IR singularities with squared amplitudes produced by MadLoop

Asked by Xiang Lv on 2020-06-15


I’m trying to use MadLoop to generate a standalone library for some processes with EW loop corrections, and have a few questions:
1. I’m confused by the IR poles in the results. For instance, for the simple QED process e- e+ -> mu- mu+, I did

define exc = g0 g+ g- h gha ghz ghwp ghwm z u c d e- b s mu- ta+ ta- w+
generate e- e+ > mu- mu+ QCD=0 [virt=QED] /exc

which produced a loop graph with a top quark loop on the photon propagator. When I did

launch -f

I got the result

|| Total(*) Born contribution (GeV^0):

| Born = 1.1012911820047121e-02

|| Total(*) virtual contribution normalized with born*alpha_S/(2*pi):

| Finite = -2.1842211753457780e-01

| Single pole = -1.1685343497245081e-01

| Double pole = 0.0000000000000000e+00

As far as I know, this loop graph should not have an IR singularity. Do you know the reason for the non-zero single pole?

2. Can I use the subtraction method in MadFKS to compute photon radiative corrections? It seems that only QCD NLO radiative corrections are available.

3. When I run MadLoop in standalone mode, how can I set a dynamical renormalization scale? The process I’m considering is again EW loop graphs. I don’t find the setscal.f file in my output.

Xiang Lv

Question information

English Edit question
MadGraph5_aMC@NLO Edit question
Hua-Sheng Shao Edit question
Solved by:
Hua-Sheng Shao
Last query:
Last reply:


I will let Hua-Sheng to answer the first question.

Concerning #2: NLO correction in EW is possible only at fixed order within the version 3 of the code which is still beta.
Note that your process definition e- e+ > mu- mu+ will likely not be compatible with that framework (see

Concerning #3:
In that mode, they are no running implemented, you have to decide yourself the scale that you use for the various running/...



Best Hua-Sheng Shao (erdissshaw) said : #2


For the first question, MadLoop not only includes the one-loop diagrams but also the renormalisation (UV) counter terms. In your case, I think the pole is from the UV counter terms not from loop diagrams. In the current implementation of the NLO EW models, there is no diagram-by-digram correspondence between loops and UVs.

For the second question, see the reply of Olivier.

For the last one, please refer to for the instructions of using MadLoop as a standalone library.



Xiang Lv (xiang123) said : #3

Thanks Hua-Sheng Shao, that solved my question.

Xiang Lv (xiang123) said : #4

Tanks for your helping.