I've done several calculations on one-loop diagrams in dimensional regularization, involving things like Feynman parameters, or using hyperspherical coordinates after a Wick rotation, s.t. you can drop terms like $p^\mu a_\mu$ in numerators, or replacing $p^\mu p^\nu$ with $\eta^{\mu\nu}/d$.

However, when it comes to two-loop calculations, I'm stumped. I would like to understand how to do a simple integral like this one: $$ \int\mathrm{d}^dp\int\mathrm{d}^dk \frac{1}{p^2-m^2}\frac{1}{k^2-m^2}\frac{c_1 p^2 + c_2k^2+c_3p\cdot k}{(k-p)^2-m^2} $$ so that I can apply similar steps to my own calculations. $d=4-\epsilon$.

In particular:

  1. How do I use Feynman parameters in this case? edit: I know about the $1/(ABC)=...$ formula. The question should read: how do I complete the squares after employing the Feynman parameters $x,y,z$?
  2. How do I deal with terms like $p\cdot k$?

If I use the Feynman parameters, is this the correct ansatz?

$$ \frac{1}{p^2-m^2}\frac{1}{k^2-m^2}\frac{1}{(k-p)^2-m²} = \int_0^1 \delta(x+y+z-1)\frac{2}{[ a_1(p+a_2)² + a_3(k+a_4)² -a_5^2 ]^3} $$

  • 2
    $\begingroup$ Did you try the Feynman parameter $\frac{1}{A B C} = \int^1_0 dx dy dz \delta (x + y + z - 1) \frac{2}{(x A + y B + z C)^3}$? $\endgroup$ Oct 28, 2020 at 16:47
  • $\begingroup$ @MicheleGrosso After rewriting the expression using Feynman parameters, how do I complete the squares in order to go to spherical coordinates? $\endgroup$
    – ersbygre1
    Nov 2, 2020 at 1:47

1 Answer 1


The answer is actually quite simple: one performs one loop integral after the other, so:

  1. Use Feynman parameters
  2. Complete the square for one (!) loop momentum, e.g. $a_1(p-a_2)^2-a_3^2$
  3. Wick rotation, $p^0=\text{i}k^0_E$, $k^2=-k_E^2$
  4. Dimensional regularization: $d^4p_E \to \lim\limits_{\epsilon\to 0^+}d^dp_E$ with $d=4-\epsilon$ (or another regularization)
  5. Use generalized spherical coordinates, s.t. $\int d\Omega$ is trivial
  6. Perform radial integration by using Beta and Gamma functions

...then do the same steps for the next loop momentum.

Finally, the integration over the Feynman parameters remains.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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