MadAnalysis 5

DJR plots

Asked by HH

Dear developers,

Q1) After generation of a matched sample, DJR plots are generated automatically. These plots show DJR distributions corresponding to some QCUT values. However these values are different from the QCUT value I choose in pythia8 card. How can I have the plot relevant for my chosen QCUT?

Q2) The DJR plots corresponding to 0-jet, 1-jet, ... for a generated matched sample (pp>tt~+0,1,2 jets) are more or less similar covering a wide range and the QCUT line (transition line) cannot be recognized easily. What is the problem? I expected a plot in which different multiplicities don't overlap much. Is this a true expectation? Below you can find the relevant modifications I made to the cards:

set ickkw 1
set xqcut 20
set drjj 0
set etaj 5
set auto_ptj_mjj T
set JetMatching:qcut 30

I regenerated the same matched sample using showerkT=ON too (xqcut=20, qcut=80 this time). But again I got the same results.

Question information

Language:
English Edit question
Status:
Solved
For:
MadAnalysis 5 Edit question
Assignee:
No assignee Edit question
Solved by:
Jack Y. Araz
Solved:
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Hi,

You are right that the DJR returned by Pythia8 is not plotting the same variable on which the cut occurs, this is only a proxy.
The py8 author claims that this is the best that they can do (I doubt about that but ok).

For the first question, do you have the following parameter:
SysCalc:fullCutVariation set on False
SysCalc:qCutList set on [30]
SysCalc:qWeed 30

If you do, then I will assign this question to Valentin which should know better than me.

Cheers,

Olivier

Revision history for this message
HH (hh-madgrapher) said : 		#2

Hi

Yes, I have these variables in the cards. As I see, SysCalc:qCutList is the list of qCuts for which the DJR plots are generated by Pythia. Modifying this variable may solve the problem. However, I preferred to use MadAnalysis5 to generate match report. It is much better indeed.

However, my main issue is the second question. I saw some DJR plots in some papers which are different from mine. In these plots, curves corresponding to different multiplicities cover different ranges, while in my report, different multiplicities more or less overlap ! However the summed curve is smooth. So, is there anything wrong with my DJR test?

Revision history for this message
HH (hh-madgrapher) said : 		#3

I think I did something wrong. At the end of the event generation, Pythia warns that the event file should be used with its weights as the events are not vetoed because of the systematic computations. Moreover, while running MadAnalysis5 I get an enormous number of the message "WARNING: Found one event with a zero weight. Skipping...".

I think I should use weights. Below is the commands I used in MadAnalysis5:

ma5>import wjets13/Events/run_01_0/tag_1_pythia8_events.hepmc
ma5>set main.merging.check = true
ma5>set main.merging.njets = 2
ma5>submit check

Do you think I should change the commands for the MadAnalysis5 to take into account the weights?

Revision history for this message
Jack Y. Araz (jackaraz) said : 		#4

Hi

Your ma5 commands looks correct, the reason that it gives you warning is because the events which are removed due to the QCUT that you are implementing, are kept in the hepmc file but the weight set to zero. So ma5 is skipping those events while preparing DJR plots. Thus you dont need to worry about those warnings.

Cheers
Jack

Revision history for this message
HH (hh-madgrapher) said : 		#5

Dear Jack,

Thanks a lot for your clarification. Yes I checked the output hepmc file (from Pythia). Weights of the vetoed events are set to 0 and the MadAnalysis5 skip those events. Moreover I found out that turning the systematic computations off, Pythia removes vetoed events. This might be good for saving disk space.

However I am still suspicious to the results as in the DJR plot, the 0-jet and 1-jet ... contributions overlap in a wide range! The summed curve is smooth though.

Revision history for this message
Best Jack Y. Araz (jackaraz) said : 		#6

Hi

Although I have limited expertiese on jet merging I believe the shape of the 0,1,2 jet samples should more or less overlap and the important curve that you should look for is combined one which is represented with the black curve in Ma5. This curve should more or less look like upside down U but if you are applying any type of hard cut on jets (other than qcut) this can shift to an upsidedown V shape. Also note that since merging process removes some of the events, lack of statistics might lead unexpected shapes as well.

Cheers
Jack

Revision history for this message
HH (hh-madgrapher) said : 		#7

Thanks Jack Y. Araz, that solved my question.