Offshell tops in LHE file

Asked by Jack Y. Araz on 2021-03-17

Dear Olivier

I hope you are well and safe! I would like to ask if there is a possible way to include off-shell particles in the LHE file as well? I'm generating the following event

generate p p > t t~ > b j j b~ l nu / a h z

with a certain FS mass cut to capture boosted top regime (since it's not possible to access tops directly from cuts.f I'm just setting a cut to the vectorially combined mass of final state particles). As expected, I do not see off-shell tops written in the LHE file nor the HEPMC file. But I need to have access to those particles somehow. Is there any way to modify the code to force it to print off-shell particles to the LHE file as well? Maybe by changing bwcutoff but can't be sure if it might ruin anything.

Thanks
Best regards
Jack

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Jack Y. Araz
Solved:
2021-03-19
Last query:
2021-03-19
Last reply:
2021-03-19

Hi,

For off-shell effects you need to be very carefull.
First your syntax is strongly discouraged because it has issue.
If I run the check command on your syntax, you will see that you will have issue with gauge and lorentz invariance

> Lorentz invariance results:
> Process Min element Max element Relative diff. Result
> g g > b d u~ b~ e6.6267301504e-23 1.1634825757e-22 4.3044010381e-01 Failed
> JAMP 0 8.7133633089e-22 1.5603827271e-21 4.4158806955e-01 Failed
> JAMP 1 1.7500977666e-22 4.1601474180e-22 5.7931832920e-01 Failed
> g g > b d u~ b~ m3.1200862865e-21 5.4347993450e-21 4.2590589120e-01 Failed
> JAMP 0 4.9739295972e-20 8.6830515571e-20 4.2716802215e-01 Failed
> JAMP 1 2.1251056905e-21 4.2654206779e-21 5.0178286014e-01 Failed
> g g > b s c~ b~ e5.5296475147e-25 1.0893625075e-24 4.9239601357e-01 Failed
> JAMP 0 7.4546164112e-24 1.5038613283e-23 5.0430160873e-01 Failed
> JAMP 1 2.3090882282e-24 4.3314808380e-24 4.6690558852e-01 Failed
> g g > b s c~ b~ m3.3011513618e-23 1.1633953048e-22 7.1624852290e-01 Failed
> JAMP 0 4.1831618998e-22 1.5888335768e-21 7.3671491081e-01 Failed
> JAMP 1 1.5721543573e-22 4.6285144775e-22 6.6033284222e-01 Failed
> Summary: 0/4 passed, 4/4 failed
> Failed processes: g g > b d u~ b~ e+ ve, g g > b d u~ b~ mu+ vm, g g > b s c~ b~ e+ ve, g g > b s c~ b~ mu+ vm
> Gauge results:
> Process matrix BRS ratio Result
> g g > b d u~ b~ e+ ve 2.1845774448e-23 1.2973571888e-23 5.9387099868e-01 Failed
> JAMP 0 3.1223630554e-22 1.5447217393e-22 4.9472841944e-01 Failed
> JAMP 1 7.4296211246e-23 7.7540516906e-23 1.0436671750e+00 Failed
> Summary: 0/1 passed, 1/1 failed
> Failed processes: g g > b d u~ b~ e+ ve
> Gauge results (switching between Unitary/Feynman/axial gauge):
> Process Unitary Feynman Relative diff. Result
> g g > b d u~ b~ e+ ve 1.6300794704e-23 1.6300794704e-23 4.1464803649e-15 Passed
> g g > b d u~ b~ mu+ vm 4.0637693677e-22 4.0637693677e-22 4.6281931653e-16 Passed
> g g > b s c~ b~ e+ ve 2.8382429648e-21 2.8382429648e-21 1.3253206189e-15 Passed
> g g > b s c~ b~ mu+ vm 3.3735649947e-19 3.3735649947e-19 5.4234421155e-15 Passed
> Summary: 4/4 passed, 0/4 failed
> Process permutation results:
> Process Min element Max element Relative diff. Result
> g g > b d u~ b~ e+ ve 8.2015398357e-21 8.2015398357e-21 1.8345735060e-16 Passed
> Summary: 1/1 passed, 0/1 failed

