Choose timezone
Your profile timezone:
Powered by Indico
v3.3.2-pre
The meeting is focused on the determination of expected numerical EFT effects from models and constraints on models from expected or obtained EFT fits, matching to specific models, BSM-driven subsets of operators, benchmarks beyond SMEFT, incl. non-linear EFT and the comparison of EFT constraints vs. direct BSM searches beyond EFT.
///////////////////////////////////
General summary of experimental talks:
Some concerns had been raised on the usage of the EFT limits to constrain directly MSSM. To be followed up on.
///////////////////////////////////
Q&A Homiller (min 22)
Luca: The log terms are independent of the renormalisation scheme so they are well defined. The one-loop matching depends on everything -- the scheme that you choose for gamma_5, the renormalisation prescription - an there are schemes that are defined to have the one-loop to be the same as the tree level --
so to make full numerical sense of this, you should go to a full NLO calculation where you include also the NLO anomalous dimension and then everything will cancel out (i.e. scheme dependence of the NLO one-loop with the anomalous dimension).
So to really go and use this matching, we should really look at these effects. e.g. what happens if you vary the matching scale by a factor of 2?
SH: Did not play too much with this. We did a couple of examples, here slide 25, we show how the matching scale changes. But you are absolutely right, that to be fully consistent you need the 2-loop anomalous dimensions (which have not been calculatated). That's a great point.
----
Tania: A comment and a question. First the comment on slide 19: The left-hand side - if you go to LEP limits, you are out of the left side limits, then you get much smaller values for the sinTheta (around 0.2), just if you include direct searches.
SH: yes, this is true.
Tania: Yes, just as a comment, not that people get too excited and try to look for this. But I guess you wanted to show something else here.
SH: Yes, we just wanted to show the plain limit you get on the mixing angle. But you are right, if this is that light, then you will have direct limits from LEP.
Tania: Question on slide 26: This is all in the SMEFT description framework, between tree and one-loop mixing. But where would be the bound if you do not take the SMEFT description but just the model as such?
SH: So just a naive limit from the Higgs couplings?
The global here contains all the Higgs couplings (the kappa measurements) and the EWPO would be just this is the idea.
Slide 23: Here we have the EWPO curve and in the paper we also have the comparison that shows you agree very well with th exact model.
Tania: Yes, this is of course also very interesting to know how wrong or wright you are in comparing the SMEFT bounds with the actual bounds.
----
Andrei: slide 23: Here you get constraints from precision measurement, but what about direct searches for singlets compare at the high mass?
SH: If this thing is really really heavy, so TeV scale or larger, then I do not thing, you'd have anyway of probing this things directly.
Slide 29 (backup) shows resonant di-higgs limits, and here sin_alpha is the same as our sin_theta.
You see these cover a lot between 600-700 GeV, but beyond it is hard to produce.
Andrei: But why is it only di-Higgs?
SH: It inherits all it's mixing from the Higgs, and if this sin_theta is small, this is suppressed by it. You could look directly via all productions, but generally the di-Higs should be best.
Tania: I disagree: The best channels are diboson searches in fact. I mean you can play around obviously, so you can tune up the scalar up to 250, here you have more parameters to play with because of your symmetries, but the phenomenology is very similar. So in the C_2 you only have 1 additional parameter, and this you can use to tune up and down the Higgs to Higgs-Higgs coupling. And then you can change it such that you have dominantly the Higgs to diboson decay which are only constrained from the signal strength.
So even 3 years ago you have quite strong constaints from diboson searches. But as I said, you can tune it up and down, so you can find points in the parameter space, where this better or worse.
SH: But if this is at the 2 TeV range, then there are really no constraints.
Tania: I don't know, because there are searches that expand higher. I get asked, why my plots only extend to 1 TeV?
Andrei: Experimental searches go far beyond 1 TeV.
Tania: I think for run-2 there are bounds beyond that. So just to say, that diboson is quite powerful.
---
chat related to this discussion:
14:22:41 From Andrei Gritsan : example of di-boson up to 3 TeV: https://arxiv.org/pdf/1804.01939.pdf
14:31:00 From Tania Robens : dear andrei, all, see e.g. htps://arxiv.org/abs/1908.10809 fig 1 for a comparison of bounds (for fixed tan be) - of course also outdated now.
