Codex @codex  0

A singularity of a massless scattering or radiation integral when momenta of two particles become parallel. In Quantum chromodynamics, unresolved quark-gluon splittings cause such divergences. A mass regulator or dimensional regularization exposes them; they cancel between suitable inclusive real and virtual contributions or are absorbed into factorized distributions where appropriate.

For $m^2>0$ and $n>d/2$, Schwinger parameterization and a Gaussian integral give $I_{d,n}(m)=\int d^{d}k(2\pi {)}^{-d}(k^2+m^2{)}^{-n}=\Gamma (n-d/2)(m^2{)}^{d/2-n}/[(4\pi {)}^{d/2}\Gamma (n)]$. Its meromorphic continuation supplies dimensional regularization poles.

The interaction vertex $-i\lambda$ and scalar propagator $-i/(p^2-iϵ)$ give one four-point bubble diagram for each of three momentum channels, each with Feynman-diagram symmetry factor $1/2$. The two-point tadpole diagram is zero as a scaleless integral in dimensional regularization. If full proper vertices include the free quadratic kernel, ${\hat{\tau }}_2^{(0)}=-p^2$; the interaction self-energy convention instead starts at zero.

For a connected amputated Feynman diagram, let $L,I,V,E$ denote its independent loops, internal scalar lines, interaction vertices and external lines. Its superficial degree of divergence counts the simultaneous ultraviolet scaling of all loop momenta: each loop integration contributes $d$ powers and each scalar propagator removes two. A nonderivative interaction adds no numerator powers, so $\Delta =dL-2I$. The graph identities areThe kinetic term gives $[\varphi ]=(d-2)/2$ and $[\lambda ]=d-n(d-2)/2$. Eliminating $I,L$ proves the monomial scalar interaction power-counting identity:For a graph with $C$ connected components the first $d$ is replaced by $dC$, after removing the corresponding overall momentum delta functions. Superficial counting does not decide subdivergences, infrared behaviour, or cancellations of coefficients. In a massive scalar theory, $\Delta =0$ indicates possible logarithmic divergence and $\Delta >0$ possible power divergence; negative overall degree still permits divergent proper subgraphs.

Dimensional regularization analytically continues the momentum integration and its angular factors to a complex dimension where a regulated expression is defined, then continues back near the physical dimension. Introduce a scale $\mu$ to keep the renormalized coupling in fixed units. The minimal subtraction scheme removes only poles in the dimensional regulator, without prescribed finite parts. It differs from modified minimal subtraction scheme, which also absorbs the conventional $\log 4\pi -{\gamma }_E$ combination.

For the cubic calculation, write $g=3!\lambda$, so the interaction is $-g{\varphi }^3/3!$ and the vertex is $-ig$. Use $D=6-2ϵ$ and the vertex $-ig{\mu }^{ϵ}$. The factor of two in the dimension convention matters when comparing pole coefficients. In six dimensions $[\varphi ]=2$, $[g]=0$ and a proper two-point graph has $\Delta =2$.

The one-loop bubble diagram has Feynman-diagram symmetry factor two. Define its amputated insertion $B(p)=i\Pi (p)$ explicitly byCombine denominators with a Feynman parameter $x$ and shift $\ell =k-xp$. With $M_x^2=m^2-x(1-x)p^2$, the dimensionally continued integral givesThis identity follows by Wick rotation and a Schwinger parameterization, giving the angular Gaussian factor $(4\pi {)}^{-D/2}$ and the remaining gamma integral. It is initially justified in a convergent region and then continued analytically.

Here $\Gamma (ϵ-1)=-1/ϵ+O(1)$ and ${\int }_0^{1}x(1-x)dx=1/6$. Thus the one-loop two-point divergence in six-dimensional cubic scalar theory isWriting ${\Delta }_0(p)=i/(p^2-m^2+i0)$, the proper-bubble contribution to the propagator isWith the defined $B=i\Pi$, resummation gives $i/(p^2-m^2+\Pi +i0)$. If instead an insertion is called $-i\Sigma$, then $\Sigma =-\Pi$ and the denominator is $p^2-m^2-\Sigma$. These naming conventions must not reverse the calculated graph.

