# Scale uncertainty in a a > X processes

Dear all,

I am wondering about the scale uncertainties in processes like a a > ll or a a > W W.

I see large scale variations (order ~10%) as well as large PDF uncertainties (150% for CT14qed but it seems to be independent from the beam, i.e. is the same for using the EPA approximation with photon beams) for the total cross section and I would like to investigate/check that everything is properly implemented and what is taken as actual scale uncertainties here (presumably this comes from the factorization scale only?

I would like to understand, if these large numbers are to be trusted?

Thanks & Best

Kristin

## Question information

- Language:
- English Edit question

- Status:
- Answered

- Assignee:
- No assignee Edit question

- Last query:
- 2020-07-22

- Last reply:
- 2020-07-29

Just to add: Looking more into this, I do see that *only* the variation of the muF gives an different event weight (looking at a random event).

I however *also* notice that it seems that the PDF uncertainties is evaluated around a scale of muF=2.0 -- is this correct/intended?

This is for a sample with 1 resolved and 1 EPA photon beam.

Hi,

1. With photon component, you need first to check what is included in the PDF set and what is not.

Some PDF include the EPA contributions inside the PDF and some others do not.

I do not know in the specific case of your particular PDF.

2. The EPA is associated to his own scale and we use muf for that one as well. In the latest version of the code we use the fix-scale value for that scale such that you can use the dynamical scale consistently for the PDF beam if you want

3. If you use dynamical scale and have very light particle, the scale selected can be quite light, for some scale choice (in particular our default one) we have a cut-off vetoing the event if the scale selected is smaller that 4GeV^2, so you are close to the limit

4. automatic scale computation, will not handle the variation of the cut of scale in the EPA case. (not sure if the code crash or not if you try).

Cheers,

Olivier

## Can you help with this problem?

Provide an answer of your own, or ask Kristin Lohwasser for more information if necessary.