DMSimp_t progress (Jakob)
Slide 4: relic density
Question: what are the white regions in the left relic density plot? Lower left: coupling would have to be O(10e-4) to get the right relic density, which is too small to be probed at colliders; Upper right: coupling would have to be in non-perturbative regime, i.e., for smaller values the relic density would be larger than measured by Planck
Slide 9: it looks like the widths are quite large for a coupling of 0.5. Experiments are using lambda=1 -- is this OK?
To be checked
Question: plans to extend to other models, particularly S3D_uR, which CMS and ATLAS are using for monojet?
Yes, plan to do studies for all models.
Discrimination analysis in the works.
Not for the white paper, but rather targeting an independent paper?
F3C (Disha)
Slides 2-3: cuts
Benedikt: what do the jet cuts do?
Disha: Not much; no effect at LO, some small (negligible?) effect for NLO.
Slide 4: K factors for XX production
Disha: not sure why the K factor is so small (0.67) in the upper left corner. Follow up with others not connected.
Slide 6: K factors for YY t-channel
Benedikt: interesting K-factor "island" in the middle right.
Exclusive K factors
Slide 8: MG commands used.
Slides 9-10: DJR plots.
Slides 11-X: different K factors vs. MET
Benedikt: how is MET defined?
Disha: DELPHES generator-level missing transverse energy function.
Benedikt: would be good to check which particles are treated as MET.
Next steps
Think towards how to define strategy for whole grid. Maybe choose a few extreme points first, to see general trends.
Sufficient computing power? Benedikt may be able to find some resources.
Needs more statistics to discern the trends (but clearly there is a slope vs MET).
Monojet constraints?
S3D_uR (Sukanya)
Inclusive K factors: note that YY t-channel is significantly lower than others.
6: MG details, same as Disha as agreed upon.
7-8: encountered an issue with Rivet, causing NLO to be O(10^2) lower than LO.
Sukanya is in contact with Rivet experts.
Taking weights by hand improves situation, but probably still some problem.
Follow up by email for next steps!
Converge on white paper goals/outline, as a guide for consolidating all the studies.
Action Items (what we would like to see in the white paper:)
look at LO 2D maps for the studies shown today
look at differential processes like on p9 and p10 of Disha’s presentation for xx, xy
Include ratio plots (ideally sub-panes)
Check the matching by plotting kinematic distributions (like pT of the leading and subleading jet, etc) and verifying that there are no kinks
higher jet pT to make sure nothing missed
Eventually add scale uncertainties
Timescale for the updates:
next meeting end of July: https://www.doodle.com/poll/8y8x5gunbqcp64t4
next-to-next meeting 2nd week of Sept (7-11 Sept)
F3C_uR model (Disha)
Is m_y > m_dm a sensible choice, no decays to DM if not fulfilled?
in t-channel models, does make sense to have the mediator lighter, since the mediator would become the DM
Q/p9: do you do matching for LO simulation, and PS simulation?
Yes, using Qcut = 30 GeV, MLM matching
Q: yy t-chan vs QCD: why don’t we have to worry about interference?
C: in principle interference is important, but in this case the interference term is small
Q: What is the advantage of splitting in your studies?
A: split is to see whether we can isolate yy_qcd and because the k-factor may behave differently across processes
Q: pT cut on the jet?
A: it’s 100 GeV
Q: Will you also provide xx, xy?
A: yes
Q/p9+10: Can we have the ratio of the plots?
A: yes
C: for xsec: OK w/o differential distributions. Would be interesting to see this this on a 2D plots
Q: how do you apply the k-factor to MC that is LO up to 1 jet?
A: there are two k-factor studies: inclusive (beginning of the presentation, full xsec ratio) and differential (second part of the presentation, in our phase space)
Q: how general are your studies, how hard are the jets?
in yy case no hard pT cut, just 20 GeV,
C: if we go to 2 jets with FxFx → may not work even at Born level
Q: Dynamic scale different at LO and NLO, is there a simple way to make this equivalent?
C: The scale are put in trivailly in NLO
Q: at LO the scale choice is not trivial at LO
A: first generate template at NLO to determine scale choice, and then move to LO accuracy using the same setup!