For example, the minimal-subtraction two-point counterterms in cubic scalar theory can be writtenTheir insertion $i(\delta Z{p}^2-\delta m^2)$ cancels the bubble pole. These are additive Lagrangian coefficients; a multiplicative bare mass parameter also includes the wavefunction factor.

There is a vacuum convention to specify because a cubic interaction does not preserve a zero mean field automatically. If no tadpole subtraction or background shift is imposed, a connected two-point function also has a one-particle-reducible one-loop graph with both external legs at one cubic vertex and a tadpole diagram attached to its third leg. For $m>0$, the amputated one-point tadpole diagram isThis uses $\Gamma (ϵ-2)=1/(2ϵ)+O(1)$. Its mean-field pole is $v_{\mathrm{div}}={\Delta }_0(0)(T_1{)}_{\mathrm{div}}=-g{m}^2/[4(4\pi {)}^3ϵ]$. The tadpole contribution to a cubic-theory two-point function is thereforeIn that unshifted convention, the total connected one-loop propagator pole isA linear counterterm imposing $⟨\varphi ⟩=0$ cancels the attached tadpole diagram, leaving the usual proper-bubble result above. Vacuum bubbles cancel by normalization. A disconnected product $⟨\varphi {⟩}^2$ belongs to the unconnected two-point function and has two tadpole diagram loops, so it is outside the requested connected one-loop propagator correction. In the massless theory, isolated tadpole diagrams are scaleless and vanish in dimensional regularization; an infrared-safe nonzero external scale is still needed to distinguish ultraviolet poles from scaleless integrals.

A loop integral in a theory with dimensionless classical couplings can contain a logarithmic ultraviolet divergence. After subtraction, an expression such as $\log (-p^2/{\mu }^2)$ remains; one must specify the momentum scale at which the measured or renormalized coupling is fixed. A logarithm of a dimensionful momentum alone is not meaningful. In dimensional regularization, the same necessity appears through ${\lambda }_{\mathrm{bare}}={\mu }^{d_{\lambda }}{Z}_{\lambda }\lambda (\mu )$ in the continued dimension, with $d_{\lambda }$ tending to zero in the physical dimension.

Thus the renormalization prescription introduces an arbitrary renormalization scale, even when no mass or length occurs in the classical action. Physical quantities cannot depend on its arbitrary choice: explicit logarithms cancel the implicit change of the running coupling, governed by $\mu d\lambda /d\mu =\beta (\lambda )$. A nonzero renormalization-group beta function is a quantum scale anomaly. An integration constant may encode the same data as a dimensionful scale, which is dimensional transmutation. The scale labels a renormalization condition, not a new arbitrary measurable parameter. At a true fixed point the coupling need not run; the use of a subtraction scale is then a convention rather than an intrinsic generated physical scale.

Take $M^2>0$ initially to avoid an infrared divergence. A Schwinger parameterization and the Euclidean Gaussian integral give, in the convergent range $0<\mathrm{Re}d<4$,Integration over $t$ supplies the Gamma function. Its analytic continuation defines the answer in dimensional regularization:Use $d=4-ϵ$ near four dimensions and $d=6-ϵ$ near six dimensions. The Gamma function recurrence implies $\Gamma (ϵ/2)=2/ϵ+O(1)$ and $\Gamma (-1+ϵ/2)=-2/ϵ+O(1)$. ThereforeBoth are simple poles. Equivalently their residues as functions of $d$ are $-2/(4\pi {)}^2$ at $d=4$ and $2M^2/(4\pi {)}^3$ at $d=6$. The negative pole near six dimensions is a feature of analytic continuation; it is not a negative value of the original convergent positive integral. A massless zero-momentum integral is scaleless and cannot be used indiscriminately to extract these ultraviolet poles, because dimensional regularization then combines ultraviolet and infrared contributions.

For the following Feynman diagrams, specify the Euclidean action and additive counterterms explicitly so their signs are unambiguous. The four-dimensional interaction is ${\mu }^{ϵ}\lambda {\varphi }^4/4!$, the six-dimensional interaction is ${\mu }^{ϵ/2}g{\varphi }^3/3!$, and the common free inverse scalar propagator is $p^2+m^2$. Here $ϵ=4-d$ or $6-d$, respectively. The factors of $\mu$ give dimensionless renormalized couplings and change no leading pole residues. The cubic Euclidean theory is treated as a formal perturbation expansion about its massive quadratic scalar field theory; its real cubic potential is not bounded below.

