NLO Higgs production in HEFT not correct with 3.X.X

a likely related issue is that the following script is also crashing:

import model HC_NLO_X0_UFO-heft
generate p p > x0 / t QNP<=1 [QCD] @0 \

add process p p > x0 j / t QNP<=1 [QCD] @1 \

output
launch
shower=off
set mt 173.0
set mass 25 125
set nevents 1000
set pdlabel lhapdf
set lhaid 244600
set reweight_scale True
set reweight_pdf False
set store_rwgt_info True
set ickkw 3
set jetradius 1.0
set ptj 10
set etaj -1
set ptgmin 20
set r0gamma 0.4

with:
 Collinear tests done for (Born) config 1
 Failures: 100 0 0 0 91
Sum of all orders: Collinear test 10 FAILED. Fraction of failures: 1.00
Split-order 2: QCD: 4; QNP: 2; QED: 0; Collinear test 10 PASSED. Fraction of failures: 0.00
Split-order 3: QCD: 2; QNP: 2; QED: 2; Collinear test 10 PASSED. Fraction of failures: 0.00
Split-order 4: QCD: 0; QNP: 2; QED: 4; Collinear test 10 FAILED. Fraction of failures: 0.91

Rik, any update on this?

Would be nice to really fixing all the small bug in 3.x version such that we can drop 2.x in a not so far future.

Olivier

Hi Olivier,

I'm not sure how to fix this in the best way. The issue is, as far as I can tell, that there is nowhere in the model/code that the new g-g-H interaction contains, in fact, a strong coupling squared. It's always simply labelled as a new order 'QNP', but not that 'QNP=QCD^2*QED'. Hence, the code doesn't know that in the QCD running, it should include two extra powers of alpha_S coming from QNP as compared to what is expected from the QCD power alone.

I guess the cleanest way would be to fix this directly in the model. But since I'm not familiar with UFO, I do not know if this feature exists and how madgraph treats it in NLO processes.

Rikkert

Hi Marco, Rikkert,

sorry to ask here. Would you have any update in fixing in the UFO model or directly in MadGraph? We (ATLAS) would be interested in having this available in MG3.

Thanks!
Ana

Hello Ana,

Olivier promised to have a look into this.
It's not a very straightforward fix, so it might take some time to get this done.

best,
Rikkert

Hi,

What I have done so far is a (not bullet proof) but I check when some coupling does depend on aS but does not have a QCD flag. (Not bullet proof since I do not check the case where the power is incorrect)

If this situation appears then I print the following message:
WARNING: coupling GC_3002h=-complex(0,1)*cosa*kHgg*gHgg has direct dependence in aS but has QCD order set to 0. Automatic computation of scale uncertainty can be wrong for such model.
WARNING: To fix the coupling order of the model, please run customize_model set_coupling_order
WARNING: coupling GC_3002a=-sina*kAgg*complex(0,1)*gAgg has direct dependence in aS but has QCD order set to 0. Automatic computation of scale uncertainty can be wrong for such model.
WARNING: To fix the coupling order of the model, please run customize_model set_coupling_order

As you can see from this warning, I want to allow to fix automatically the model via a command
but the command is not yet implemented (and the command syntasx might change)

The idea here would be to set the QCD coupling order to the power of aS (which will impact the syntax needed to generate the events in some cases)

Cheers,

Olivier

Hi Rik,

I'm afraid that the problem is deeper or require something even more complex.
In the model in attachment, I have modified the original model in order to increase by "2" every coupling (and counterterm/R2/...) which did have an hidden dependency in G**2.
I believe that I catch all of them. The procedure will be complex to fully automate but in principle easy enough to modify the few problematic models.

However, if I do
import model HC_as_UFO-heft
generate p p > x0 / t QNP<=1 QCD<=2 [QCD] @0
then I know have a "pole do not cancel". Do you think that I have missed some modification of the coupling order? or do you think that this is even more complex than I though?

Cheers,

Olivier