Q-Cohomology and Path-Integral Deformation

For a selected $Q$-closed observable, a $Q$-exact deformation can leave the path integral independent of its coefficient. The proof is a Ward identity at finite deformation parameter. Turning that identity into a saddle formula as the parameter tends to infinity requires additional convergence, contour, and completeness arguments; it is not contained in the algebraic statement “$QV$ is exact.”

Required background. Finite-dimensional equivariant localization states the rigorous model and its compactness hypotheses. $Q$-cohomology and Hodge complexes supplies the cohomological interpretation of protected insertions.

Helpful background. Localized transformations and Ward–Takahashi identities supplies the change-of-variables derivation of a Ward identity and its anomaly term.

The finite-parameter Ward identity

Let $\Phi$ denote all fields, including auxiliaries, and let $\Gamma$ be a specified middle-dimensional cycle in the complexified Euclidean field space. Consider an even observable $\mathcal O$ and

$$N_{\mathcal O}(t)= \int_\Gamma\mathcal D\Phi\; \mathcal O\,e^{-S-tQV}.$$

Assume

$$QS=0, \qquad Q\mathcal O=0, \qquad Q^2V=0.$$

Then the integrand is $Q$ invariant and

$$\frac{dN_{\mathcal O}}{dt} =-\int_\Gamma\mathcal D\Phi\; \mathcal O\,QV\,e^{-S-tQV} =-\int_\Gamma\mathcal D\Phi\; Q\!\left(\mathcal O V e^{-S-tQV}\right).$$

If the measure has zero $Q$ divergence and the change of variables $\Phi\mapsto\Phi+\epsilon Q\Phi$ preserves $\Gamma$ without a contribution from its boundary or infinity, the last integral vanishes. Thus $N_{\mathcal O}(t)$ is independent of $t$. The same argument applies to $Z(t)=N_1(t)$, so the normalized expectation value $\langle\mathcal O\rangle_t=N_{\mathcal O}(t)/Z(t)$ is also independent wherever $Z(t)\ne0$.

The condition $Q^2V=0$ is substantive. If $Q^2=\mathcal H+\delta_{\rm gauge}(\Lambda)$, then $V$ must be invariant under $\mathcal H$ and gauge transformations. A functional that is $Q$ exact but not invariant under $Q^2$ does not define a supersymmetric deformation.

This change-of-variables argument is the field-theory analogue of the odd-symmetry localization theorem of Schwarz and Zaboronsky 1997, §§1–3. In an infinite-dimensional path integral it remains conditional on the measure and cycle.

Every possible remainder

It is useful to write the failed Ward identity schematically as

$$\int_\Gamma\mathcal D\Phi\;QX =\mathcal A_Q[X] +\int_{\partial\Gamma} \iota_Q\Omega_X +\mathcal B_{\rm spacetime}[X].$$

The three terms have different origins:

Measure anomaly $\mathcal A_Q$. A regulator may not preserve $Q$, or the fermion measure may have an anomalous Jacobian. The anomaly must vanish, be cancelled by inflow, or be represented explicitly in the Ward identity.

Field-space boundary. Noncompact directions, singular gauge orbits, poles, or an integration cycle with endpoints can leave a surface term. Exponential decay on $\Gamma$ must be uniform enough to remove it.

Spacetime boundary. Integrating the local current identity can produce a normal supercurrent flux. Boundary conditions and boundary degrees of freedom must cancel it.

There are two further elementary failures: an insertion can be only classically $Q$ closed, or it can collide with another insertion and produce contact terms. Neither is repaired by increasing $t$.

From independence to a saddle expansion

Choose a fermionic $V$ such that on the selected cycle

$$(QV)_{\rm bos}\ge0.$$

In Euclidean signature this inequality is a property of $V$ and $\Gamma$. Since barred and unbarred fields are independent complex variables, it cannot be inferred from a Lorentzian adjoint relation. If the inequality is strict transverse to its zero set, configurations away from

$$\mathcal M_Q=\{\Phi\in\Gamma:(QV)_{\rm bos}=0\}$$

are exponentially suppressed as $t\to\infty$.

To replace the original integral by fluctuations around $\mathcal M_Q$, one must additionally establish:

1. the complete set of connected components and topological sectors of $\mathcal M_Q$;
2. uniform convergence permitting differentiation and the $t\to\infty$ limit to be exchanged with integration;
3. control of flat, negative, and complex directions;
4. a gauge-fixed quadratic complex with all zero modes separated;
5. absence of new boundary contributions as the contour is deformed;
6. a regulator compatible with the $Q^2$ symmetry.

Only after these steps does deformation independence justify evaluating the answer at large $t$. Pestun’s four-sphere computation exhibits the full sequence—off-shell $Q$, positive deformation on a chosen real slice, locus, gauge fixing, and index-theoretic determinant—rather than using exactness alone Pestun 2012, §§3–4.

Dependence that localization does not remove

Even a valid localization argument can leave dependence on $Q$-closed couplings, background moduli, discrete flux sectors, boundary polarizations, and the integration cycle. It also does not remove finite local counterterm ambiguity. If $S_{\rm ct}$ is an allowed supersymmetric background functional, then

$$Z\longmapsto e^{-S_{\rm ct}}Z$$

can change a partition function while preserving every Ward identity. The protected quantity must therefore be stated with its scheme and normalization, or in a ratio or derivative where the allowed ambiguity cancels.

Common pitfalls

“$QV$ is exact, so its bosonic part is positive.” Exactness is algebraic; positivity is analytic and contour dependent.

“The derivative vanishes, so only one saddle contributes.” Finite-$t$ independence neither classifies saddles nor fixes the thimble coefficients of the original cycle.

“Nonzero modes cancel.” They cancel only after gauge fixing, with compatible domains and regulators. Kernels, cokernels, phases, and anomalous Jacobians remain.

Exercises

1. Normalized correlator. Starting from $\langle\mathcal O\rangle_t=N_{\mathcal O}(t)/Z(t)$, derive its derivative before using the Ward identity.

Solution

Differentiating the quotient gives $d\langle\mathcal O\rangle_t/dt=-\langle\mathcal O\,QV\rangle_t+\langle\mathcal O\rangle_t\langle QV\rangle_t$. Each term vanishes by writing the corresponding numerator as a $Q$ variation, under the stated measure and boundary hypotheses.

2. A cycle with endpoints. In a finite-dimensional analogue, let $Q$ act as the de Rham differential on an interval. Why need the integral of a $Q$-exact one-form not vanish?

Solution

If the form is $df$, then $\int_{[a,b]}df=f(b)-f(a)$. The endpoints are the field-space boundary term. Compactness without boundary, decay at infinity, or boundary conditions setting this difference to zero are required.

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.

Next step

For a gauge theory the relevant odd symmetry is a combined differential. Construct it on the gauge-fixed localization deformation complex.