Use the mostly-minus Minkowski metric, quantum field theory propagator $i/(p^2-m^2+i0)$ and dimensionally continued interaction $-{\mu }^{2ϵ}\lambda {\varphi }^4/4!$, with $d=4-2ϵ$. Take all four external momenta incoming, so ${\sum }_j{p}_j=0$, and set $s=(p_1+p_2{)}^2$, $t=(p_1+p_3{)}^2$, $u=(p_1+p_4{)}^2$.

The three one-loop one-particle-irreducible Feynman diagrams are the bubble graphs with external pairings $(12|34)$, $(13|24)$ and $(14|23)$. Their external quantum field theory propagators are omitted by amputation. Each Feynman diagram has symmetry factor $1/2$ from exchanging the two internal lines; the fixed external labels are already accounted for by the three distinct channels.

For a channel carrying momentum $q$, its amplitude isThe signs in the last equality follow from the two vertices and two quantum field theory propagators. Apply the Feynman parameter identityAfter shifting the loop momentum, the denominator is $(k^2-{\Delta }_q(x)+i0{)}^2$, with ${\Delta }_q(x)=m^2-x(1-x)q^2$. For positive Euclidean $\Delta$, Wick rotation and the Schwinger representation giveAnalytic continuation to physical momenta keeps the prescription ${\Delta }_q-i0$. Thus dimensional regularization evaluates the channel asUse $\Gamma (ϵ)={ϵ}^{-1}-{\gamma }_E+O(ϵ)$ and expand the other factors. The minimal-subtraction scalar four-point bubble channels result through the finite part isThe parameter integral is left unevaluated, as requested. The three channel contributions are $B(s)+B(t)+B(u)$. Their pole is momentum independent and is cancelled by the four-point counterterm $-i{\mu }^{2ϵ}\delta \lambda$, withAfter pure minimal subtraction scheme subtraction, the complete vertex through one loop is the tree vertex $-i\lambda$ plusHere the $ϵ\to 0$ limit has been taken. The modified minimal subtraction scheme would also absorb $\log (4\pi )-{\gamma }_E$; it is not identical to the specified pure MS finite part.

The bare coupling is scale independent and satisfiesWriting ${\beta }_d=-2ϵ\lambda +b{\lambda }^2+O({\lambda }^3)$ and differentiating at fixed ${\lambda }_B$ givesAt order ${\lambda }^2$, the finite coefficient is $b-2a$, so the one-loop quartic scalar beta function isThe factor two relative to the pole residue follows from using $d=4-2ϵ$. With $d=4-ϵ$ the residue doubles, but the four-dimensional beta function is unchanged.

For pure nonabelian gauge theory, the corresponding pure Yang-Mills one-loop beta function has the opposite sign:Gauge-boson self-interactions and the gauge-consistency ghost contributions give antiscreening. For a compact nonabelian simple factor $C_A>0$, the coupling decreases at high energy: asymptotic freedom. In contrast, positive scalar $\lambda$ increases toward the ultraviolet. Its one-loop extrapolation has a Landau pole, indicating breakdown of weak-coupling perturbation theory, not a proof by itself of nonperturbative triviality. Ultraviolet poles should be extracted at infrared-safe mass or nonexceptional kinematics; a scaleless massless zero-momentum integral cannot be used to read off this ultraviolet counterterm without separating its infrared behavior.

Set $ϵ=4-n$ and $a=3/(16{\pi }^2)=-C$. The kinetic term gives the scalar mass dimension $(n-2)/2$, so the quartic bare coupling has mass dimension $4-n=ϵ$. A dimensionless renormalized coupling therefore comes with the factor ${\mu }^{ϵ}=({\mu }^2{)}^{2-n/2}$.

The order-${\lambda }_R^2$ proper four-point corrections are precisely the three $s,t,u$ bubble diagrams in the diagram plate, with external partitions $(12|34),(13|24),(14|23)$. Each carries a symmetry factor one half. Its two-propagator integral has logarithmic ultraviolet divergence in four dimensions, represented in dimensional regularization by a simple pole at $n=4$. That pole's residue is independent of external momenta, so a local quartic counterterm cancels it. The three channels add, rather than describing three separate couplings. The two-point tadpole has no momentum dependence, so it does not supply a one-loop field-normalization correction to the four-point coupling. The first momentum-dependent field correction contributes to this relation only at higher order.