14:37:05 From sallydawson : Ah but this doesn’t have the diboson limits. The point of Sam’s talk wasn’t the singlet model but an exploration of the SMEFT ingredients
14:39:45 From Tania Robens : hi sally - did this refer to me (sorry too many parallel discussions...) ? :-) the green line in that figure is basically VV bounds by that time. just an answer to andrei. obviously nothing to do w the main topic of samuels talk...
14:41:23 From Andrei Gritsan : From my side: I was curious to compare EFT constraints to direct searches
14:42:21 From Tania Robens : ?? but you can also compare direct searches to the eft description of the respective process - what do you mean by "constraints" ?
14:42:36 From Tania Robens : "eft constraints" (sorry)
14:48:21 From Andrei Gritsan : I should be more careful: compare direct search for a new state (singlet) with “indirect" constraints without information from that search. In any case, thanks for the followup information!
14:52:04 From Tania Robens : i think you have 9 categories: a1) direct searches, direct model b1) indirect experimental bounds (stu/...), direct model, c1) bounds from theory, direct model. then the same in eft, leading order mathcing. and then eft, one-loop mathcing...
15:05:17 From Tania Robens : hi andrei - what you want is basically done in many many bsm parameter scan papers...... i think more or less each single scan paper has at least one plot where things like this are compared. :-)
///////////////////////////////////
Q&A Joydeep Chakrabortty
no discussion
///////////////////////////////////
Q&A Jose Santiago
Sally: Are there any prospects to expand this to model which have 2 new particles or more?
JS: Unfortunately not enough time to include everything, but we are working up to include this to have more. At dim6 or tree-level it is trivial to add as many particles as you want. Loop-level or dim8 is more complicated.
Admir: This numerical exercise, that you have shown, the small fit with thte 5 directions. This was focusing on the directions which are purely constrained at the moment.
JS: Yes, the idea was to set those to 0 which were very small, so being able to reduce from 22 to 5. And then we added some U(2) flavour symemtries for th quarks for simplicity to get some numbers.
Ken: I wnated to come back to the inherent scheme dependence due to the one-loop matching. You mentioned this CDE, color derivative expansion, and I would like you clarify: what is the scheme there? For the diagram matching it's clear, but for the CDE the scheme is not so obvious, so would be nice to clarify that.
JS: maybe the next 2 speakers will answer this better than me. But basically you are expanding in momentum integrals, and there you decide what you want to do. e.g. with the gamma_5. And you need to define the EW-scheme, e.g. MS-bar, but you get very explicit momentum integrals.
Ken: Ok, so in a sense the same as diagrammatic matchting.
JS: Yes, this is just a different way of organisin the diagrams, but in the end you get the momentum integral that you want.
Alexander: I have a question on the mass scales. You start from EFT limits, where we have the SM and an EFT expansion at a scale way above LHC energies. Now, I mean in the first talks where we saw limits for energies that were reachable by the LHC. Can you use these kinds of models on hybrid cases, where we have masses appearing at the LHC but not all of them. Eg. take a SUSY case, where you could find a few particles, but not the rest, because they are at higher energies.
can you then still use the EFT limits?
JS: This depends a lot on which data you are using and what the masses of the particles are. The restriction is that the scale or the masses of the particles need to be much higher than the masses you are experimentally probing. If I expect some particles, I could only compare the direct limits at the LHC with the indirect ones from LEP.
Alexander: So you really need to make a decision which is the data to use.
JS: Yes, and indeed afterwards, once you have found the models/particles allowed by the global fit, you should go back and check that they are within the validty of the EFT and that you are not violating any direct search limits.
Alexander: A quick a follow-up: Assume we would assume a new particle and then would we be able to incorporate this and re-apply some other EFT to check for other stuff the UV scale?
JS: Yes exactly, you would include the lighter states in the SMEFT and based on that build a new EFT model.
Aleksander: that would require to re-interpret the global fit.
///////////////////////////////////
Q&A Anders Eller Thomsen
Admir: In the first talk, we had the talk from SH, for benchmark models and the matching there was done at tree level. Given your example here: how difficult would it be to write down the loop level for the vector-like top quark?
AET: It should not be very hard, but it's mostly just a matter of writing down the X-matrices, but it should be doable in a matter of a day. Then you'd need to match the output to the smeft and that might take a little longer. but it's certainly much quicker than doing it the traditional way from first principles.
Admir: It would be very good to have some benchmark examples, where different groups could compare their approaches.
AET: Yes, just for peace of mind very useful. It's nice to compare the different tools to compare them to see they all work well.
///////////////////////////////////
Q&A Xiaochuan Lu
no discussion