Off-Shell Closure and Auxiliary Fields

Auxiliary fields extend a component transformation system so that supersymmetry closes without using equations of motion. They are nonpropagating in a specified action and are eliminated algebraically, but their primary representation-theoretic role is to supply the missing off-shell components and cancellations. This mechanism works finitely for the four-dimensional $\mathcal N=1$ chiral and vector multiplets; degree counting exposes failures, yet equality of counts alone neither constructs auxiliaries nor proves a no-go theorem.

Required background. Component Multiplets and Closure Ledgers supplies the chiral closure calculation. Chiral, Vector, Linear, and Field-Strength Superfields supplies the vector prepotential, Wess–Zumino compensation, and standard component counts.

Helpful background. The BRST Differential and the Gauge-Fixed Complex helps distinguish gauge and ghost variables from supersymmetry auxiliaries.

What “off shell” means

Let $\varphi^I$ denote all fields in a proposed representation. The strongest local closure statement is

$$[\delta_1,\delta_2]\varphi^I =\xi^\mu\partial_\mu\varphi^I$$

for arbitrary field configurations. For a gauge system, the appropriate statement is

$$[\delta_1,\delta_2]\varphi^I =\xi^\mu\partial_\mu\varphi^I +\delta_{\rm gauge}(\Omega)\varphi^I,$$

with $\Omega$ displayed and no field equation used. A kinematic constraint may define the field space, but a term proportional to

$$\mathcal E_I=\frac{\delta S}{\delta\varphi^I}$$

makes the reduced algebra on shell.

An auxiliary field $Y^a$ has two linked properties in a given formulation:

1. its transformation participates in off-shell closure;
2. the action contains no kinetic operator for $Y^a$, so $\delta S/\delta Y^a=0$ is algebraic.

Neither property alone is enough. A gauge-removable component can be nonpropagating without being auxiliary; a Lagrange multiplier can impose a constraint; a compensator restores a gauge choice; a ghost belongs to the gauge-fixed complex. The name follows the full transformation-and-action record.

The chiral auxiliary completes the algebra

For $(A,\psi_\alpha,F)$,

$$\begin{aligned} \delta A &=\sqrt2\,\epsilon\psi,\\ \delta\psi_\alpha &=i\sqrt2(\sigma^\mu\bar\epsilon)_\alpha\partial_\mu A +\sqrt2\,\epsilon_\alpha F,\\ \delta F &=i\sqrt2\,\bar\epsilon\bar\sigma^\mu\partial_\mu\psi. \end{aligned}$$

The $F$ variation cancels the nontranslation part produced when two transformations act on $\psi$. The result is

$$[\delta_1,\delta_2](A,\psi,F) =\xi^\mu\partial_\mu(A,\psi,F)$$

without dynamics. The off-shell count is

$$(2_A+2_F)_{\rm B}=4_{\rm B}, \qquad (\psi_\alpha)_{\rm F}=4_{\rm F}.$$

For canonical Kähler kinetic terms and superpotential $W(A)$, the part of the Lorentzian Lagrangian involving $F$ has the schematic convention-compatible form

$$\mathcal L_F =F^*F+F\,W'(A)+F^*\,\overline{W'(A)}.$$

Its algebraic equations are

$$F=-\overline{W'(A)}, \qquad F^*=-W'(A).$$

Substitution gives the scalar potential

$$V_F=\lvert W'(A)\rvert^2.$$

After substituting $F(A^*)$ into $\delta\psi$, the commutator on $\psi$ contains a term proportional to its interacting Euler–Lagrange equation. The same action-level supersymmetry survives, but the reduced field transformations form an on-shell representation. This is why eliminating an auxiliary and then reporting the original off-shell closure is incorrect. The calculation is explicit in Weinberg 2000, § 26.4, pp. 75–82.

