b-jet tag matching dr parameter?

Asked by Billy Ford on 2019-03-12

Dear all,

I've been working on some events with 4 b's in the final state, looking to explore the boosted merging in b-jet multiplicity distributions.

I recently came across the parameter 'main.fastsim.bjet_id.matching_dr', which as I understand sets a cone around a B-hadron and looks for a jet, if one is found it tags the jet as a b and moves on (with the bjet_id.exclusive parameter set to true). Just wondering if there is any information regarding what values one should use for this?

I've performed a scan on my events over values between 0.4 and 2.0, finding that the b-jet multiplicity changes, in particular for my sample one finds as the matching_dr increases, the proportion of events with 4 b-jets increases, up to a plateau around dr= 1.6. To note the tagging efficiency is set to 1.0 and mistag is 0.0.

This suggests to me that for lower dr's, there are events where the jet from the b lies outside of the dr cone, and are therefore (incorrectly) not tagged as b's. It is peculiar as the default option is 0.4, which in my case seems to 'miss' a significant amount of jets? Alternatively I wonder if I have simply stumbled upon a parameter that isn't really supposed to be tampered with!

Would appreciate any insight into the workings of this parameter and/or suggestions as to which values to use (maybe the same as the clustering algorithm radius?)

Thanks,
Billy

Question information

Language:
English Edit question
Status:
Solved
For:
MadAnalysis 5 Edit question
Assignee:
No assignee Edit question
Solved by:
Benjamin Fuks
Solved:
2019-03-13
Last query:
2019-03-13
Last reply:
2019-03-12
Best Benjamin Fuks (fuks) said : #1

Hi Billy,

The matching_dr parameter is how close the B-hadron should be from the jet direction. If it is too small with respect to the radius parameter used for the clustering, you will have b-jets that won't be tagged as b-jets. I have never really investigated which value to use here. Usually, I am very naive: if my radius parameter is R, I require the matching_dr parameter to be R too. In other words, it is sufficient that the reconstructed jet contains a B-hadron to be tagged as a b-jet.

Cheers,

Benjamin

> On 12 Mar 2019, at 13:22 , Billy Ford <email address hidden> wrote:
>
> New question #679140 on MadAnalysis 5:
> https://answers.launchpad.net/madanalysis5/+question/679140
>
> Dear all,
>
> I've been working on some events with 4 b's in the final state, looking to explore the boosted merging in b-jet multiplicity distributions.
>
> I recently came across the parameter 'main.fastsim.bjet_id.matching_dr', which as I understand sets a cone around a B-hadron and looks for a jet, if one is found it tags the jet as a b and moves on (with the bjet_id.exclusive parameter set to true). Just wondering if there is any information regarding what values one should use for this?
>
> I've performed a scan on my events over values between 0.4 and 2.0, finding that the b-jet multiplicity changes, in particular for my sample one finds as the matching_dr increases, the proportion of events with 4 b-jets increases, up to a plateau around dr= 1.6. To note the tagging efficiency is set to 1.0 and mistag is 0.0.
>
> This suggests to me that for lower dr's, there are events where the jet from the b lies outside of the dr cone, and are therefore (incorrectly) not tagged as b's. It is peculiar as the default option is 0.4, which in my case seems to 'miss' a significant amount of jets? Alternatively I wonder if I have simply stumbled upon a parameter that isn't really supposed to be tampered with!
>
> Would appreciate any insight into the workings of this parameter and/or suggestions as to which values to use (maybe the same as the clustering algorithm radius?)
>
> Thanks,
> Billy
>
> --
> You received this question notification because you are an answer
> contact for MadAnalysis 5.

Billy Ford (billyford) said : #2

Thanks for your insight Ben! I'll keep in mind to set R and dr to the same value in future, but I do wonder what exactly is going on when scanning over values.

Billy Ford (billyford) said : #3

Thanks Benjamin Fuks, that solved my question.