If you use
> check p p > b j j b~ l nu
> Lorentz invariance results:
> Process Min element Max element Relative diff. Result
> g g > b d u~ b~ e2.1676410978e-18 2.1676410978e-18 3.5895042695e-14 Passed
> g g > b d u~ b~ m4.5078632244e-19 4.5078632244e-19 2.9906651865e-15 Passed
> g g > b s c~ b~ e1.9334775228e-18 1.9334775228e-18 1.0757840810e-14 Passed
> g g > b s c~ b~ m8.9088751643e-19 8.9088751643e-19 7.7825176264e-15 Passed
> u u > b u d b~ e+1.6489362508e-19 1.6489362508e-19 6.2779100442e-15 Passed
> u u > b u d b~ mu8.4360479008e-18 8.4360479008e-18 1.8263812316e-14 Passed
> u c > b u s b~ e+1.3467443021e-20 1.3467443021e-20 1.3183398398e-14 Passed
> u c > b u s b~ mu9.8811680577e-19 9.8811680577e-19 1.9490913758e-14 Passed
> u c > b c d b~ e+2.8089375650e-20 2.8089375650e-20 2.3568997284e-15 Passed
> u c > b c d b~ mu4.2080711048e-18 4.2080711048e-18 3.3867967612e-14 Passed
> u d > b d d b~ e+6.3289304396e-20 6.3289304396e-20 1.5215287660e-15 Passed
> u d > b d d b~ mu1.2289323287e-17 1.2289323287e-17 1.9056629545e-14 Passed
> u s > b d s b~ e+6.7187954042e-19 6.7187954042e-19 1.9635394926e-14 Passed
> u s > b d s b~ mu3.1568246287e-18 3.1568246287e-18 1.1469589608e-14 Passed
> c c > b c s b~ e+2.1645764453e-17 2.1645764453e-17 4.0430199110e-14 Passed
> c c > b c s b~ mu1.2116655823e-16 1.2116655823e-16 7.0598773836e-14 Passed
> c d > b d s b~ e+5.2375237743e-20 5.2375237743e-20 1.4019235481e-14 Passed
> c d > b d s b~ mu1.7032788270e-19 1.7032788270e-19 2.6995919102e-14 Passed
> c s > b s s b~ e+1.0616179684e-16 1.0616179684e-16 3.1348442123e-15 Passed
> c s > b s s b~ mu5.9170091679e-18 5.9170091679e-18 1.3540402493e-14 Passed
> Summary: 20/20 passed, 0/20 failed
> Gauge results:
> Process matrix BRS ratio Result
> g g > b d u~ b~ e+ ve 3.8686168009e-17 1.5741498213e-48 4.0690249314e-32 Passed
> Summary: 1/1 passed, 0/1 failed
> Gauge results (switching between Unitary/Feynman/axial gauge):
> Process Unitary Feynman Relative diff. Result
> g g > b d u~ b~ e+ ve 1.6981519064e-17 1.6981519064e-17 3.6292252765e-16 Passed
> g g > b d u~ b~ mu+ vm 1.6335078523e-19 1.6335078523e-19 5.8950740324e-16 Passed
> g g > b s c~ b~ e+ ve 3.4335439485e-16 3.4335439485e-16 1.2923506029e-15 Passed
> g g > b s c~ b~ mu+ vm 2.1254446896e-18 2.1254446896e-18 9.0613035184e-16 Passed
> u u > b u d b~ e+ ve 6.6666881455e-18 6.6666881455e-18 0.0000000000e+00 Passed
> u u > b u d b~ mu+ vm 9.3388183540e-19 9.3388183540e-19 4.1245688081e-16 Passed
> u c > b u s b~ e+ ve 1.6216392263e-20 1.6216392263e-20 3.7113872055e-16 Passed
> u c > b u s b~ mu+ vm 1.1802216709e-18 1.1802216709e-18 1.6318374691e-16 Passed
> u c > b c d b~ e+ ve 5.0073447968e-18 5.0073447968e-18 4.6154519552e-16 Passed
> u c > b c d b~ mu+ vm 1.1366699350e-18 1.1366699350e-18 1.3554893184e-15 Passed
> u d > b d d b~ e+ ve 9.3891267407e-19 9.3891267407e-19 1.2307406200e-15 Passed
> u d > b d d b~ mu+ vm 6.3123426153e-21 6.3123426153e-21 2.8603633119e-15 Passed
> u s > b d s b~ e+ ve 1.9111184262e-17 1.9111184262e-17 3.2248005867e-16 Passed
> u s > b d s b~ mu+ vm 3.4338414945e-19 3.4338414945e-19 2.8043372815e-16 Passed
> c c > b c s b~ e+ ve 8.4409672550e-19 8.4409672550e-19 0.0000000000e+00 Passed
> c c > b c s b~ mu+ vm 4.4410407500e-17 4.4410407500e-17 6.9386616437e-16 Passed
> c d > b d s b~ e+ ve 9.5671407941e-20 9.5671407941e-20 0.0000000000e+00 Passed
> c d > b d s b~ mu+ vm 6.2553171837e-19 6.2553171837e-19 4.6183028482e-16 Passed
> c s > b s s b~ e+ ve 1.3139395645e-18 1.3139395645e-18 1.4657675257e-16 Passed
> c s > b s s b~ mu+ vm 4.8030147456e-16 4.8030147456e-16 4.1060716394e-16 Passed
> Summary: 20/20 passed, 0/20 failed
> Process permutation results:
> Process Min element Max element Relative diff. Result
> g g > b d u~ b~ e+ ve 6.7772900192e-19 6.7772900192e-19 7.1043512191e-16 Passed
> u u > b u d b~ e+ ve 1.4733145600e-20 1.4733145600e-20 6.1275418431e-16 Passed
> u c > b u s b~ e+ ve 3.3403270821e-19 3.3403270821e-19 4.3242695185e-16 Passed
> u c > b c d b~ e+ ve 6.6471377946e-19 6.6471377946e-19 4.3460734617e-16 Passed
> u d > b d d b~ e+ ve 9.3893538737e-19 9.3893538737e-19 1.4358293225e-15 Passed
> u s > b d s b~ e+ ve 6.2225076896e-20 6.2225076896e-20 5.8033172892e-16 Passed
> c d > b d s b~ e+ ve 6.0154438401e-19 6.0154438401e-19 2.0810674942e-15 Passed
> Summary: 7/7 passed, 0/7 failed

