Thursday, 24 December 2015

750 and what next

A week has passed since the LHC jamboree, but the excitement about the 750 GeV diphoton excess has not abated. So far, the scenario from 2011 repeats itself. A significant but not definitive signal is spotted in the early data set by the ATLAS and CMS experiments. This announcement is wrapped in multiple layers of caution and skepticism by experimentalists, but is universally embraced by theorists. What is unprecedented is the scale of theorist's response, which took a form of a hep-ph tsunami.    I still need time to digest this feast, and pick up interesting bits among general citation fishing.  So today I won't write about the specific models in which the 750 GeV particle could fit: I promise a post on that after the New Year (anyway, the short story is that, oh my god, it could be just anybody). Instead, I want to write about one point that was elucidated by the early papers,  namely that the diphoton resonance signal is unlikely to be on its own, and there should be accompanying signals in other channels. In the best case scenario, confirmation of the diphoton signal may come by analyzing the existing data in other channels collected this year or in run-1.

First of all, there should be a dijet signal. Since the new particle is almost certainly produced via gluon collisions,  it must be able to decay to gluons as well by time-reversing the production process. This would show up at the LHC as a pair of energetic jets with the invariant mass of 750 GeV. Moreover, in simplest models the 750 GeV particle decays to gluons most of the times. The precise dijet rate is very model-dependent, and in some models it  is too small to ever be observed, but typical scenarios predict order 1-10 picobarn dijet cross-sections. This would mean that thousands of such events have been produced in the LHC run-1 and this year in run-2. The plot on the right shows one example of a parameter space (green) overlaid with contours of dijet cross section (red lines) and limits from dijet resonance searches in run-1 with 8 TeV proton collisions (red area). Dijet resonance searches are routine at the LHC, however experimenters usually focus on the high-energy end of the spectrum, far above 1 TeV invariant mass. In fact, the 750 GeV region is not covered at all by the recent LHC searches at 13 TeV proton collision energy.

The next important conclusion is that there should be matching signals in other diboson channels at the 750 GeV invariant mass. For the 125 GeV Higgs boson, the signal was originally discovered  in  both the γγ and  the ZZ final states, while in  the WW channel the signal is currently similarly strong. If the 750 GeV particle were anything like the Higgs, the resonance should actually first show in the ZZ and WW final states (due to the large coupling to longitudinal polarizations of vector bosons which is a characteristic feature of Higgs-like particles).  From the non-observation of anything interesting in run-1 one can conclude that there must be little Higgsiness in the 750 GeV particle, less than 10%.  Nevertheless, even if the particle has nothing to do with the Higgs (for example, if it's a pseudo-scalar), it should still decay to diboson final states once in a while. This is because a neutral scalar cannot couple directly to photons, and the coupling has to  arise at the quantum level through some other new electrically charged particles, see the diagram above. The latter couple not only to photons but also to Z bosons, and sometimes to W bosons too.  While the details of the branching fractions are highly dependent, diboson signals  with comparable rates as the diphoton one are  generically predicted.  In this respect, the decays of the 750 GeV particle to one photon and one Z boson emerge as a new interesting battleground.  For the 125 GeV Higgs boson, decays to Zγ have not been observed yet, but in the heavier mass range the sensitivity is apparently better.  ATLAS made a search for high-mass Zγ resonances in the run-1 data,  and their limits already put non-trivial constraint on some models explaining the 750 GeV excess. Amusingly, the ATLAS Zγ search has a 1 sigma excess at 730 GeV...   CMS has no search in this mass range at all, and both experiments are yet to analyze the run-2 data in this channel.  So, in principle,  it is well possible that we learn something interesting even before the new round of collisions starts at the LHC.

Another generic prediction is that there should be vector-like quarks or other new colored particles just behind the corner. As mentioned above, such particles are necessary to generate an effective coupling of the 750 GeV particle to photons and gluons. In order for those couplings to be large enough to explain the observed signal,  at least one of the new states should have mass below ~1.5 TeV. Limits on vector-like quarks depend on what they decay to,  but the typical sensitivity in run-1 is around 800 GeV. In run-2, CMS already presented a search for a charge 5/3 quark decaying to a top quark and a W boson, and they were able to improve the run-1 limits on the new quark's mass from 800 GeV up to 950 GeV. Limits on other type of new quarks should follow shortly.

On a bit more speculative side, ATLAS claims that the best fit to the data is obtained if the 750 GeV resonance is wider than the experimental resolution. While the statistical significance of this statement is not very high, it would have profound consequences if confirmed. Large width is possible only if the 750 GeV particle decays to other final states than photons and gluons. An exciting possibility is that the large width is due to decays to a new hidden sector with new light particles very weakly or not at all coupled to the Standard Model. If these particles do not leave any trace in the detector then the signal is the same monojet signature as that of dark matter: an energetic jet emitted before the collision without matching activity on the other side of the detector. In fact, dark matter searches in run-1 practically exclude the  possibility that the large width can be accounted for uniquely by invisible decays (see comments #2 and #13 below).  However, if the new particles in the hidden sector couple weakly to the known particles, they can decay back to our sector, possibly after some delay, leading to complicated exotic signals in the detector. This is the so-called hidden valley scenario that my fellow blogger has been promoting for some time. If the 750 GeV particle is confirmed to have a large width, the motivation for this kind of new physics will become very strong. Many of the possible signals that one can imagine in this context are yet to be searched for.    

