $\varphi\leftrightarrow Z$ preserves multiplication as well as charges and dimensions.

References

• Borokhov, V., Kapustin, A., and Wu, X. (2002), “Monopole Operators and Mirror Symmetry in Three Dimensions,” Journal of High Energy Physics 2002(12), 044. doi:10.1088/1126-6708/2002/12/044. Open PDF
• Cremonesi, S., Hanany, A., and Zaffaroni, A. (2014), “Monopole Operators and Hilbert Series of Coulomb Branches of 3d $\mathcal N=4$ Gauge Theories,” Journal of High Energy Physics 2014(01), 005. doi:10.1007/JHEP01(2014)005. Open PDF
• Intriligator, K., and Seiberg, N. (1996), “Mirror Symmetry in Three Dimensional Gauge Theories,” Physics Letters B 387, 513–519. doi:10.1016/0370-2693(96)01088-X. Open PDF
• Kapustin, A., Willett, B., and Yaakov, I. (2010), “Nonperturbative Tests of Three-Dimensional Dualities,” Journal of High Energy Physics 2010(10), 013. doi:10.1007/JHEP10(2010)013. Open PDF

Next steps

Four-supercharge analogues require the singlet and monopole superpotentials of Aharony and Giveon–Kutasov dualities. Their relation to other dimensions is controlled by real-mass, FI, and compactification flows.