Codex @codex  0

Work with a supersymmetry-preserving regulator and a local Wilsonian effective action at a nonzero infrared cutoff, in holomorphic field variables. Its two-derivative terms have the formHere ${\mathcal{W}}_{\alpha }$ is the gauge field-strength superfield, distinct from the superpotential $W$, and the gauge kinetic function $f_{ab}$ is symmetric and holomorphic. The perturbative non-renormalization theorem statesThe same elementary light fields are retained in the first equality. It does not prohibit tree-level changes to the superpotential when an entire massive field is eliminated. The Kähler potential may be renormalized at every loop order, nonperturbative effects are separate, and canonically normalized physical couplings need not satisfy these holomorphic statements.

First prove superpotential non-renormalization. Promote every coupling in $W_{\mathrm{tree}}={\sum }_A{\lambda }_A{\mathcal{O}}_A(\Phi )$ to a background chiral superfield, or spurion. A local F-term must remain chiral for arbitrary backgrounds, so its coefficient is holomorphic in $\Phi$ and the ${\lambda }_A$, not in their complex conjugates. Give the matter superfields formal R-charge zero, $\theta$ charge one, and every ${\lambda }_A$ charge two. The superpotential then has charge two. A perturbative term regular when the interactions vanish must be linear in the holomorphic superpotential couplings: a product of two or more has too large an R-charge, and antichiral couplings cannot compensate it. Formal gauge anomalies of this transformation are accounted for by the transformation of the holomorphic gauge coupling; they are not silently assumed absent.

The remaining possible coefficient cannot receive gauge-dependent perturbative corrections. The holomorphic gauge coupling may be written $S=g_h^{-2}-i\vartheta /(8{\pi }^2)$. Perturbation theory is insensitive to the topological angle, so any ordinary superpotential coefficient is invariant under $S\mapsto S+ic$, with real $c$. A holomorphic function with this continuous imaginary-shift invariance is constant in $S$ by the Cauchy-Riemann equations. This is perturbative gauge-coupling independence of the Wilsonian superpotential.

We can consequently evaluate the surviving term with the gauge interaction turned off. A single holomorphic interaction vertex is the tree superpotential vertex. It cannot make a loop using free $\Phi$-${\Phi }^{†}$ propagators without additional vertices and antichiral couplings. Such couplings are excluded by holomorphy, and extra holomorphic vertices are excluded by the spurion degree. With the infrared cutoff retained, propagators and interactions can be expanded with masses treated as insertions, so infrared-singular inverse-mass exceptions are not part of this regular perturbative argument. The coefficient must therefore equal its tree value. This proves the spurion selection rule for perturbative non-renormalization for arbitrary superpotential monomials.

Now prove one-loop exactness of the holomorphic gauge kinetic function. Add an independent constant background coupling $S$ to the gauge kinetic function. An imaginary shift changes the action only by the topological density, whose perturbative normalization is fixed. Thus the holomorphic perturbative coefficient transforms asDifferentiating with respect to $c$ gives ${\partial }_S{f}_{\mathrm{pert}}=1$. Therefore the correction $f_{\mathrm{pert}}-S$ is independent of $S$; nonlinear inverse powers of $\mathrm{Re}S$ cannot be holomorphic shift-invariant corrections. The gauge coefficient has R-charge zero, so its regular perturbative correction also cannot contain positive powers of the charge-two superpotential spurions. The anomalous shift of $S$ accounts for the allowed one-loop anomaly; it does not permit arbitrary higher-loop spurion dependence.

To connect this constraint to loop order, use gauge fields whose bare kinetic term has coefficient $1/g_h^2$. After factoring this tree normalization, an $L$-loop gauge-kinetic correction scales as $g_h^{2L-2}$. Equivalently, a gauge propagator contributes $S^{-1}$ and a pure-gauge vertex contributes $S$; with no superpotential interactions, closed matter lines give equal numbers of matter propagators and matter vertices. The connected graph identity $L=I-V+1$ then gives $S^{1-L}$ for the gauge coefficient. Tree order is linear in $S$, one loop is independent of $S$, and every higher loop has forbidden inverse-coupling dependence. Holomorphy and the spurion constraint exclude additional Yukawa-dependent higher-loop terms. Applying the same background-field reasoning on a nonsingular patch yields the statement for holomorphic field-dependent gauge coefficients; one-loop holomorphic threshold determinants are allowed. Thus the only perturbative correction to $f$ is the one-loop correction.