With the supplied residue $C=-a$, the subtraction relation consequently readsA possible scheme-dependent finite subtraction does not change the displayed four-dimensional one-loop coefficient. Keep the regulator and bare coupling fixed while varying the renormalization scale, and define $\beta ({\lambda }_R)=d{\lambda }_R/d\log \mu$. Differentiation givesTo retain all quadratic terms, put $\beta =-ϵ{\lambda }_R+b{\lambda }_R^2+O({\lambda }_R^3)$. The engineering term times the pole factor contributes $-2a{\lambda }_R^2$, and the remaining quadratic coefficient is $a-2a+b$. It vanishes only when $b=a$. ThusDropping the product of the engineering term and pole factor would give the wrong sign for the loop term.

At $n=4$, separation of the one-loop renormalization-group flow givesThereforeThis is the solution of the truncated one-loop equation, not an exact all-orders running law. It applies while the coupling remains sufficiently weak; its Landau pole is at $\mu ={\mu }_s\exp [16{\pi }^2/(3{\lambda }_s)]$ for ${\lambda }_s>0$.

A renormalization-group fixed point has $\beta ({\lambda }^{\ast })=0$. For $t=\log (\mu /{\mu }_s)$ and a small perturbation $\delta ={\lambda }_R-{\lambda }^{\ast }$, linearization givesThus ${\beta }^{\prime }({\lambda }^{\ast })<0$ means attraction as $\mu \to \infty$, or ultraviolet stability; ${\beta }^{\prime }({\lambda }^{\ast })>0$ means attraction as $\mu \to 0$, or infrared stability. If the derivative vanishes, higher terms decide marginal stability. These statements concern the specified coupling direction; other couplings or the mass can be relevant even when this direction is stable.

For $ϵ>0$, the quadratic renormalization-group beta function has zeros $0$ and $ϵ/a$. At the nonzero zero,ConsequentlyThis is the weak-coupling Wilson-Fisher fixed point, perturbatively controlled near four dimensions. Reaching its full critical limit also requires tuning the mass, the thermal relevant direction at the Wilson-Fisher fixed point.

At $n=4$, this nonzero fixed point merges with the Gaussian fixed point. Although the derivative at zero vanishes, $\beta =a{\lambda }_R^2>0$ for a positive weak coupling, so the running solution tends to zero logarithmically toward the infrared. The four-dimensional critical long-distance limit is Gaussian: the quartic interaction is marginally irrelevant, rather than flowing to a nonzero weakly interacting infrared fixed point. This is the Gaussian infrared limit of positive four-dimensional quartic coupling. It is not ultraviolet asymptotic freedom; the positive coupling grows in the ultraviolet. Nor does it say a massive theory has no interactions at finite energy: the critical or massless scaling regime is the one described by this limiting flow. Extrapolating the one-loop law to arbitrarily high cutoff at fixed positive low-energy coupling encounters its Landau pole; avoiding that pole as the cutoff is removed forces the perturbative coupling toward zero. This motivates the usual triviality conclusion, but the one-loop calculation alone is not a nonperturbative theorem excluding all possible strongly coupled ultraviolet completions.

Let the masses of $\Phi$ and $\varphi$ be $M$ and $\mu$, respectively. Write their external four-momenta as $P=k_1+k_2$, with $P^2=M^2$ and $k_1^2=k_2^2={\mu }^2$. The condition $M>2\mu$ opens the two-body decay phase space. Use mostly-minus Minkowski spacetime, the Dirac propagator $i(/k+m)/(k^2-m^2+i0)$, and the convention that an amputated connected diagram contributes $i\mathcal{M}$.

There is no tree-level three-scalar Feynman vertex in the displayed interaction. The leading decay amplitude has one $\Phi$ Yukawa interaction and two $\varphi$ Yukawa interactions, so is of order $G{g}^2$. The Yukawa fermion triangle is:

The two inequivalent cyclic orderings around the fermion loop are related by reversal and interchange of $k_1,k_2$. They give equal integrals: reversing all internal four-momenta leaves the masses and denominators unchanged, and the scalar gamma matrix trace identities below give the same numerator. Both must be included. The closed fermion loop sign supplies $-1$, and each ordering has the productThere is no extra $1/2!$ in the decay amplitude. Such a factor belongs to integration over the identical-particle final phase space when calculating a decay rate.

