N=2 Gauge Dynamics and Seiberg–Witten Geometry

Four-dimensional $\mathcal N=2$ gauge dynamics turns strongly coupled infrared physics into geometry. On a generic Coulomb-branch patch the light theory is abelian and controlled at two derivatives by special coordinates. Globally, electric and magnetic charges form an integral local system, while the central charge is a holomorphic section of its dual; a Seiberg–Witten curve and differential often realize its local components as periods. Singular fibers identify candidate vanishing charges and possible light sectors, but the existence and chamber of a BPS state remain separate physical data. The pure-$SU(2)$ construction is developed in Seiberg and Witten 1994, §§2–6 and reconstructed pedagogically in Bilal 1996, §§3–6.

Helpful background. Review Hamiltonian group actions and moment maps, branches and monodromy, and monopoles and dyons.

Enter through the physical layer

Fields and vacuum branches. Begin with $\mathcal N=2$ multiplets, Lagrangians, and branches. It separates Coulomb, Higgs, and mixed branches and explains why their geometries have different protection.

Low-energy photons. Use Coulomb branches and the abelian theory to identify the Wilsonian two-derivative action, central charges, and the scale at which additional states must be restored.

Metric and couplings. Use prepotentials and special Kähler geometry locally. A prepotential exists in a chosen polarization; the period vector and symplectic structure are the global data.

Global electric–magnetic structure. Use charge lattices, duality frames, and local systems. Charges transform contragrediently to periods so the central charge stays invariant.

Singular points. Use vanishing cycles and monodromies to infer local integral monodromy from a massless BPS charge and check the ordered global product.

Curves and periods. Use Seiberg–Witten curves, differentials, and periods for the geometric construction, then the pure $SU(2)$ solution for the canonical complete example.

Beyond rank one. Use matter, higher rank, and integrable systems only after the period and monodromy conventions are fixed. A spectral curve is useful when its ultraviolet origin, polarization, residues, and exceptional loci are known. The extension to massive hypermultiplet matter, including singularities, curves, and residue conditions, is worked out in Seiberg and Witten 1994, §§2–10 and §§11–17.

BPS states. Use spectra, chambers, and wall crossing for protected one-particle indices in a named chamber. The charge lattice does not populate itself.

Strongly coupled endpoints. Use Argyres–Douglas theories and class S to extract scaling data at mutually nonlocal singularities and to separate intrinsic four-dimensional claims from six-dimensional constructions.

Controlled confinement. Use $\mathcal N=1$ deformations and monopole condensation for the small-deformation regime. It is a precise supersymmetric confinement mechanism, not a derivation of nonsupersymmetric QCD confinement.

The global structure in one view

Let the Coulomb branch have complex rank $r$. On its regular locus $\mathcal B^\circ$, electric and magnetic charges form a local system

$$\Gamma\longrightarrow\mathcal B^\circ$$

with integral antisymmetric pairing. In a local symplectic basis, collect special coordinates as

$$\Pi(u)=\binom{a_{D,I}(u)}{a^I(u)}, \qquad I=1,\ldots,r.$$

Invariantly, the central-charge map is a holomorphic section

$$Z\in H^0\!\left(\mathcal B^\circ, \Gamma^\vee\otimes_{\mathbb Z}\mathcal O_{\mathcal B^\circ}\right),$$

and $\Pi$ is its column of components in the chosen local basis. This distinction prevents one from confusing the lattice that carries charges with the dual holomorphic object that assigns their central charges.

An electromagnetic charge $\gamma_{\mathrm{em}}=(p^I,q_I)$ has

$$Z_{\gamma_{\mathrm{em}}}(u)=q_Ia^I(u)+p^Ia_{D,I}(u).$$

If flavor charges are present, write the extended charge as $\widehat\gamma=(p^I,q_I;s_a)$ and add $s_am^a$ to the central charge. The column $\Pi$ above contains only the electromagnetic section; flavor masses belong to the affine flavor extension and must not be hidden inside that $2r$-component vector.

Transport around a singularity acts by

$$\Pi\longmapsto M\Pi, \qquad M\in\operatorname{Aut}(\Gamma,J).$$

In a principal symplectic basis this automorphism group is $Sp(2r,\mathbb Z)$. For polarization type $D$, it is instead the integral group preserving $J_D$.

while the charge coordinates transform by the inverse transpose in the compatible convention. Thus $Z_\gamma$ is single-valued for a transported physical charge even though individual period components are not.

The coupling matrix

$$\tau_{IJ}=\frac{\partial a_{D,I}}{\partial a^J}$$

is symmetric and has positive imaginary part on a physical local patch. The metric is

$$ds^2=\operatorname{Im}\tau_{IJ}\,da^I\,d\bar a^J.$$

These equations describe the two-derivative Wilsonian action away from singularities. Higher-derivative terms and the spectrum of massive states require additional data.

Why singularities are informative

At a discriminant locus, the homology cycle $\delta_\gamma$ associated with a charge $\gamma$ can vanish and

$$Z_\gamma\to0.$$

The corresponding BPS state becomes massless if it exists in the chamber. Integrating it out produces a logarithmic coupling and monodromy. Restoring it in a duality frame where it is electric yields a regular local effective theory when all light charges are mutually local.

