Wednesday, 14 November 2012

Higgs: what's new

I know, there's already a dozen of nice summaries on blogs (for example here, here, and here) so why do you need another one? Anyway... the new release of LHC Higgs results is the clue of this year's HCP conference (HCP is the acronym for  Human CentiPede). The game is completely different than a few months ago: there's no doubt that a 126 GeV Higgs-like particle is there in the data,  and nobody gives a rat's ass whether the signal significance is 5 or 11 sigma. The relevant question now is whether the observed properties of the new particle match those of the standard model Higgs.  From that point of view, today's update brought some new developments, all of them depressing.

The money plots from ATLAS and CMS summarize it all:

We're seeing the Higgs in more and more channels, and the observed rates are driven, as if by magic, to  the vertical line denoting the standard model rate. 

It came to a point where the most exciting thing about the new Higgs release was what wasn't there :-) It is difficult not to notice that the easy Higgs search channels, h→γγ and ATLAS h→ZZ→4l,    were not updated.  In ATLAS, the reason was the discrepancy between the Higgs masses measured in  those 2 channels: the best fit mass came out 123.5 GeV in the  h→ZZ→4l, and 126.5 GeV in the  h→γγ channel. The difference is larger than the estimated mass resolution, therefore ATLAS decided to postpone the update in order to carefully investigate the problem.  On the other hand in CMS, after unblinding the new analysis in the h→γγ channel, the signal strength went down by more than they were comfortable with; in particular the new results are not very consistent with what was presented on the 4th of July. Most likely, all these analyses will be released before the end of the year, after more cross-checking is done.

Among the things that were there, the biggest news is the h→ττ decay. Last summer there were some hints that the ττ  channel might be suppressed, as the CMS exclusion limit was reaching the standard model rate. It seems that the bug in the code has been corrected:  CMS, and also ATLAS, now observe an excess of events over the non-Higgs backgrounds consistent with what we expect from the standard model Higgs.  The excess is not enough to claim observation of this particular decay, but enough to suppress the hopes that some interesting physics is lurking here. 

Another important update concerns the h→bb decay, for the Higgs produced together with a  W or Z boson. Here, in contrast, earlier  hints from the Tevatron suggested that the rate might be enhanced by a factor of 2 or so. The LHC experiments are now at the point of surpassing the Tevatron sensitivity in that channel, and they don't see any enhancement: CMS observes the rate slightly above the standard model one (though again, the excess is not enough to claim observation), while ATLAS sees a large negative fluctuation. Also, the Tevatron has revised downward the reported signal strength, now that they know it should be smaller.  So, again, it's "move on folks, nothing to see here"...

What does this all mean for new physics? If one goes beyond the standard model, the Higgs couplings to matter can take in principle arbitrary values, and the LHC measurements can be interpreted as constraints on these coupling. As it is difficult to plot a multi-dimensional parameter space, for presentation purposes one makes simplifying assumptions.  One common ansatz is to assume that all tree-level Higgs couplings to gauge bosons get rescaled by a factor cV, and all couplings to fermions get rescaled by an independent factor cf.  The standard model corresponds to the point cf=cV=1. Every Higgs measurement selects a preferred region in the  cV-cf parameter space, and measurements in different channels constrain different combinations of cV and cf.  The plot on the right shows 1-sigma bands corresponding to individual decay channels, and the 68%CL and 99%CL preferred regions after combining all LHC Higgs measurements. At the end of the day,  the standard model agrees well with the data. There is however a lower χ2 minimum in the region of the parameter space where the relative sign between the Higgs couplings to gauge bosons and to fermions  is flipped. The sign does not matter for most of the measurements, except in the h→γγ channel. The reason is that h→γγ is dominated by two  1-loop processes, one with the W boson and one with the top quark in the loop. Flipping the sign changes the interference between these two processes from destructive to constructive, the latter leading to an enhancement of the h→γγ rate in agreement with observations. On the down side, I'm not aware of any model where the flipped sign would come out naturally (and anyway the h→γγ will  go down after CMS updates h→γγ, probably erasing the preference for the non-SM minimum).