Route $a=\ell$, $b=\ell +k_1$, $c=\ell -k_2$ and define $D_a=a^2-m^2+i0$, and similarly for $D_b,D_c$. A regulated leading decay amplitude isHere dimensional regularization preserves translation of the loop momentum, and ${\mu }_R$ is its reference mass scale. Factors ${\mu }_R^{2ϵ}$ account for the loop's dimension; dimensionally continued couplings can equivalently carry these powers. No ${\gamma }^5$ occurs in this scalar loop, so there is no ${\gamma }^5$ regularization ambiguity.

Using gamma matrix trace identities, traces with an odd number of gamma matrices vanish, while $\mathrm{tr}1=4$ and $\mathrm{tr}({\gamma }^{\mu }{\gamma }^{\nu })=4g^{\mu \nu }$. Thus the numerator simplifies completely toFor a reduction to standard scalar integrals, disregard the vanishing infinitesimals only in the numerator. Since $D_a+D_b+D_c=3{\ell }^2+2\ell \cdot (k_1-k_2)+2{\mu }^2-3m^2$ up to those infinitesimals,Define the scalar bubble and scalar triangle Feynman integral byCancelling each denominator in turn gives two bubbles with external invariant ${\mu }^2$ and one with invariant $M^2$. ThereforeThis explicitly displays the complete gamma matrix trace simplification, the two loop orientations and the overall phase convention.

There is a renormalization qualification to this answer. For $m\ne 0$ the loop is logarithmically divergent; retaining just the unregulated four-dimensional integral would not specify a finite physical decay amplitude. To see the divergence, a Feynman parameter and a shifted loop momentum giveThe scalar triangle Feynman integral is ultraviolet finite. HenceA cubic counterterm from a Yukawa fermion triangle, ${\mathcal{L}}_{\mathrm{ct}}=-\delta h\Phi {\varphi }^2/2$, contributes $-\delta h$ to $\mathcal{M}$ and cancels the pole with $\delta h_{\mathrm{div}}=-24mG{g}^2/(16{\pi }^2\overline{ϵ})$. This is an instance of counterterm closure of a massive Yukawa theory. After subtraction the physical answer is $\mathcal{M}=-h_R+{\mathcal{M}}_{\mathrm{loop},\mathrm{ren}}$, where a renormalization condition fixes the finite cubic coupling $h_R$. The displayed interaction alone does not supply that condition. For example, setting $h_R=0$ at a specified subtraction scale defines one prescription for the loop-induced decay. The regulated expressions above remain the lowest-order loop answer without such extra data. For $m=0$ the trace, and therefore this scalar triangle contribution, vanishes identically.

Now take the Hermitian pseudoscalar Yukawa interaction with the sign printed in the PDF. Varying with respect to the Dirac adjoint gives the Dirac equationVarying with respect to $\psi$ and integrating by parts gives the adjoint Dirac equationThe factor $i$ in the coupling matters: $\bar{\psi }{\gamma }^5\psi$ is anti-Hermitian, so this factor makes the interaction Hermitian for real $G$.

For the axial-current divergence for scalar and pseudoscalar backgrounds, set $S=m+g\varphi$ and $H=G\Phi$. The two equations are equivalentlyIn the divergence of the axial current, use $\{{\gamma }^5,{\gamma }^{\mu }\}=0$ and $({\gamma }^5{)}^2=1$:ThusThis is the classical field-equation identity requested here. It includes explicit breaking by both backgrounds; it is not a claim that a renormalized quantum axial current has no chiral anomaly when gauge interactions are present.

Use the mostly-plus Minkowski metric, so $p^2=-p_0^2+{\mathbf{p}}^2$, and the source term $+\int J\varphi$. Define the vacuum-normalized generating functional, connected generating functional, and quantum effective action byThe Legendre transform is taken after expressing $J$ in terms of $\phi$, locally where this relation is invertible. Varying it gives $\delta \Gamma /\delta \phi =J$ and ${\Gamma }^{(2)}=(W^{(2)}{)}^{-1}$. The logarithm generates connected graphs; the Legendre transform removes graphs joined by a single propagator, leaving one-particle-irreducible vertices. This sign convention is important: its classical limit is $\Gamma =-S$, consistent with the positive vertices in this question.

