Rigid Backgrounds, Topological Twists, and Localization

Localization is not a single theorem applied to an infinite-dimensional integral. It is a chain of constructions: choose a supersymmetric background, obtain a globally defined odd symmetry, identify its cohomology, combine it with gauge fixing, determine every saddle and zero mode, and only then specify the integration cycle and regularization that turn the formal answer into an observable. This chapter develops that chain and makes clear which statements are exact theorems and which remain controlled path-integral arguments.

Helpful background. Fredholm and Dirac index theorems explain how unpaired modes replace a formal quotient of infinitely many eigenvalues. Heat kernels, zeta functions, and spectral determinants supply determinant regulators. BRST cohomology and physical observables supplies the gauge-theory differential used after a localizing supercharge is chosen.

Enter this chapter

The central diagnostic is simple:

A localization formula is only as well defined as its least specified ingredient.

For a Euclidean theory, fields of opposite Lorentzian reality type are generally independent complex variables. Thus a positive bosonic deformation is a statement about a chosen middle-dimensional cycle, not an automatic consequence of Wick rotation. Likewise, $Q^2$ need not vanish: it may be an isometry, an R rotation, a flavor transformation, and a field-dependent gauge transformation. Every deformation, insertion, boundary condition, and regulator must be invariant under that full even symmetry.

The pages are ordered so that each construction supplies the input of the next.

Stage	Question answered	Output needed downstream
Supercurrent compatibility	Which current multiplet can couple to the desired off-shell supergravity formulation?	Admissible nondynamical sources and obstruction classes
Rigid curved backgrounds	Which frozen bosonic backgrounds set the gravitino variation to zero?	Generalized Killing-spinor equation and off-shell algebra
Global spin-R bundles	Does the local spinor patch to a global supercharge?	Bundle, flux, zero-locus, and boundary data
Topological and holomorphic twists	Can Lorentz and R symmetry be recombined to produce a scalar or holomorphic differential?	Twisted fields, $Q$-complex, and protected observables
Finite-dimensional localization	What does the Atiyah–Bott–Berline–Vergne theorem actually prove?	The precise fixed-locus model and its hypotheses
$Q$-exact path-integral deformation	Why should a selected correlator be deformation independent?	Ward identity and a list of possible failure terms
Gauge fixing and the deformation complex	How are supersymmetry and gauge redundancy combined?	A graded elliptic or transversely elliptic complex
Loci and one-loop determinants	Which saddles, collective coordinates, and unpaired modes remain?	Sector sum, classical weights, and regulated determinant
Contours and regularization	Which integration cycle and spectral prescription define the answer?	Thimble coefficients, determinant phase, and scheme dependence
Boundaries, gluing, and JK residues	How do cutting, boundary anomalies, and residue chambers fit together?	A gluing pairing or chamber-qualified residue formula

The validity chain

It is useful to write a proposed computation as the following data rather than as the slogan “add a $Q$-exact term”:

$$\mathfrak L= \bigl(M,\mathcal B,\Gamma,Q,Q^2,\mathcal O,V, \mathcal M_Q,\mathcal C_{\rm def},\operatorname{Reg},S_{\rm ct},\mathfrak b\bigr).$$

Here $M$ is the spacetime (possibly with boundary), $\mathcal B$ its spin-R and background bundles, $\Gamma$ the field-space cycle, $Q$ the odd symmetry, $\mathcal O$ the insertion, and $V$ the deformation functional. The remaining entries are the full BPS locus $\mathcal M_Q$, the gauge-fixed fluctuation complex $\mathcal C_{\rm def}$, the spectral regulator, allowed local counterterms, and boundary or gluing data $\mathfrak b$. A claimed exact quantity should remain qualified until every relevant entry is fixed.

Three distinct conclusions must not be conflated:

1. Cohomological independence: a Ward identity gives $d\langle\mathcal O\rangle_t/dt=0$ for finite $t$.
2. Asymptotic localization: the $t\to\infty$ integral is controlled by the zeros of the bosonic deformation on the chosen cycle.
3. Evaluation: the resulting finite-dimensional integral, sum, or residue has been assigned a contour, regulator, phase, and normalization.

The first does not by itself prove the second, and neither fixes the third. This distinction is emphasized in the finite- and infinite-dimensional treatments of Schwarz and Zaboronsky 1997, pp. 463–476 and in the gauge-theory construction of Pestun 2012, §§3–4.

Review the chapter

Take a localization formula you know and answer, in order:

1. What is the globally defined bundle of the supersymmetry parameter?
2. What is $Q^2$ on physical fields, ghosts, and boundary fields?
3. Which integration cycle makes the bosonic deformation convergent?
4. Are all topological sectors and fixed components included?
5. Which modes are removed from $\det'$ and how are they integrated instead?
6. Which regulator fixes the determinant phase and scale?
7. Which supersymmetric local counterterms can change the reported observable?
8. If the space is cut, what polarization, anomaly inflow, and gauge quotient enter the pairing?

If any answer is missing, the formula may still be a useful formal expression, but it is not yet a complete definition.

References

• Pestun, Vasily. “Localization of Gauge Theory on a Four-Sphere and Supersymmetric Wilson Loops.” Communications in Mathematical Physics 313 (2012): 71–129. doi:10.1007/s00220-012-1485-0. Open preprint.
• Schwarz, Albert, and Oleg Zaboronsky. “Supersymmetry and Localization.” Communications in Mathematical Physics 183 (1997): 463–476. doi:10.1007/BF02506415. Open preprint.

Further reading

• Pestun, Vasily, and Maxim Zabzine, eds. “Localization Techniques in Quantum Field Theories.” Journal of Physics A: Mathematical and Theoretical 50 (2017), special issue. Foreword and chapter guide.