Finally,  we learned at the HCP that the LHC is taking precision Higgs measurements to a new level, probing not only the production rates but also more intricate properties of the Higgs signal. In particular,  CMS presented an analysis of the data in the h→ZZ→4l channel that discriminates between a scalar and a pseudoscalar particle. What this really means is that they discriminate between 2 operators allowing a decay of  the Higgs into Z bosons: 

The first operator occurs in the standard model at tree level, and leads to a preference for decays into  longitudinally polarized Z bosons. The other is the lowest order coupling possible for a pseudoscalar, and leads to decays into transversely polarized Z bosons only. By looking at the angular distributions of the leptons from Z decays (a transverse Z prefers to emit leptons along the direction of motion, while a longitudinal Z - perpendicularly to the direction of motion) one can determine the relative amount of transverse and longitudinal Z bosons in the Higgs sample, and thus discriminate between the two operators.  CMS observes a slight 2.5 sigma preference for the standard model operator, which is of course not surprising (it'd be hard to understand why the  h→ZZ rate is so close to the standard model one if the other operator was responsible for the decay). With more data we will obtain more meaningful constraints on the higher dimensional couplings of the Higgs.

To summarize,  many particle theorists were placing their bets that Higgs physics is the most likely place where new physics may show up. Unfortunately, the simplest and most boring version of the Higgs predicted by the standard model is emerging from the LHC data. It may be the right  time to start scanning job ads in condensed matter or neuroscience ;-)

All Higgs parallel session talks are here (the password is given in the dialog box).


Anonymous said...

you know too much about atlas internal discussion.

Jester said...

it's all over facebook :-)

Dan D. said...

Frustrating that the diphoton channel wasn't updated for either experiment. What are the chances we'll have to wait until Moriond for the next update?

Jester said...

My guess is they will release the 13fb-1 diphoton results in a few weeks. In Moriond they will surely update all channels with the full 8 TeV dataset.

Robert L. Oldershaw said...

Are we ready for new ideas yet?

Not S.O.S. "new ideas", but truly different approaches to understanding nature that seriously question all inadequately tested assumptions, like strict reductionism.

Anonymous said...

Condensed matter or neuroscience would be fine. Please not finance and economics this time.

chris said...

i really love your summaries :-)

going to scan a few more job ads now...

Tony Smith said...

Alexey Drozdetskiy at HCP 2012 said about background in the ZZ to 4l channel
"… ZZ to 4lpeak is in place and in agreement with prediction …".

Felipe in comment on Tommaso Dorigo blog said about diphoton channel
"… The two photon signal was ... suspect because the background was not calculated in the SM
but fit to some ad hoc polynomials …".

Since the diphoton channel sees only a small bump on a very large background curve that seems to be smooth and therefore be well represented by polynomial fitting
a question arises:

Which is better representation of background:
polynomial fit or SM calculation ?

To see that the answer may not be obvious, consider that the diphoton situation may be

different from ZZ to 4l which has easy SM background calculations verifiable by checking the Z to 4l peak
more like the A_FB situation as to which Bryan Webber at ETH Oct 2012 said
"… Is The Top Quark Asymmetry Just Standard-Model Physics?"
... Asymmetry larger than NLO SM seen by CDF in several independent data sets ...
HO SM prediction not yet clear (recoils) ...".

Could the SM calculations needed for diphoton background also require Higher Order results than presently available ?

If so, might it be that given the present state of SM calculations
a polynomial fit diphoton background might show the diphoton bump more accurately ?


cb said...

The most boring version of the Higgs ? Come on, falks! Have a closer look on its mass ... Let's remember Jean Iliopoulos :
... and instead of looking for another job why not scanning other serious alternative theories not necessary incompatible with supersymmetry and going beyond the standard model (with non-commutative geometry for example)


Anonymous said...

Can you say when we'll know the particle's spin for certain, or near certain? Very likely to be 0, but if there's a major surprise waiting, perhaps that's one of the few possibilities. Maybe you could tell us about other possibilities of that kind as well, thank you.

Jester said...

Yes it 99.99% certain, in the Bayesian sense, because the production rate of the 125 GeV particle observed by the LHC are close to those predicted for the standard model Higgs. Particles with other spin or parity than the Higgs would have a completely different tensor structure of their couplings to matter, even the dimension of the couplings would be different. It would be an incredible coincidence if those different couplings led to the production rates similar (even within a factor of 2) as in the standard model.

Anonymous said...

Thanks. And other surprises - will we know next March that there aren't any that can appear without a large coincidence of the kind you mention, or do we know that already?

Jester said...

We know the Higgs looks like the standard model one in the first approximation, but of course there is still a lot of room for surprises. Some processes we have not seen at all (e.g. tth production), and in others the experimental error on the cross section is order 100%. So we can easily have order 1 deviations from the standard model predictions, without much conspiracy or coincidences. What I meant in the previous comment is that for spin 2 or 0- one would naturally expect deviations much larger than order 1.

Anonymous said...

I guess you're saying it's got to be either a standard Higgs or a non-standard Higgs, and that it'd take a large coincidence for it not be to be a Higgs at all.

Jester said...