Renormalons, OPE Ambiguities, and Transseries

Renormalon growth arises when a perturbative coefficient samples an ever wider range of loop momenta and logarithmic moments generate $n!$. A bubble-chain calculation makes the mechanism and its power scale explicit, but it is a diagnostic approximation, not a proof of the full theory’s Borel singularities. For an infrared singularity, the resulting prescription dependence must have the same power in $1/Q$ as an allowed operator-product contribution so that the complete observable can be prescription independent.

Required background. Borel singularities, lateral sums, and Stokes data supplies the directional ambiguity. The free-field OPE preview supplies the separation into Wilson coefficients and local-operator matrix elements.

Helpful background. Running couplings and dimensional transmutation fixes the one-loop scale relation. Renormalons, OPE ambiguities, and power corrections owns the broader perturbative and effective-field-theory treatment.

Logarithmic momentum moments generate a factorial

Use the one-loop convention

$$\frac{\mathrm d\alpha(\mu^2)}{\mathrm d\log\mu^2} =-\beta_0\alpha^2(\mu^2)+O(\alpha^3), \qquad \beta_0>0,$$

so that

$$\alpha(k^2) =\frac{\alpha(Q^2)} {1+\beta_0\alpha(Q^2)\log(k^2/Q^2)}.$$

A simplified infrared contribution to a dimensionless Euclidean observable is

$$R_p(Q) =\int_0^{Q^2}\frac{\mathrm dk^2}{Q^2} \left(\frac{k^2}{Q^2}\right)^p\alpha(k^2), \qquad p>-1.$$

Writing $x=k^2/Q^2$ and formally expanding the running coupling gives

$$\begin{aligned} R_p(Q) &\sim \alpha(Q^2)\sum_{n=0}^{\infty} \bigl[\beta_0\alpha(Q^2)\bigr]^n \int_0^1\mathrm dx\,x^p(-\log x)^n,\\ \int_0^1\mathrm dx\,x^p(-\log x)^n &=\frac{\Gamma(n+1)}{(p+1)^{n+1}}. \end{aligned}$$

The last identity follows from differentiating $\int_0^1x^p\,\mathrm dx=(p+1)^{-1}$ $n$ times. Thus

$$R_p(Q) \sim \sum_{n=0}^{\infty}r_n\alpha^{n+1}(Q^2), \qquad r_n=\frac{\beta_0^n n!}{(p+1)^{n+1}}.$$

For the Borel transform with respect to $\alpha$,

$$\widehat R_p(\zeta) \equiv\sum_{n=0}^{\infty}\frac{r_n}{n!}\zeta^n =\frac1{p+1-\beta_0\zeta}.$$

The pole lies at

$$\zeta_{\mathrm{IR}}=\frac{p+1}{\beta_0}, \qquad u\equiv\beta_0\zeta=p+1.$$

Because it is on the positive Borel ray, its two lateral sums differ by a term proportional to

$$\exp\!\left[-\frac{p+1}{\beta_0\alpha(Q^2)}\right].$$

With the one-loop definition

$$\frac{\Lambda^2}{Q^2} =\exp\!\left[-\frac1{\beta_0\alpha(Q^2)}\right],$$

the ambiguity scales as

$$\left(\frac{\Lambda^2}{Q^2}\right)^{p+1} =\left(\frac{\Lambda}{Q}\right)^{2(p+1)}.$$

This derivation explains both the factorial and the power, but also exposes the weakness: expanding $\alpha(k^2)$ before integrating extends perturbation theory into the region near its infrared pole. The high-order coefficients are diagnosing sensitivity to a region where that expansion is not uniformly valid.

Matching the ambiguity to the OPE

For a dimensionless short-distance observable,

$$R(Q) =C_{\mathbb 1}(Q,\mu) +\sum_d\frac{C_d(Q,\mu)}{Q^d} \langle O_d(\mu)\rangle.$$

If the coefficient of the identity has an infrared-renormalon ambiguity of order $(\Lambda/Q)^d$, the definition of the dimension-$d$ operator contribution must carry the opposite prescription dependence. Only their sum is physical. This is not a numerical coincidence: changing the separation prescription reallocates contributions between short-distance coefficients and long-distance matrix elements.

The bubble model above predicts

$$d=2(p+1)=2u_{\mathrm{IR}}.$$

For an Adler-function-like weight with $p=1$, the first allowed infrared pole is at $u=2$ and has scale $(\Lambda/Q)^4$, matching a dimension-four operator contribution. In massless QCD there is no gauge-invariant local scalar of dimension two, so a putative $u=1$ ambiguity cannot be matched by the standard OPE and its residue must be absent in that observable. Detailed residues and branch exponents depend on anomalous dimensions, the scheme, and diagrams beyond the one-chain model.

David made the composite-operator ambiguity and its OPE cancellation explicit in controlled large-$N$ two-dimensional models; see David 1984, pp. 237–251. Beneke’s review derives the momentum-region diagnostic and its OPE power matching in Beneke 1999, §§2–3.

