Supersymmetric Sigma Models and Kähler Geometry

A two-dimensional nonlinear sigma model turns scalar fields into a map $\phi:\Sigma_2\to X$. Requiring ordinary $(2,2)$ supersymmetry for a model built only from chiral multiplets forces the target metric to be Kähler, packages the metric and closed $B$-field into superspace data, and ties quantum R-symmetry anomalies to $c_1(TX)$. The Kähler condition is an off-shell statement for this multiplet choice; conformal invariance is a stronger quantum condition.

Required background. We use $(2,2)$ chiral superspace and the general construction of supersymmetric Kähler sigma models. Helpful background. Regulated Jacobians and measure variation explain the axial anomaly used below.

From superspace to Kähler geometry

Let $\Phi^i$ be local holomorphic coordinates on a complex target $X$. The most general two-derivative D-term action made only from these chiral fields is

$$S_D=\frac1{4\pi\alpha'}\int d^2x\,d^4\theta\,K(\Phi,\bar\Phi).$$

Its bosonic part contains

$$S_{\mathrm{bos}}= \frac1{4\pi\alpha'}\int d^2x\, g_{i\bar j}(\phi,\bar\phi) \partial_\mu\phi^i\partial^\mu\bar\phi^{\bar j}+\cdots,$$

with

$$g_{i\bar j}=\partial_i\partial_{\bar j}K, \qquad g_{ij}=g_{\bar i\bar j}=0.$$

The associated two-form

$$\omega=i g_{i\bar j}\,d\phi^i\wedge d\bar\phi^{\bar j}$$

is closed because mixed partial derivatives commute. Thus $(X,g,J,\omega)$ is Kähler. Conversely, every Kähler metric is locally of this form. On overlaps $U_a\cap U_b$, local potentials may differ by

$$K_a-K_b=f_{ab}(\Phi)+\bar f_{ab}(\bar\Phi),$$

whose full-superspace integral vanishes. The action therefore glues even when no global Kähler potential exists.

This is the superspace version of Zumino’s result that a second supersymmetry requires a covariantly constant complex structure compatible with the metric Zumino 1979.

Component geometry and fermions

The right- and left-moving fermions are sections

$$\psi_+^i\in\Gamma(S_+\otimes\phi^*T^{1,0}X), \qquad \psi_-^i\in\Gamma(S_-\otimes\phi^*T^{1,0}X),$$

with conjugates valued in $T^{0,1}X$. Their kinetic terms use the pullback Levi-Civita connection, and supersymmetry fixes a four-fermion curvature interaction schematically of the form

$$R_{i\bar j k\bar l}\,psi_+^i\bar\psi_+^{\bar j} \psi_-^k\bar\psi_-^{\bar l}.$$

This term is not optional: varying the connection in the fermion kinetic term produces curvature, whose cancellation requires the four-fermion coupling. It is also the local source of characteristic classes in the twisted theory.

If a holomorphic superpotential $W$ is added, the bosonic potential is $g^{i\bar j}\partial_iW\partial_{\bar j}\bar W$. A pure sigma model has $W=0$ and a continuous target; a sigma model with $W$ is more properly a curved-target Landau–Ginzburg model.

The $B$-field and torsion

A closed two-form $B$ contributes

$$S_B=\frac{i}{2\pi\alpha'}\int_{\Sigma_2}\phi^*B$$

in Euclidean signature. Locally $B$ may be shifted by $d\Lambda$, and globally its gauge-invariant information is a gerbe connection; periods are defined modulo large gauge transformations. When $H=dB=0$, the complexified Kähler class is conventionally written $B+i\omega$ up to normalization.

The elementary chiral-only superspace action above has Kähler target and no local torsion. More general $(2,2)$ models containing both chiral and twisted-chiral fields—or semichiral multiplets—can support $H\ne0$ and generalized Kähler, or bi-Hermitian, geometry. In that case two complex structures $J_\pm$ are covariantly constant with respect to connections with torsion $\pm H$. This generalization was derived in Gates, Hull, and Roček 1984. It does not contradict the Kähler result; it changes the off-shell multiplet content and geometric hypotheses.

R-symmetry anomalies

Classically both $U(1)_V$ and $U(1)_A$ act on the fermions. In an ordinary Kähler sigma model, the vector symmetry is non-anomalous, whereas the axial fermion measure transforms by an index proportional to

$$\int_{\Sigma_2}\phi^*c_1(TX).$$

Hence $U(1)_A$ is non-anomalous for all worldsheet maps if $c_1(TX)=0$ in the relevant integral cohomology. A weaker torsion condition can leave only a discrete subgroup. This distinction controls the twists:

• the A-twist uses $U(1)_V$ and is available for a generic Kähler target;
• the B-twist uses $U(1)_A$ and globally requires the axial anomaly to vanish.

For a Calabi–Yau target, a nowhere-vanishing holomorphic volume form trivializes the canonical bundle and sets $c_1(TX)=0$. The converse can require global qualifications, especially for noncompact or singular spaces.