as you can see the situation is much better. Actually it is not yet perfect since those check are done with 0 width for the top/W (which might be ok in your case for the top? but likely not for the W)

> with a certain FS mass cut to capture boosted top regime (since it's not possible to access tops directly from cuts.f I'm just setting a cut to the vectorially combined mass of final state particles). As expected, I do not see off-shell tops written in the LHE file nor the HEPMC file. But I need to have access to those particles somehow. Is there any way to modify the code to force it to print off-shell particles to the LHE file as well? Maybe by changing bwcutoff but can't be sure if it might ruin anything.

This makes me feel uncomfortable, if your top is not onshell then what is the meaning to call it a top and more importantly what is the meaning to remove any other diagram that interferes with the one of the top? (which goes back to the first question). In top of that in that regime you have broken gauge invariance (by selecting a wrong subset of diagram).
At that level, you also start to care about the impact of the finite width of the top/W, so I would suggest to use the complex-mass scheme (as stated the above check was setting all widths to zero to avoid the breaking of gauge invariance associated to the width, using the complex-mass scheme will avoid those gauge invariant issue linked to the width).

Now to answer your question, yes increasing the bwcutoff should do the trick. However this can impact the phase-space integrator which is not designed for large value of that parameter (so the generation of events has a risk to be bias).
Second the parton-shower will not be identical if the events has/has not an top written in the lhef file.
In theory the invariant mass of any particle written in the lhef file needs to be preserved by the parton-shower, now some PS sometimes takes freedom to not do that and to internally remove again that line and not to preserve the invariant mass/ consider it for the parton-shower.