What bubble chains establish—and what they do not

In a large-fermion-flavor or related expansion, a chain of vacuum-polarization insertions can be a controlled leading contribution within that expansion. In full QCD, replacing the fermion-chain coefficient by the complete $\beta_0$ is a useful “large-$\beta_0$” model, but it is not a controlled expansion at the physical flavor number. It can reveal:

• a mechanism for $n!$ from momentum logarithms;
• candidate positive- and negative-axis Borel locations;
• the power of the associated prescription dependence; and
• consistency conditions with allowed OPE operators.

It does not prove:

• that the exact Borel transform has a pole rather than a cut at that location;
• that the modeled residue has the right sign or magnitude;
• that no cancellation removes the singularity;
• that every power correction is a renormalon; or
• that a finite-action semiclassical saddle realizes the singularity.

Ultraviolet momentum regions similarly generate alternating factorial growth and negative-axis singularities in this convention. They constrain large order but do not obstruct the positive-ray Borel integral in the same way. The canonical systematic discussion, including scheme dependence and known cancellations, is Beneke 1999, §§2.2–2.4.

Semiclassical proposals need a regime label

On $\mathbb R^4$, no accepted ordinary finite-action saddle universally realizes the leading QCD infrared renormalon. In certain asymptotically free gauge theories compactified on $\mathbb R^3\times S^1$ with specified holonomy and fermion boundary conditions, neutral-bion and correlated-event amplitudes occur in a weakly coupled semiclassical regime. Their Borel scales motivate a proposed continuity to renormalon physics; Argyres and Ünsal 2012 explicitly presents that four-dimensional identification as a conjectural continuation.

The proposal is valuable precisely because its controls are visible. It must not be promoted to a universal proof for infinite-volume QCD. Global form, matter content, circle size, holonomy, and adiabatic continuity are part of the claim.

The Borel and transseries map keeps the bubble diagnostic separate from semiclassical interpretations. The exact and rigorous status comparison supplies the corresponding evidence vocabulary.

Ambiguity matching is not a completed transseries

Finding an $O((\Lambda/Q)^d)$ ambiguity and an allowed dimension-$d$ OPE term is a necessary consistency check. It does not compute the real part of the matrix element, determine all higher powers, or establish a unique transseries completion. A complete result also needs an operator scheme, renormalization scale, lateral prescription, and independently defined nonperturbative matrix elements.

This page therefore stops at the OPE interface. Observable-specific precision phenomenology belongs with the QCD application, while a universal claim about singularity structure would require evidence beyond bubble chains.

Common pitfalls

Treating the Landau pole as the calculation. The factorial comes from logarithmic moments of the formal expansion. The pole signals why exchanging expansion and integration is nonuniform; it is not itself a regulated answer.

Matching only an exponential. The power $d$, anomalous logarithms, quantum numbers, and operator mixing must all agree. A scale resemblance alone does not establish cancellation.

Equating a compactified bion with a four-dimensional renormalon without qualification. The relation is controlled on the small circle and conjectural under continuation unless the required continuity is independently established.

Exercises

1. Evaluate the ultraviolet analogue

$$U_p(Q)=\int_{Q^2}^{\infty}\frac{\mathrm dk^2}{Q^2} \left(\frac{Q^2}{k^2}\right)^{p+2}\alpha(k^2), \qquad p>-1,$$

and show that its logarithmic moments alternate.

Solution

Set $x=Q^2/k^2$, so the measure and weight reduce to a positive constant times $x^p\,\mathrm dx$. Now $\log(k^2/Q^2)=-\log x>0$, and

$$\alpha(k^2) =\alpha(Q^2)\sum_{n\ge0} \bigl[-\beta_0\alpha(Q^2)(-\log x)\bigr]^n.$$

The moment is again $n!/(p+1)^{n+1}$ but carries $(-1)^n$, placing the modeled singularity on the negative Borel axis.

2. A positive-axis singularity at $u=3$ appears in a dimensionless Euclidean observable. What OPE power must be available to cancel its ambiguity in the one-loop convention?

Solution

Since $e^{-u/(\beta_0\alpha)}=(\Lambda^2/Q^2)^u$, the ambiguity is $(\Lambda/Q)^{2u}=(\Lambda/Q)^6$. A dimension-six operator contribution with the same quantum numbers and logarithmic running is therefore required. Its existence is necessary for matching but does not determine the residue.

References

• Argyres, Philip C., and Mithat Ünsal. “A Semiclassical Realization of Infrared Renormalons.” Physical Review Letters 109 (2012): 121601. arXiv:1204.1661; doi:10.1103/PhysRevLett.109.121601.
• Beneke, Martin. “Renormalons.” Physics Reports 317 (1999): 1–142. arXiv:hep-ph/9807443; doi:10.1016/S0370-1573(98)00130-6.
• David, François. “On the Ambiguity of Composite Operators, IR Renormalons and the Status of the Operator Product Expansion.” Nuclear Physics B 234 (1984): 237–251. doi:10.1016/0550-3213(84)90235-9.