Dijets, dibosons, monojets, vector-like quarks, hidden valley...  experimentalists will have hands full this winter.  A negative result in any of these searches would not strongly disfavor the diphoton signal, but would provide important clues for model building. A positive signal would break all hell loose, assuming it hasn't yet. So, we are waiting eagerly for further results from the LHC,  which should show up  around the time of the Moriond conference in March. Watch out for rumors on blogs and Twitter ;)

34 comments:

Somdatta Bhattacharya said...

Jester,
Why did neither ATLAS nor CMS bother to do the dijet analysis for 750 GeV and nearby? It would have been the obvious thing to do once they found the diphoton signal. Any ideas?

Anonymous said...

I don't understand why you write that most of the width cannot be into invisible states. A counterexample is: Gamma_tot ≈ 0.06 M, BR_gg≈ 1%, BR_gammagamma ≈ 0.3%, BR_inv ≈ 98.7% = all the rest

Unknown said...

In general jet energy calibration is a hectic job in both ATLAS and CMS, specially in the high pileup scenario. It takes quite an effort for precise determination of the jet energy scale and hence people tend to play safe on reporting multi-jet final states. This photon and lepton momentum scales are far better understood in a hadron collider. So the next best channel should be 4-lepton (ZZ resonance search ~ 750 GeV) or di-lepton + MET channels.

Somdatta Bhattacharya said...

OK, thanks for the clarification.

Anonymous said...

Are there any models pre announcement that make any such predictions of a particle like this? Also what about that slight bump at around 1.9Tev, is that now considered false?

Chris Austin said...

"a neutral scalar cannot couple directly to photons"

That would be in a renormalizable model. But if the 750 GeV particle is a component of an extra-dimensional graviton, for example if it's a dilaton, would it not then generically have a tree-level coupling to 2 photons? e.g. from within the sqrt{-G} in sqrt{-G} G^{I K} G^{J L} F_{I J} F_{K L}?

Could you explain a bit more about why 750 GeV decay to a hidden sector would give a monojet signature? Is it just that a monojet is the easiest way to balance large MET? Would the mass of the particle decaying to the hidden sector be largely unidentified with this signature?

Tommaso Dorigo said...

Hi Jester,

I don't see how you can say we did not cover 750 GeV dijet resonances. Atlas and CMS have limits in the few pb range there. Plus, a data scouting result to come out soon does exclude even more there.

Cheers,
T.

Anonymous said...

Hello Jester,

What is your estimate at the moment for this anomaly being new physics ?

Anonymous said...

Hi Jester, the gamma-Z excess you point to in the ATLAS paper (run 1) peaks at around 714 GeV. At 750 GeV, it is between 0 to 0.5 sigma depending on which figure I look at.

Jester said...

That's right, but a particle with mass a bit below 750 GeV also gives a good to the diphoton data, especially if the width is larger than the resolution.

Jester said...

Anon-2: 10% (larger than ever)

Jester said...

Tommaso: I'm confused. E.g. in CMS's Fig.3 of [arXiv:1512.01224] all plots are cut off at ~1.5 TeV. Similar for ATLAS. Maybe you mean run-1 searches? These indeed go below 1 TeV and set useful limits there.

Jester said...

Anon 24/12/16:32: it's not excluded but very implausible. Suppose the width is dominated by invisible decays. Then, due to run 1 monojet bounds, the total cross section should be less than 4pb at 13 Tev. Then, to fit the diphoton excess, the branching fraction to photons must be at least 0.1%, which implies the diphoton width must be at least 50 Mev. You need very large loop contributions to the effective photon coupling to get this large width. E.g. you would need a triplet of 300 GeV vector-like leptons with Yukawa couplings to the 750 GeV particle of order 6.

Jester said...

Chris: that's right, the statement is true for renormalizable theories. But it amounts to the same: if you have the non-renormalizable coupling then your theory stops making sense near the weak scale and you need to complete it with new degrees of freedom that explain the effective coupling to photons. E.g., your graviton or dilaton would really be some composite states, and at some energy scale you would need to introduce the full dynamical model which would contain new charged particles coupled to photons and gluons.

Concerning monojets: when you produce only dark matter the detector sees nothing and does not trigger. You need dark matter to be produced together with something visible, and the most common process in hadron colliders is jet emission. This can e.g. happen when one of the colliding gluons emits an energetic gluon before the collision. Then the dark matter particles get a sideway kick and recoil against that gluon, and you can see the missing momentum in the detector. So monojets is just the simplest topology where you can actually 'see' the invisible particles. In vanilla models of dark matter this is the most sensitive channel, but you can search for dark matter in other final states as well.