The vector auxiliary and gauge closure

For the Abelian vector multiplet $(A_\mu,\lambda_\alpha,D)$,

$$\begin{aligned} \delta A_\mu &=i\epsilon\sigma_\mu\bar\lambda -i\lambda\sigma_\mu\bar\epsilon,\\ \delta\lambda_\alpha &=(\sigma^{\mu\nu}\epsilon)_\alpha F_{\mu\nu} +i\epsilon_\alpha D,\\ \delta D &=-\epsilon\sigma^\mu\partial_\mu\bar\lambda -\partial_\mu\lambda\,\sigma^\mu\bar\epsilon. \end{aligned}$$

The real $D$ cancels the would-be gaugino equation in the commutator. One obtains

$$\begin{aligned} [\delta_1,\delta_2]\lambda &=\xi^\mu\partial_\mu\lambda,\\ [\delta_1,\delta_2]D &=\xi^\mu\partial_\mu D,\\ [\delta_1,\delta_2]A_\mu &=\xi^\nu\partial_\nu A_\mu +\partial_\mu(-\xi^\nu A_\nu). \end{aligned}$$

Thus the vector multiplet closes off shell modulo gauge. In Wess–Zumino gauge the component transformations already include the compensating supergauge transformation that restores the gauge slice; Gates, Grisaru, Roček, and Siegel 1983, §§ 3.9–3.10 separates the physical, auxiliary, gauge, and compensating components explicitly.

With matter, $D$ enters algebraically and is sourced by the moment map. Eliminating it produces a $D$-term potential. The exact sign and factor of the gauge coupling depend on whether $g$ is placed in $V$, the covariant derivative, or the gauge kinetic term, so they belong to the action convention on the next chapter.

The count is

$$\big[(4-1)_{A_\mu}+1_D\big]_{\rm B} =4_{\rm B} =4_{\rm F}(\lambda).$$

Removing $D$ leaves only three bosonic off-shell functions after the gauge quotient, which flags the missing completion before the commutator is calculated.

Degree counting is a diagnostic

For a finite-dimensional off-shell representation at a generic momentum, one can choose a real linear combination of supercharges whose square is a nonzero translation. That invertible odd operator pairs the bosonic and fermionic component spaces. This motivates

$$n_{\rm B}^{\rm off}=n_{\rm F}^{\rm off}$$

after accounting for gauge equivalence and any gauge-for-gauge structure. The qualification matters:

System	Bosonic count	Fermionic count	What must be included
Chiral	$2_A+2_F=4$	$4_\psi$	Complex auxiliary $F$
Abelian vector	$4_{A_\mu}-1_{\rm gauge}+1_D=4$	$4_\lambda$	Gauge quotient and real $D$
Real linear	$1_C+3_{H_\mu}=4$	$4_\chi$	Constraint $\partial_\mu H^\mu=0$

This test can reveal an impossible proposed table, but it does not determine the Lorentz representation of the missing auxiliaries or the transformation coefficients. It also cannot by itself decide locality, covariance, finite versus infinite cardinality, or compatibility with an action.

Gauge systems require particular care. Subtracting two on-shell photon polarizations from an off-shell fermion count mixes representation spaces. One must count covariant field functions modulo gauge before field equations, or use a gauge-fixed complex including ghosts and compensating symmetries consistently.

Algebraic elimination as a controlled operation

Suppose

$$S[\phi,Y] =\int\mathrm d^4x\, \left[ \mathcal L_0(\phi) +\frac12Y^aM_{ab}(\phi)Y^b +Y^aJ_a(\phi) \right],$$

with no derivatives acting on $Y$. If $M$ is invertible on the chosen branch,

$$Y^a_{\rm cl} =-(M^{-1})^{ab}J_b.$$

The reduced action

$$S_{\rm red}[\phi]=S[\phi,Y_{\rm cl}(\phi)]$$

has equivalent classical Euler–Lagrange equations for $\phi$ on that branch. Nevertheless, the induced reduced transformation

$$\delta_{\rm red}\phi =\delta\phi\big|_{Y=Y_{\rm cl}}$$

need not close without the $\phi$ equations, because

$$\delta Y_{\rm cl}(\phi) \ne \delta Y\big|_{Y=Y_{\rm cl}}$$

off shell. Their difference is precisely what becomes the equation-of-motion remainder.

Three failure cases must be separated:

• if $M$ is singular, the would-be auxiliary may instead impose a constraint or label branches;
• if quantum corrections generate derivatives of $Y$, it is no longer auxiliary in that effective description;
• if eliminating $Y$ introduces inverse powers of a field that can vanish, the reduced formula is only patchwise valid.

