automatic ptj and mjj cuts from xqcut are also applied to jets that are not relevant for the matching

Hi Rikkert,

Let me be sure of your problem:
let's assume that you have
e+ e- > w+ j1, w+ > j2 j3

and that you have in your run_card:
cut_decays = T
auto_ptj_mjj = T
ptj= 10
xqcut= 20

Then you were hoping to have the following cuts:
ptj1 : 20
ptj2: 10
ptj3: 10
while you have (This is at least what I think should happen)
ptj1: 20
ptj2:20
ptj3:20

Is it the only scenario of combination of parameters where you have trouble/different expectation?:
cut_decays = F
auto_ptj_mjj = T
ptj= 10
xqcut= 20
you should have
ptj1 : 20
ptj2: 0
ptj3: 0
I don't see any possible confusion on here.

>Moreover, it's not so easy to spot that this is going wrong, because all the matching plots are rather smooth. Only the cross section is a bit lower than expected.

Yes this is expected since ptj2 and ptj3 didn't enter to the matching. So the smootness of the matching shouldn't be impacted by this effect. In fact you can spot them via the log file of each single run where you can see the pt cut for each particle.

If I understand your problem correctly this is not a bug (in the sense that the code produce the output that I expect). So I have tagged this in the wishlist. If you agree with this classification, I will not fix it right away and start a discussion with the CMS team to know what they think about this behavior before changing it. (This might creates trouble for them if we change it without noticing them at least)

Cheers,

Olivier

Hi Olivier,

What I did was p p > t t~, t > b j j, t~ > b~ j j. I removed all the cuts in the run_card (apart from the xqcut), so I had expected that it would be irrelevant how the cut_decays was set. To my surprise it did make a difference, because the auto_ptj_mjj flag was set to true. So it's equivalent to your first scenario above.

This is certainly a very strange behavior that does not make sense at all: There is one jet merged, but the jets from the top decay products have a generation cut that was not specifically specified in the run_card. "Experimentally" there is no difference between the jets, so this is a rather strange situation. Therefore I would consider it a bug instead of a missing feature.

Cheers,
Rik

Hi Rik,

I found in the documentation that such case is explicitly stated that auto_ptj_mjj should be set to F.

> Note: For most processes, the generation speed can be improcved by setting ptj and mjj to xqcut, which is done automatically if the flag auto_ptj_mjj is set to T. If some jets should not be restricted this way (as in single top or vector boson fusion (VBF) production, where some jets are not radiated from QCD), auto_ptj_mjj should be set to F.

So this is adding a new feature, which is indeed to put as a whishlist. But the good news is that we are fully allowed to change this behavior (I will inform CMS anyway).
But I don't see clearly how to do it, so it will take me a bit of time to change this behavior.

Cheers,

Olivier

On Oct 23, 2013, at 1:09 PM, Rikkert Frederix <email address hidden> wrote:

> Hi Olivier,
>
> What I did was p p > t t~, t > b j j, t~ > b~ j j. I removed all the
> cuts in the run_card (apart from the xqcut), so I had expected that it
> would be irrelevant how the cut_decays was set. To my surprise it did
> make a difference, because the auto_ptj_mjj flag was set to true. So
> it's equivalent to your first scenario above.
>
> This is certainly a very strange behavior that does not make sense at
> all: There is one jet merged, but the jets from the top decay products
> have a generation cut that was not specifically specified in the
> run_card. "Experimentally" there is no difference between the jets, so
> this is a rather strange situation. Therefore I would consider it a bug
> instead of a missing feature.
>
> Cheers,
> Rik
>
> --
> You received this bug notification because you are a member of MadTeam,
> which is subscribed to MadGraph5.
> https://bugs.launchpad.net/bugs/1243543
>
> Title:
> automatic ptj and mjj cuts from xqcut are also applied to jets that
> are not relevant for the matching
>
> Status in The MadGraph Matrix Element Generator version 5:
> Triaged
>
> Bug description:
> Hi Olivier,
>
> When doing matching, for example for ttbar production, including hadronic decays, the auto_ptj_mjj flag and the cut_decays flag are not working as expected. I would have expected that, no matter what I put for the cut_decays flag, the ptj and mjj cuts that are set automatically from the xqcut with the auto_ptj_mjj are always only applied to the jets that enter the matching, and certainly not to the top quark decay products.
> Moreover, it's not so easy to spot that this is going wrong, because all the matching plots are rather smooth. Only the cross section is a bit lower than expected.
>
> Cheers,
> Rikkert
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/madgraph5/+bug/1243543/+subscriptions

Dear Olivier,

I think in the auto_ptj_mjj documentation refers to directly-produced jets (e.g. the forward jets in WBF) and not decay products. I would expect any cuts on the latter to be solely controlled by the cut_decays flag. In this case, I would agree with Rikkert that this is a bug. Moreover MG/ME gives (me at least) a wildly incorrect cross section for the process. This is exactly the problem I was describing in my bug report #1311186.

Cheers,

Rakhi

Dear Rakhi,

This is a feature which works as describe in the manual. So this is not my definition of bug.
But the definition of a feature which doesn’t work as some people expects.
So this is indeed a feature request and not a bug.

Cheers,

Olivier

On Apr 30, 2014, at 1:06 AM, Rakhi Mahbubani <email address hidden> wrote:

> Dear Olivier,
>
> I think in the auto_ptj_mjj documentation refers to directly-produced
> jets (e.g. the forward jets in WBF) and not decay products. I would
> expect any cuts on the latter to be solely controlled by the cut_decays
> flag. In this case, I would agree with Rikkert that this is a bug.
> Moreover MG/ME gives (me at least) a wildly incorrect cross section for
> the process. This is exactly the problem I was describing in my bug
> report #1311186.
>
> Cheers,
>
> Rakhi
>
> --
> You received this bug notification because you are a bug assignee.
> https://bugs.launchpad.net/bugs/1243543
>
> Title:
> automatic ptj and mjj cuts from xqcut are also applied to jets that
> are not relevant for the matching
>
> Status in MadGraph5_aMC@NLO Generator:
> Triaged
>
> Bug description:
> Hi Olivier,
>
> When doing matching, for example for ttbar production, including hadronic decays, the auto_ptj_mjj flag and the cut_decays flag are not working as expected. I would have expected that, no matter what I put for the cut_decays flag, the ptj and mjj cuts that are set automatically from the xqcut with the auto_ptj_mjj are always only applied to the jets that enter the matching, and certainly not to the top quark decay products.
> Moreover, it's not so easy to spot that this is going wrong, because all the matching plots are rather smooth. Only the cross section is a bit lower than expected.
>
> Cheers,
> Rikkert
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/mg5amcnlo/+bug/1243543/+subscriptions