Cheers,

Olivier

> On 17 Mar 2021, at 14:01, Jack Y. Araz <email address hidden> wrote:
>
> New question #696110 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/696110
>
> Dear Olivier
>
> I hope you are well and safe! I would like to ask if there is a possible way to include off-shell particles in the LHE file as well? I'm generating the following event
>
> generate p p > t t~ > b j j b~ l nu / a h z
>
> with a certain FS mass cut to capture boosted top regime (since it's not possible to access tops directly from cuts.f I'm just setting a cut to the vectorially combined mass of final state particles). As expected, I do not see off-shell tops written in the LHE file nor the HEPMC file. But I need to have access to those particles somehow. Is there any way to modify the code to force it to print off-shell particles to the LHE file as well? Maybe by changing bwcutoff but can't be sure if it might ruin anything.
>
> Thanks
> Best regards
> Jack
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Jack Y. Araz (jackaraz) said : #2

Dear Olivier

Thanks a lot for the detailed answer, it was very useful! Now that you said it might create a problem to increase bwcut, excluding off-sell tops would be much more useful for the physics that I'm interested in. This, I believe, can be done by selecting the decay products of top and requiring each of the group to satisfy Mt +- X condition through cuts.f. However, I'm not sure if it will be accurate if I do it solely based on phase-space, is there any way to access the particles' pdgID so that I can match them with respect to their charge? This way, I believe I can fix all the problems that you mentioned above.

I believe its also possible to define the process as

p p > t t~, t > b l nu, t~ > j j b~

but in this notation, I'm not able to set NP^2==2 to isolate the interference term for both production and decay vertices.

Thank you so very much!
Best regards
Jack

Hi,

> is there any way to access the particles' pdgID

We do have pdgID information for final state particles but thise pdgID can (and are) change later in the code.
Now if you care only the electric charge of the associated particle you are fine.

The cleanest would be to split the variable IS_A_LEPTON to IS_A_LEPTON_PLUS and IS_A_LEPTON_MINUS
(and same for is_a_jet in three category) and define them accordingly in setcuts.f

> I believe, can be done by selecting the decay products of top and
> requiring each of the group to satisfy Mt +- X condition through cuts.f.

Obviously setting only a cut in cuts.f is very dangerous since the phase-space integrator will not know in advance this cut and therefore you hope that the machine learning part (VEGAS) of the code is smart enough to handle it.
given that you are already trying a VERY complicated matrix-element (pure interference + decay-chain), you will need to be very carefull with the phase-space integrator.

I believe that it would make more sense to
first implement intereference with decay-chain in madgraph
and then study how to integrate those (my latest paper gives me hope in that direction)
than trying to use risky work around that would need even more time to validate and would not be transferable to the community.

Two year ago, I had money to start such project but no one volunteer and we are now out of money.
If you are interested in implementing that, I will still be interested to coaching someone but I can not pay anymore.

Cheers,

Olivier

> On 19 Mar 2021, at 01:45, Jack Y. Araz <email address hidden> wrote:
>
> Question #696110 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/696110
>
> Status: Answered => Open
>
> Jack Y. Araz is still having a problem:
> Dear Olivier
>
> Thanks a lot for the detailed answer, it was very useful! Now that you
> said it might create a problem to increase bwcut, excluding off-sell
> tops would be much more useful for the physics that I'm interested in.
> This, I believe, can be done by selecting the decay products of top and
> requiring each of the group to satisfy Mt +- X condition through cuts.f.
> However, I'm not sure if it will be accurate if I do it solely based on
> phase-space, is there any way to access the particles' pdgID so that I
> can match them with respect to their charge? This way, I believe I can
> fix all the problems that you mentioned above.
>
> I believe its also possible to define the process as
>
> p p > t t~, t > b l nu, t~ > j j b~
>
> but in this notation, I'm not able to set NP^2==2 to isolate the
> interference term for both production and decay vertices.
>
> Thank you so very much!
> Best regards
> Jack
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Jack Y. Araz (jackaraz) said : #4

Dear Olivier

Thanks a lot for your help! Let's discuss about the project you had in mind through email!

Best regards
Jack