Extended supersymmetry changes the problem

Finite auxiliaries exist for many multiplets, including the four-dimensional $\mathcal N=2$ vector multiplet, whose bosonic auxiliary fields form an $SU(2)_R$ triplet. The hypermultiplet is different: a manifestly Lorentz-covariant formulation with all eight supercharges and no central-charge qualification is naturally described in harmonic or projective superspace with an infinite auxiliary expansion.

For four-dimensional $\mathcal N=4$ Yang–Mills, no finite set of conventional Lorentz-covariant auxiliary fields is known that makes all sixteen supercharges manifest off shell. Siegel and Roček’s counting argument rules out a broad class of previously known finite constructions Siegel and Roček 1981, pp. 275–277. Its conclusion must retain its assumptions: it is not a theorem forbidding every formulation with infinite towers, extra coordinates, partial manifest supersymmetry, nonstandard gauge structure, or weakened locality.

Harmonic and projective superspaces relocate the auxiliary data into functions of additional bosonic coordinates. This is a genuine off-shell construction for important $\mathcal N=2$ systems, but “one analytic superfield” does not mean finitely many component auxiliaries. The next extended-superspace page makes the tower explicit.

Closure checks

Before accepting an auxiliary completion, perform these checks in order:

1. state dimension, signature, real form, and supersymmetry count;
2. list every independent field and gauge equivalence;
3. count off-shell bosonic and fermionic functions on the same space;
4. calculate every component commutator;
5. label translations, gauge transformations, constraints, and equations of motion separately;
6. identify which fields are algebraic in the named action;
7. eliminate them only on a branch where the algebraic system is invertible;
8. recompute the reduced commutator rather than inheriting the old label;
9. state whether the auxiliary set is finite, infinite, or formulation dependent.

The strongest justified conclusion is the weakest closure class found among the components.

Common pitfalls

Auxiliary elimination is not an invertible field redefinition. It uses an Euler–Lagrange equation. The reduced action can be classically equivalent while the reduced transformations close only on shell.

Gauge-fixing variables are not supersymmetry auxiliaries. A Nakanishi–Lautrup field belongs to the BRST gauge-fixing complex; $D$ belongs to the supersymmetry vector multiplet. Their algebraic equations serve different symmetries.

A no-go statement has hypotheses. Finite, local, Lorentz-covariant, manifest, conventional, and without central charge are material qualifiers. Dropping one can change the answer.

Exercises

1. Count before and after elimination

Count the off-shell chiral multiplet before and after setting $F=0$.

Solution

Before elimination, $A$ and $F$ contribute two real components each, matching the four real components of $\psi$. After deleting $F$, the bosonic count is two while the fermionic off-shell count remains four. The mismatch predicts that the reduced transformations cannot close off shell; the fermion equation supplies the missing reduction on shell.

2. Distinguish gauge from dynamics

Does

$$[\delta_1,\delta_2]A_\mu =\xi^\nu\partial_\nu A_\mu+\partial_\mu\Omega$$

require the Maxwell equation?

Solution

No. The second term is tangent to the gauge orbit. The result is off-shell closure modulo gauge provided $\Omega$ is displayed and no equation was used to obtain it.

3. Test an “auxiliary” matrix

What happens if $M_{ab}(\phi)$ in the algebraic system has a zero eigenvalue?

Solution

The formula $Y=-M^{-1}J$ is invalid. Along the null direction the equation may impose a constraint on $J$, leave an undetermined multiplier, or signal a new branch. One must solve the algebraic system by rank strata before calling every $Y^a$ eliminated.

Continue

Constrained, On-Shell, and Nonlinear Superfields studies constraints that remove components, including the nilpotent goldstino multiplet. Extended Superspace Methods and Off-Shell Limits explains how harmonic and projective variables encode infinite auxiliary towers.

References

• Gates, S. James, Jr., Marcus T. Grisaru, Martin Roček, and Warren Siegel. Superspace, or One Thousand and One Lessons in Supersymmetry. Frontiers in Physics 58. Reading, MA: Benjamin/Cummings, 1983. Corrected open edition, 2001. arXiv:hep-th/0108200.

• Siegel, Warren, and Martin Roček. “On Off-Shell Supermultiplets.” Physics Letters B 105 (1981): 275–277. DOI.

• Weinberg, Steven. The Quantum Theory of Fields. Volume III: Supersymmetry. Cambridge: Cambridge University Press, 2000, §§ 26.1–26.4 and §§ 27.1–27.4. DOI.