Marco Frasca said...

I would not exclude the possibility that this is another fatter Higgs particle. This because the Higgs sector of the Standard Model, without any other interaction than itself, can be solved exactly. Exact solution can do better with respect to perturbative ones, as it is known from textbooks.

Tommaso Dorigo said...

Right, I was talking of Run 1 search results... They still count, right ? Or do they expire after 3 years ? ;-)

Cheers,
T.

Jester said...

that's right, the statement is true for renormalizable theories

as for monojets: when you produce only dark matter the detector sees nothing and does not trigger. you need dark matter to be produced together with something visible, and the most common process in hadron colliders is jet emission. so monojets is just the simplest topology where you can actually see the invisible particles.

Anonymous said...

Throes of desperate theorists hammering the arxiv, trying to stake their claim to...
Hopefully something but likely nothing.

Anonymous said...

Are you the new person drawn toward me, and asking
something significant from me?
To begin with, take warning—I am probably far
different from what you suppose;
Do you suppose you will find in me your ideal?
Do you think it so easy to have me become your
lover?
Do you think the friendship of me would be unalloyed
satisfaction?
Do you suppose I am trusty and faithful?
Do you see no further than this façade—this smooth
and tolerant manner of me?
Do you suppose yourself advancing on real ground
toward a real heroic man?
Have you no thought, O dreamer, that it may be all
maya, illusion? O the next step may precipitate
you!

CALAMUS. Walt Whitman

Alex Small said...

Is it likely that 750 GeV particle could be produced by cosmic rays? If so, could current experiments detect their decay products?

My inclination is that cosmic rays searches wouldn't be much help because you need to detect two photons with different momenta to nail down the mass, but maybe some characteristic decay spectra could be detected? But not my field, so I'm putting the question out there for the experts.

Jester said...

people are thinking whether one can learn something from cosmic rays but so far we don't even know how to "discover" the Z boson. All particles are produced by cosmic rays and there is enough energy to make heavy ones. The main problem is a low "luminosity" so the production rate of weak scale particles is too low.

clayton said...

Oooh, I *really love* the Feynman diagram you posted. Did this deep, perceptive, inspirational combination of straight and squiggly lines come from the hep-ph tsunami, and if so, which of the papers is it ;) ??!!?!??!!!!?

Anonymous said...

best explanation until now: https://xkcd.com/1621/

Robert L. Oldershaw said...

The deluge of papers on this subject at arxiv is a bit of a concern.

Like the gold rush days the idea is to strike a claim in any halfway likely spot and hope you get lucky enough to claim ownership of something valuable.

Somewhat unseemly, in my opinion.

Anonymous said...

I think this tsunami of papers, as you put it Jester, is giving us more insight into the bad shape of high energy physics phenomenology as a field rather than uncovering new valuable knowledge (for which I think you are overoptimisitc). The status by now is that this is simply an statistical fluctuation, one can be interested of course, but in no case it justifies the avalanche of low quality papers. The quality of these papers is so low other scientist will laugh. Remember physicists laughing about Medicine or other fields scientific practices, this is similar or worst.

Alex Small said...

Not that I have followed it too closely, but the takeaway messages I have seen indicate that the 750 GeV particle can vindicate pretty much every speculation out there except (oddly) SUSY. Why is that? Is SUSY less malleable than we'd been led to believe? If so, that would actually raise my respect for SUSY as a meaningfully falsifiable model.

Anonymous said...

And another dozen of articles about the excess on arxiv today. Maybe we should ask the arxiv admins to create a new category hep-gw (guesswork). That would make it easier to find the serious papers among all the spam these days.

Anonymous said...

I find annoying that hep-ph is still accepting papers on topics unrelated to the new S_750GeV

Jester said...

Alex: I think the statement is that that the 750 GeV diphoton resonance does not fit in the simplest and most popular SUSY models, such as the MSSM or the NMSSM. But the excess does not disfavor (or favor) supersymmetry. I suppose that any of the 100 models proposed so far can be embedded in a supersymmetric framework.

Jester said...

Alex: I think the statement is that that the 750 GeV diphoton resonance does not fit in the simplest and most popular SUSY models, such as the MSSM or the NMSSM. But the excess does not disfavor (or favor) supersymmetry. I suppose that any of the 100 models proposed so far can be embedded in a supersymmetric framework.

mfb said...

What about a category hep-750?
You can still crosslist everything to hep-ex just to annoy experimentalists looking for experimental results.

Anonymous said...

Are leptoquarks looking less likely now?

Jester said...

Leptoquarks may be hinted at by other, completely independent measurements related to B-meson decays. Nothing has changed in this respect. Subjectively, these hints have less weight for me than the diphoton resonance, however both (or none) may well be true.

Anonymous said...

You youngsters never perused PRL in the month after the J/psi was found...