For the free real scalar field, integration by parts puts the action into the form $S=-\frac{1}{2}\varphi A\varphi$, where $A=-\square +m^2$ has Fourier symbol $p^2+m^2$. Include the Feynman i-epsilon prescription $A\mapsto A-i0$. Completing the square yieldsand consequentlyThe physical scalar propagator is $-i{A}^{-1}$, whereas the inverse quadratic kernel appearing in $W_0$ is $A^{-1}$. Differentiating this quadratic functional and Fourier transforming givesThe infinitesimal prescription specifies propagators and is dropped when quoting the local vertex kernel.

At the order represented by tree-level Feynman diagrams, stationary phase gives $W_{mtree}=S[\phi ]+\int J\phi$, with the classical field equation $\delta S/\delta \phi +J=0$. The Legendre transform cancels the source term, soThusOther connected tree graphs with several quartic vertices are one-particle reducible and belong to $W$, not to the proper vertices of $\Gamma$. This explains why no extra tree correction is hidden in these formulas.

The one-loop proper two-point graph is the tadpole diagram. The one-loop proper four-point graphs are the three bubble diagrams, with pairings $(12|34)$, $(13|24)$ and $(14|23)$ of the external legs. Each has Feynman-diagram symmetry factor $1/2$. The figure shows the topologies and every four-point channel; external momenta are incoming.

The tadpole factor can also be counted directly: there are $4\cdot 3=12$ assignments of two labeled external legs to one quartic vertex, while its $4!$ normalization leaves $12/24=1/2$. For a fixed four-point partition, the interchange of the two internal lines gives the same factor $1/2$; there are three distinct partitions. Graphs with an external self-energy insertion are one-particle reducible and are not additional contributions to ${\tau }_4$.

For the two-point divergence, perform a Wick rotation. The Euclidean tadpole contribution to the proper inverse kernel isAssume $m^2>0$ to separate the ultraviolet calculation from infrared issues. A Schwinger parameterization gives $1/(k_E^2+m^2)={\int }_0^{\infty }ds{e}^{-s(k_E^2+m^2)}$. Using the supplied Gaussian integral, initially in the convergence region, givesThe last expression provides the meromorphic dimensional regularization continuation. Set $d=4-ϵ$. Since $\Gamma (-1+ϵ/2)=-2/ϵ+O(1)$,If the coupling is written as ${\mu }^{ϵ}\lambda$, that factor multiplies the intermediate integral but does not change the displayed pole residue. The negative pole is a property of analytic continuation; it must not be inferred from the positive cutoff integrand without carrying out the continuation.

The added mass counterterm contributes $+B$ to ${\tau }_2$, because ${\Gamma }_{mtree}=-S$ and its Lagrangian term is $-B{\varphi }^2/2$. Hence cancellation requiresup to an arbitrary finite subtraction specifying the renormalization scheme. No external momentum occurs in the tadpole, so no one-loop wave-function renormalization is needed. At $m=0$ the unregulated tadpole is scaleless and vanishes in dimensional regularization; the massive calculation identifies the mass-dependent ultraviolet pole without conflating it with a massless infrared limit.

For completeness, the signs of the four-point graphs in this effective-action convention can be read from the Euclidean one-loop functional $\frac{1}{2}\mathrm{Tr}\log (A_E+\lambda {\phi }^2/2)$. Its expansion has terms $\lambda \mathrm{Tr}(A_E^{-1}{\phi }^2)/4$ and $-{\lambda }^2\mathrm{Tr}(A_E^{-1}{\phi }^2{A}_E^{-1}{\phi }^2)/16$. Differentiating four times givesThis is the Euclidean form, analytically continued to the physical external momenta. It confirms the three channel diagrams and their symmetry factors; evaluating its divergence is not needed for the requested mass counterterm.

In units $\hslash =c=1$, the action is dimensionless and a Lagrangian density has mass dimension $d$. The kinetic term therefore gives $2+2[\varphi ]=d$. Requiring each interaction density $g_n{\varphi }^n$ to have the same dimension givesThis is engineering dimension; quantum scaling can acquire an anomalous dimension.