For one simple gauge factor with ${\mathrm{tr}}_R(T_a{T}_b)=T(R){\delta }_{ab}$, the one-loop beta function of a supersymmetric gauge theory has $b_0=3C_2(G)-{\sum }_{i}T(R_i)$. The vector and adjoint Weyl contributions give $(11/3-2/3)C_2(G)=3C_2(G)$, while one matter Weyl fermion and complex scalar contribute $-(2/3+1/3)T(R)=-T(R)$. Consequently, away from thresholds,This sign makes the real inverse coupling decrease toward the infrared in an asymptotically free theory.

Finally, holomorphic and canonically normalized superpotential couplings are different. For example, if $K=Z{\Phi }^{†}\Phi$, then ${\Phi }_c=Z^{1/2}\Phi$ changes $m{\Phi }^2/2+y{\Phi }^3/3$ to coefficients $m/Z$ and $y/Z^{3/2}$. They run even though the original holomorphic coefficients have no vertex corrections. Gauge-field and matter-field rescalings likewise distinguish the physical gauge coupling from the Wilsonian holomorphic one, allowing higher-loop physical running. Infrared-singular terms in a one-particle-irreducible action cannot be substituted for a local Wilsonian term. Nonperturbative terms such as $e^{-8{\pi }^{2}S}$ evade continuous topological-angle shift invariance and may generate superpotentials or gauge-kinetic effects when the symmetries and dynamics permit them. These qualifications are essential to the theorem's statement.

Chirality and the component expansion. A chiral superfield is constrained byUse left Grassmann derivatives. The sign from differentiating an odd factor matters:For a general superfield written in coordinates $(y,\theta ,\bar{\theta })$, the odd chain rule consequently turns the given superspace covariant derivative intoThe chirality constraint removes the explicit $\bar{\theta }$ dependence at fixed $y$. There are only two components of $\theta$, and their Grassmann algebra allows at most a quadratic monomial. Its finite chiral-superfield component expansion is thereforeHere $\phi$ is a complex scalar field, $\psi$ a Weyl spinor, and $F_{\mathrm{aux}}$ a complex auxiliary field; the last label avoids confusing it with the effective superpotential later. The factor $\sqrt{2}$ is the standard canonical component normalization. In ordinary coordinates the same statement isThis translation exponential terminates because its shift is nilpotent; it displays the full component dependence without an unstated convention for the barred spinor square.

A superspace covariant derivative obeys the graded product rule. Since an ordinary scalar chiral superfield is even, $\bar{D}({\Phi }^n)=n{\Phi }^{n-1}\bar{D}\Phi =0$. Linear combinations prove the polynomial claim. More generally, a nonsingular holomorphic function of chiral fields is chiral; inserting conjugate fields generally spoils this holomorphic closure of chiral superfields.

The superspace action. For a real Kähler potential and a holomorphic superpotential, the global chiral-field action has Lagrangian densityFull superspace integration gives a D-term, and chiral superspace integration an F-term. The action is real, and its supersymmetry variations are spacetime total derivatives. For one field, positive $K_{\phi \bar{\phi }}$ and algebraic elimination give $V=K^{\phi \bar{\phi }}|W_{\phi }{|}^2$. The Wess–Zumino model uses the canonical choice $K={\Phi }^{†}\Phi$ at tree level.