If mutually nonlocal charges become massless simultaneously, no single electric frame contains them all as local elementary fields. The infrared endpoint can be an interacting Argyres–Douglas SCFT. The charge pairing distinguishes this situation from an ordinary weakly coupled singularity.

The pure SU(2) benchmark convention

The rank-one pages use one frozen convention unless they explicitly say otherwise. This card is deliberately more detailed than a usual notation list: changing only one row changes charges, matrices, or period values.

Datum	Chapter convention
Scale and base point	$\Lambda>0$ and $u_b=2\Lambda^2$ on the weak-coupling positive real axis
Curve and differential	$y^2=(x-u)(x^2-\Lambda^4)$ and $\lambda_{\mathrm{SW}}=(\sqrt2/(2\pi))(x-u)\,dx/y$
Base-fiber cuts	$[-\Lambda^2,+\Lambda^2]$ and $[u_b,\infty]$ in the $x$-plane
Cycles and periods	$A\circ B=+1$, $a=\oint_A\lambda_{\mathrm{SW}}>0$, and $a_D=\oint_B\lambda_{\mathrm{SW}}\in i\mathbb R_{>0}$ at $u_b$
Charge labels	$\gamma=(n_m,n_e)^T$, $\delta_\gamma=n_mB+n_eA$, and $Z_\gamma=n_ma_D+n_ea$
Pairings	$\omega(\gamma,\gamma')=n_mn'_e-n_en'_m=-\delta_\gamma\circ\delta_{\gamma'}$ and $\langle\gamma,\gamma'\rangle_{\mathrm{part}}=2\omega(\gamma,\gamma')$
Distinguished particles	$\gamma_W=(0,1)$, $\gamma_m=(1,0)$, and $\gamma_d=(1,-1)$ in root-unit electric labels
Based loops	Counterclockwise meridians $\ell_+$ and $\ell_-$ use the stems shown on the shared monodromy figure; $\rho(\ell_\infty)=\rho(\ell_+)\rho(\ell_-)$
Matrix action	Period columns transform on the left, so the rightmost matrix in a product is applied first
Chamber	$u_b$ is in the exterior weak-coupling chamber; pages that state a spectrum also declare a central-charge half-plane

The factor two in the physical particle pairing is the $SU(2)$ root–coroot pairing in this adjoint, root-unit normalization. It is not a claim that the displayed $W$ label equals two. In weight-unit labels one instead writes $n_e^{\mathrm{wt}}=2n_e$ and $a^{\mathrm{wt}}=a/2$; charges and every monodromy matrix must then be transformed together.

The two finite discriminant points are $u=\pm\Lambda^2$. With the paths above, a monopole becomes massless at the positive point and a dyon at the negative point. At large positive $u$, $a\sim\sqrt{2u}$, and the one-loop logarithm fixes the monodromy at infinity. The based product of the two finite monodromies must reproduce it. This global constraint is as important as either local calculation.

Exact results and bounded claims

Result	Scope
Holomorphic prepotential	Local two-derivative abelian vector-multiplet action
Special Kähler metric	Regular Coulomb-branch locus; singular EFT patches require light fields
Curve and differential	Periods and monodromies after normalization and cycle choices
BPS mass formula	Exact conditional on state existence and chamber
Wall-crossing formula	Protected index, not the full unprotected spectrum
Hyperkähler Higgs metric	Protected under standard rigid $\mathcal N=2$ assumptions
$\mathcal N=1$ monopole condensation	Controlled for a small adjoint-mass deformation near the solved singularity
Integrable-system or class-S realization	Family-specific; ultraviolet origin and global data must be stated

This table prevents the exactness of one protected layer from spreading to every observable.

Reproducibility checklist

A Seiberg–Witten calculation should specify:

1. gauge group and global form, matter, masses, and dynamical-scale convention;
2. base point and coordinates on the Coulomb branch;
3. charge basis, intersection pairing, and central-charge convention;
4. curve, differential, residues, and any exact-form ambiguity;
5. branch cuts, cycle orientations, continuation paths, and monodromy order;
6. asymptotic normalization and numerical error if periods are integrated;
7. BPS chamber and the distinction between allowed charge, state existence, and protected index.

Without these, matrices and period values cannot be compared across references.

Review the chapter

For a rank-one curve with two finite discriminant points:

• choose a weak-coupling base point;
• assign a primitive vanishing charge to each point;
• derive the two Picard–Lefschetz transformations;
• order their product according to explicit loops;
• compare with the semiclassical monodromy at infinity;
• identify a chamber before naming any BPS spectrum.

The pure-$SU(2)$ pages carry out each step and show where a convention change conjugates every matrix consistently.

References

• Bilal, Adel. “Duality in $\mathcal N=2$ SUSY $SU(2)$ Yang–Mills Theory: A Pedagogical Introduction to the Work of Seiberg and Witten.” 1996. arXiv:hep-th/9601007.
• Seiberg, Nathan, and Edward Witten. “Electric–Magnetic Duality, Monopole Condensation, and Confinement in $\mathcal N=2$ Supersymmetric Yang–Mills Theory.” Nuclear Physics B 426 (1994): 19–52; erratum 430 (1994): 485–486. arXiv:hep-th/9407087.
• Seiberg, Nathan, and Edward Witten. “Monopoles, Duality and Chiral Symmetry Breaking in $\mathcal N=2$ Supersymmetric QCD.” Nuclear Physics B 431 (1994): 484–550. arXiv:hep-th/9408099.