2nd presentation (Adil, title TBC)
Q/p3: what are the entries?
Q/p6: no dependence on mDM?
Q/p7: which kinematic cuts in generation?
eta < 2.5, pT > 30 GeV, Njet > 0,
Need to check jet pT distributions for smoothness for confirming the choice o fmatching parameters
Q/p8: what is the LO generation? Matching? Which processes including?
A: xy+j, but not the full one; DM DM + jet → would need to include everything.
Q: can show MET?
A: I have produced the plots, see p8
Q: have you checked the width of the mediator? Is it narrow enough?
A: no, set to auto in MG
the width can be quite large, it should be small enough (<20%), else MadSpin will break as decay and production will not factorise
A: confirmed the width small
Q: Is the idea to compare the 1 jet + MET and 2 jet + MET sensitivities
C: yes
Very nice to see all the studies performed!
General comments on all pheno studies:
AI: Encourage everyone doing these studies to pick one restriction card, 3 representative parameter point across the mX,mY plane, and reproduce each others’ results, just to make sure the individual studies provide and apples-to-apples comparison
AI: would be good to agree on common set of mass points and plot the different spin hypotheses against each other
AI: Adopt uniform plotting style
Eventually compare all spin hypotheses on the same point (see AI above)
S3D_uR studies (Sukanya)
Q: Difference between slides 6 and 7?
A: p6 is the QCD contribution (quartic couplings)
Add generation commands for the cross section plots
AI: look up matching efficiency for MLM, which matching threshold?
Characterization Study & LO vs NLO calculations for DM+jet production (Adil)
(going through the slides and noting comments)
p2: why not use lambda = 1 for consistency with other studies?
p2+: clarify which processes are shown. Is this XX, XY, YY? Give generation command on slides.
p7:
use consistent xqcut = 30 GeV
AI: David will look up which xqcut is used at CMS if they use the MLM matching scheme for old t-channel signals
why use particle-level mono-jet analysis rather parametrised detector simulations and mono-jet selections + results from ATLAS/CMS, or model-independent limits from ATLAS/CMS?
p8: is this is sum of all XX, XY, YY processes? Give generation command on slides...
p13+: what is going into these processes?
F3C_uR studies (Disha):
p5:
Q: MET? ME?
It’s MET → will fix label
add ratio panel under plot like here
C: YY channel is dominant now, unlike before. Non-trival effect!
A: if the coupling reduced, QCD will survive, but t-channel will reduce
Compare sum of all subprocesses for LO and NLO
both tables and plots (broken down into sub-processes and total)
mapping the SCC_q model to the ÜberUFO (Shih-Chieh):
Q/p6 missing diagrams -- should be modified in general UFO?
A: talked and coordinated with Benjamin, but did bring up making it a common fix for all channels. Overall impact small
Q/p9: decay width -- mass mPhi = 10 GeV, for most other points with higher masses not relevant?
depends on choice of mDM and mPhi, can be checked easily
Discussion of choices for first grid (slides of David)
For the NLO vs LO decision, agree to wait for further studies, as currently the picture is inconclusive -- some work indicates flat k-factors (favouring LO), other work suggests otherwise (favouring NLO)
Agree as compromise to use S3D_uR for future interpretations
main arguments in favour:
studied in detail in Über-UFO pheno paper, restriction card readily available
simpler from a theory perspective
Broadly speaking, the physics between the SCC_q and S3D_uR models is very similar, as indicated by the exclusion curves
Still valuable to consider SCC_q, as it probes a Lagrangian with LH couplings, and tests for the first two generations
AI: Shih-Chieh will send the param card to Benedikt & David
Agreed to use SCC_b like model for b-quark couplings in the future
Potentially interesting to investigate
Agreed to generate split into sub-processes (XX, XY, YY, etc)
Agreed on lambda = 1
Agree to establish a common set of points
AI: David will make a suggestion
Tentatively agree that NNPDF3.0 would be OK
AI: will confirm with CMS analysers
AI: check if negative weights prevalent