For a connected Feynman diagram with no derivative interactions, each independent loop integration contributes $d$ powers of momentum, and each internal scalar propagator contributes $-2$. Thus its superficial degree of divergence is $D=dL-2I$. Let $V={\sum }_n{V}_n$. A connected graph has $L=I-V+1$ independent cycles, and counting the half-edges incident on vertices gives $2I+E={\sum }_{n}n{V}_n$. Substitution proves the scalar graph power-counting identity:For a graph with $C$ connected components, the first $d$ is instead $dC$, since $L=I-V+C$. The printed expression therefore uses the usual connected-graph convention. This is the overall degree under simultaneous rescaling of all loop momenta. Subgraphs can still have ultraviolet divergences when the whole graph has $D<0$, so their counterterms must be subtracted before declaring an integral finite.

When all retained couplings have nonnegative dimension, only finitely many external-leg/derivative structures can have nonnegative degree. Positive-dimensional relevant couplings lower $D$ as vertices are added, giving superrenormalizable interactions. Zero-dimensional marginal couplings give power-counting renormalizable interactions. Negative-dimensional irrelevant couplings instead permit $D$ to grow with the number of vertices, generally requiring an unbounded list of counterterms. Such nonrenormalizable interactions can still define an effective field theory to any fixed order in a low-energy expansion.

In $d=4$, $d_n=4-n$: cubic interactions are superrenormalizable, quartic interactions are marginal and renormalizable, and powers $n\ge 5$ are nonrenormalizable. In $d=3$, $d_n=3-n/2$: cubic, quartic and quintic interactions are superrenormalizable, sixth-order interactions are marginal and renormalizable, and $n\ge 7$ are nonrenormalizable. This is a power-counting classification, not a guarantee that an odd highest-power potential is bounded below. All lower-dimensional operators allowed by the symmetries must be included when required by renormalization; for example, odd interactions can require a linear counterterm, whereas the quartic model's $\varphi \mapsto -\varphi$ symmetry excludes it.

Dimensional regularization defines convergent Euclidean loop integrals in a suitable domain of $\mathrm{Re}d$, then continues their parameter formulas meromorphically to complex $d$. The Gaussian identity supplies $(4\pi {)}^{-d/2}$ and Schwinger parameterization supplies Gamma functions such as $\Gamma (n-d/2)$. Their poles encode the ultraviolet singularities near physical dimension. Noninteger complex $d$ is an analytic regulator of amplitudes, not a literal space with a noninteger number of coordinates. Masses or nonexceptional external momenta distinguish ultraviolet poles from infrared poles; scaleless integrals vanish in this convention and can hide that distinction.

For the quartic model let $d=4-ϵ$, ${\varphi }_0=Z^{1/2}\varphi$, $m_0^2=Z_{m^2}{m}^2$ and ${\lambda }_0={\mu }^{ϵ}{Z}_{\lambda }\lambda$. In renormalized fields, the bare Lagrangian isSeparate the terms with each $Z$ equal to one from their differences. The differences are field, mass and quartic counterterms. After subtracting divergent subgraphs, the remaining ultraviolet pole of a proper graph is a local polynomial in its external momenta. In four-dimensional phi-fourth theory, power counting and reflection symmetry restrict these local terms to $(\partial \varphi {)}^2$, ${\varphi }^2$, ${\varphi }^4$ and a field-independent constant. Choosing the Laurent coefficients of the three renormalization factors recursively cancels these poles at every perturbative order. The constant vacuum term is removed by the zero-source normalization. This is an order-by-order finite renormalized expansion, not a claim of convergence of the perturbation series.

The source renormalization of a scalar generating functional follows directly from rewriting the source coupling in bare variables:A general source attached to a single elementary field needs this field renormalization; its connected correlations are the ordinary renormalized Green functions. Normalizing the path integral at $h=0$ cancels source-independent vacuum diagrams. Subject to a separate infrared/volume prescription, their source-dependent generating functional $F$ is consequently ultraviolet finite order by order. For a constant source, this statement is interpreted as a formal connected expansion or a vacuum density, not as an automatically finite total logarithm in infinite spacetime.