Renormalization and the conformal question

At one loop in the standard sigma-model normalization, the metric runs by the Ricci tensor Friedan 1980:

$$\mu\frac{d}{d\mu}g_{i\bar j} =\alpha'R_{i\bar j}+O(\alpha'^2),$$

up to the sign convention for RG time. A Ricci-flat metric cancels the leading metric beta function, but several distinctions matter:

1. Kähler is not Ricci-flat. Kähler geometry is required by supersymmetry; Ricci-flatness addresses conformal invariance.
2. One loop is not an all-orders proof. Extended supersymmetry strongly constrains higher corrections, yet the precise conformal representative can differ from a classical Ricci-flat metric by scheme-dependent field redefinitions.
3. Vanishing beta functions are not enough without a well-defined QFT. Noncompact targets can have continuous spectra and infrared divergences.
4. The dilaton changes the equations. A nonconstant dilaton contributes to Weyl-invariance conditions and cannot be omitted in general string backgrounds.

The Kähler class runs proportionally to $c_1(X)$ perturbatively. For $c_1(X)=0$, this leading running vanishes; for Fano targets such as $\mathbb{CP}^{N-1}$, the model is asymptotically free and generates a scale.

Worldsheet instantons

In Euclidean signature, an A-type BPS configuration is a holomorphic map $\bar\partial\phi=0$. The bosonic action admits the bound

$$S_E\ge \frac1{2\pi\alpha'} \left|\int_{\Sigma_2}\phi^*\omega\right|,$$

with equality for a holomorphic or antiholomorphic map according to orientation. Its weight combines area and $B$-field phase,

$$\exp\left[-\frac1{2\pi\alpha'}\int\phi^*\omega +\frac{i}{2\pi\alpha'}\int\phi^*B\right].$$

Instantons can deform A-model products into quantum cohomology. B-model local observables are independent of Kähler moduli and do not receive the same worldsheet-instanton corrections, though global anomalies, boundaries, and noncompactness still require control; the localization and observable structure of the sigma-model twist were established in Witten 1988.

For $X=\mathbb{CP}^{N-1}$, the classical cohomology relation $H^N=0$ becomes the quantum relation

$$H^N=q,$$

where $q$ records the complexified Kähler parameter in a chosen normalization. The GLSM and effective twisted-superpotential pages derive the same relation from a gauge-theory Coulomb branch.

Domain of validity

Statement	Assumptions	Typical failure
Chiral multiplets imply Kähler target	Two-derivative off-shell $(2,2)$ action using ordinary chirals	Twisted or semichiral multiplets allow torsionful generalized Kähler targets
B-twist exists	Quantum non-anomalous $U(1)_A$	$c_1(TX)\ne0$ or boundary anomaly
Instanton sum is discrete	Compact moduli after stable-map completion	Noncompact zero modes or bubbling at uncontrolled boundaries
Ricci-flatness suggests an SCFT	Compact, unitary model with controlled quantum corrections	Continuum, dilaton gradient, singular target
Classical geometry is a valid EFT	Curvatures small in sigma-model units and omitted modes heavy	Singular GLSM wall or small-cycle regime

Exercises

1. Show directly that $\omega=i\partial\bar\partial K$ is closed.

Solution

$d=\partial+\bar\partial$, while $\partial^2=\bar\partial^2=0$ and $\partial\bar\partial=-\bar\partial\partial$. Therefore $d\omega=i(\partial+\bar\partial)\partial\bar\partial K=0$.

2. For the Fubini–Study potential $K=r\log(1+|z|^2)$ on a patch of $\mathbb{CP}^1$, compute the metric and Kähler form.

Solution

Two derivatives give

$$g_{z\bar z}=\frac{r}{(1+|z|^2)^2}, \qquad \omega=\frac{ir\,dz\wedge d\bar z}{(1+|z|^2)^2}.$$

On the opposite patch the potentials differ by a holomorphic plus antiholomorphic term, so the metric and $\omega$ agree globally.

3. Why can a noncompact Ricci-flat Kähler target fail to define a compact SCFT spectrum?

Solution

Ricci-flatness controls the local beta function, not normalizability. A noncompact target permits wavefunctions to escape to infinity and generally produces a continuum of states. Partition functions and indices then need infrared boundary conditions or regulators.

References

• Friedan, D. H. “Nonlinear Models in $2+\epsilon$ Dimensions.” Physical Review Letters 45 (1980): 1057–1060. doi:10.1103/PhysRevLett.45.1057.
• Gates, S. J., Hull, C. M., and Roček, M. “Twisted Multiplets and New Supersymmetric Nonlinear Sigma Models.” Nuclear Physics B 248 (1984): 157–186. doi:10.1016/0550-3213(84)90592-3.
• Witten, E. “Topological Sigma Models.” Communications in Mathematical Physics 118 (1988): 411–449. doi:10.1007/BF01466725.
• Zumino, B. “Supersymmetry and Kähler Manifolds.” Physics Letters B 87 (1979): 203–206. doi:10.1016/0370-2693(79)90964-X.