In the paper "Scattering into the Fifth Dimension of $\mathcal{N}=4$ Super Yang-Mills", the authors give the following result for an integral:

$$\begin{align} I^{(1)}(x_{13}^2,x_{24}^2,m) =& \left( x_{13}^2 +m_{13}^2 \right) \left( x_{24}^2 + m_{24}^2 \right)\cr & \times\int d^4 x_5 \frac{1}{(x_{15}^2+m^2) (x_{25}^2+m^2) (x_{35}^2+m^2) (x_{45}^2+m^2)} \notag \\ =& 2 \ln \left( \frac{m^2}{x_{13}^2} \right) \ln \left( \frac{m^2}{x_{24}^2} \right) - \pi^2 + \mathcal{O}(m^2) \tag{1} \end{align}$$

where $x_{ij} := x_i - x_j$, $m_{ij} := m_i - m_j$. I would like to take the limit $m \to 0$ of this integral and compare it to a computation done with dimensional regularization. The following relation between cutoff and dim reg is given in the paper "Cutoff Regularization Method in Gauge Theories" :

$$\ln \Lambda^2 = \frac{1}{\epsilon} - \gamma + \ln(4\pi^2 \mu^2) + 1 \tag{2}$$

Can I use this relation for translating the "mass cutoff" of eq.(1)?


1 Answer 1


A few things:

$\bullet$ Your definition of the $I^{(1)}$ integral, and then the expression involving $m$ that follows, is missing information. The paper you quote says that the latter expression holds only in the case where all the masses are the same ($m_i = m$), and also only in the limit where $m$ is small compared to $\hat{x}_{13}^2$ and $\hat{x}^2_{24}$

$\bullet$ As I think you are aware, it is in this sense that the latter expression already has the limit $m \to 0$ taken. That is what the $\mathcal{O}(m^2)$ means.

$\bullet$ In general, I don't think there is a consistent mapping between different regularization schemes: ie. Although it is often true that a $1/\epsilon$ in dim reg corresponds to a $\log(\Lambda)$ in a cutoff scheme, I think this is only true for simple loop integrals and I don't think you can always trust this.

$\bullet$ In the second paper, your relation between cutoff $\Lambda$ and dim-reg $\epsilon$ seems to be for an ultraviolet cutoff $\Lambda$ (which is supposed to be large compared to the scales in your integral). The mass $m$ in your loop integral parametrizes an infrared divergence (in that $m$ is small compared to the other scales in your integral). This further undermines using your last expression. I would definitely be wary of using it.

  • $\begingroup$ Mm okay, yeah that makes sense. I will try to think of another way to proceed. Thanks for the clarification! $\endgroup$
    – Pxx
    Nov 2, 2019 at 0:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.