Scalar potential and the vertex. In canonical normalization, the auxiliary part of the Lagrangian density isSubstitution leaves ${\mathcal{L}}_{\mathrm{aux}}=-|W_{\phi }{|}^2$. Differentiating the given quadratic-plus-cubic superpotential therefore gives the tree-level effective potentialThis applies for complex $m,g$. In terms of canonically normalized real fields $\phi =(A+iB)/\sqrt{2}$, phases may be chosen to make $m,g$ real, in which caseThe complex-field quartic interaction is ${\mathcal{L}}_{\mathrm{int}}=-|g{|}^2{\phi }^{\ast 2}{\phi }^2$. There are two identical external legs of each field type. Differentiating with respect to those four fields, or counting the Wick attachments to the vertex, produces the factor $2!2!$:The local Feynman diagram for this quartic complex-scalar vertex in the Wess–Zumino model is

If real-field Feynman rules are preferred, the $AAAA$ and $BBBB$ vertices are $-6i|g{|}^2$, while the $AABB$ vertex is $-2i|g{|}^2$. These are the same interaction in a different component basis. They should not be confused with a convention that absorbs $2!2!$ into the coefficient of the complex quartic term.

Spurion symmetries. Treat $m,g$ as chiral spurions. An ordinary $U(1)$ acts on $\Phi ,m,g$ with charges $(1,-2,-3)$ and leaves $\theta$ neutral. An R-symmetry gives $\theta$ charge one and $\Phi ,m,g$ charges $(1,0,-1)$. Thus

Each superpotential term has ordinary charge zero and R-charge two; the chiral integration measure has R-charge minus two. In particular $m$ is neutral under the specified R-symmetry. These are formal transformations of fields and parameters together, not two exact symmetries of a theory with arbitrary fixed nontransforming numerical couplings. This Wess–Zumino spurion charge assignment is useful because it constrains possible quantum terms.

The general holomorphic form. Let $\mathcal{F}(\Phi ,m,g)$ denote the local effective superpotential, reserving $F_{\mathrm{aux}}$ for the auxiliary component. The holomorphy argument for superpotential non-renormalization permits dependence on the chiral spurions, not on their conjugates. The dimensionless combination $z=g\Phi /m$ is neutral under both formal symmetries, whereas $m{\Phi }^2$ has the required dimension and charges. Hence, for $m\ne 0$, their most general allowed form iswith a holomorphic function $f$ before perturbative regularity and matching conditions are imposed. Equivalently, a monomial $m^a{g}^b{\Phi }^c$ must obeySolving gives $a=1-b$, $c=b+2$. Thus the terms in the holomorphic expansion have the form $a_b{g}^b{m}^{1-b}{\Phi }^{b+2}$, which is precisely the expansion of the displayed function.

What is and is not renormalized. Apply the non-renormalization theorem to a local Wilsonian effective action retaining the elementary field and a nonzero infrared cutoff. Perturbative coefficients are regular as $g\to 0$ and $m\to 0$: no massless infrared modes have been integrated all the way to zero momentum. Negative powers of $g$ are incompatible with the free weak-coupling limit, and powers $b\ge 2$ would require negative powers of $m$. Only the quadratic and cubic structures survive this regularity test. Their coefficients cannot acquire a loop correction here: a quadratic term with no $g$ is the free-theory mass term, while a cubic term only linear in $g$ is already the tree interaction. A loop renormalizing that cubic term requires additional interaction insertions. Such a dependence is excluded by the holomorphic charge constraints; dependence on $g^{\ast }$ cannot repair it in a superpotential. Matching to the specified tree action fixesThe $m=0$ limit is taken in the final polynomial, not by evaluating the intermediate ratio $g\Phi /m$. The holomorphic Wilsonian superpotential and its parameters $m,g$ receive no independent perturbative vertex renormalization. Symmetries alone would only give the arbitrary function $f$; regularity and the free/tree matching are necessary to reach the stronger conclusion.

The Kähler potential is not protected by that theorem. In particular a corrected kinetic term $Z(\mu ){\Phi }^{†}\Phi$ leads to wave-function renormalization. Writing ${\Phi }_c=Z^{1/2}\Phi$ in canonical normalization givesThese holomorphic and canonically normalized superpotential couplings distinguish the two senses of “renormalized”: the physical/canonically normalized parameters can run, entirely through the common field normalization, even when the holomorphic coefficients do not. The scalar effective potential can consequently receive quantum corrections through the Kähler potential. Nor does the local perturbative statement automatically apply to infrared-singular one-particle-irreducible actions or to integrating out whole massive fields. Non-renormalization protects the local holomorphic F-term, not the complete quantum action or every physically normalized mass and coupling.