preparing for the early years of the large hadron collider overview lecture given at the 4 th rtn...
TRANSCRIPT
Preparing for the Early Years of the
Large Hadron Collider Overview Lecture Given at the 4th RTN Workshop
Varna, Bulgaria
Matthew J. StrasslerRutgers University (New Jersey, USA)
The ContextThe LHC:
the best possible place to look for new physicsthe worst possible place to look for new physics What makes it so horrible?! How do we deal with the challenges
Jets are everywhere What are jets anyway?
The Standard Model is a source of large backgrounds to most signals What are the most important, and in what sense?
And now we want to find new physics But what does it look like? What kind of backgrounds must be understood? What should we expect? What should we be careful of?
What I won’t talk about
Heavy Ion Collisions and the connection with String Theory [see H. Liu’s talk]
Diffractive Higgs production and the connection with string theory [Pomeron] – not early LHC
All those different models of extra dimensions, deconstructed extra dimensions, theories dual to extra dimensions fermionic extra dimensions (well, a few words)
Black Holes
BBC Reporting
LHC Cross sections vary over many orders
of magnitude
Every aspect of the experiment is influenced
by this graph
Note the ratios of
• inelastic to b pairs
• b pairs to W
• W to top pairs
• top pairs to Higgs
• top pairs to TeV-scale SUSY
• Higgs to Higgs photons
If we kept all the events
we’d have 1015 / year
In fact 99.9999% must be instantly discarded
The trigger is necessary, crucial, and introduces unavoidable bias
Even after the trigger,
~109 events/year
~103 physicists
An automated system must quickly analyze the huge amount of data from each event.
Another necessary, crucial, and potentially biasing stage
Standard Model produces huge backgrounds: hard to calculate, or measure, or model
Theory is way behind
Experimentalists will determine many backgrounds by measuring
This is not always possible and is fraught with dangerous assumptions
Theory bias can creep in here as well.
ATLAS, CMS, LHCb, ALICE,... General purpose detectors ATLAS, CMS our main focus
What can these detectors do? What can’t they do?
Quick review: LHC Kinematics
Everything is oriented along (longitudinal) and perpendicular (transverse) to beam-pipe
The c.m. frame of proton-proton collision is the lab frame
But c.m. frame of scattering quarks/antiquarks/gluons is not the lab frame
Typical scattering is boosted along beampipe therefore total energy, z-momentum not known
Quick review: LHC Kinematics
7 TeV
7 TeV
Everything is oriented along (longitudinal) and perpendicular (transverse) to beam-pipe
The c.m. frame of proton-proton collision is the lab frame
But c.m. frame of scattering quarks/antiquarks/gluons is not the lab frame
Typical scattering is boosted along beampipe therefore total energy, z-momentum not known
Quick review: LHC Kinematics
x2 7 TeV
x1 7 TeV
The scattering “partons” carry fractions x1, x2 of their protons momentum
Everything is oriented along (longitudinal) and perpendicular (transverse) to beam-pipe
The c.m. frame of proton-proton collision is the lab frame
But c.m. frame of scattering quarks/antiquarks/gluons is not the lab frame
Typical scattering is boosted along beampipe therefore total energy, z-momentum not known
Quick review: LHC Kinematics
Can only apply conservation of transverse momentum Energy and z-momentum of scattering partons not known Not observable: destroyed proton debris lost down beampipe
Transverse momentum (2-d vector) : pT NOT longitudinal momentum NOT energy
Missing transverse momentum if pT not zero. Called “MET” or Missing (Transverse) Energy
x2 7 TeV
x1 7 TeV
The scattering “partons” carry fractions x1, x2 of their protons momentum
What’s easy, what’s hard?
Relatively easy: Detecting and measuring isolated electrons, muons, photons
Very hard Measuring jet energies, momenta for jets (pT> 50 GeV) Interpreting them as quark/gluon energies and momenta
Extremely hard Detecting/measuring low-pT jets Measuring missing transverse momentum
Virtually impossible Telling quark jets from gluon jets or antiquark jets Seeing electrons or photons inside of jets
What’s easy, what’s hard
Easy: Measuring cross-sections times branching ratios times efficienciesNumber of events proportional to Cross section for production , times Branching fraction into particular final state , times “Efficiency” for detecting the particular final state
Hard: Measuring cross-sections times branching ratios Measuring ratios of branching ratios
Need model for the shape of distributions to determine efficiency
Extremely hard: measuring cross-sections or branching ratios separately Must generally do accounting for 100 percent of the produced particle’s decays
But determining theory often requires cross-sections, branching ratios
What’s easy, what’s hard
Easy: Measuring masses of resonances in e, mu, gamma Measuring certain combinations of mass in dilepton decays Measuring charge quantum numbers of particles
Hard: Measuring masses of resonances in jets, taus Measuring non-resonant masses directly – but see new methods! Measuring spins of particles – but see new methods!
Extremely hard: Measuring small mass differences Measuring quark flavor quantum numbers (except t, b, maybe c) Measuring mixing angles [requires many measurements]
Unfortunately, extracting theory often requires masses and mixing angles
Example: Z’ resonance
A 1.5 TeV electron-positron resonance could be discovered by December 2008, or June 2009
Is it spin one? What are its couplings to Quarks vs Leptons? Special couplings to third generation? Tops? Bottoms? Taus? Couplings to Right-handed vs Left-handed Fermions? Does it decay to W+W-? Does it decay to Z Higgs? Does it decay to superpartners or other new particles? Does it decay invisibly, and if so, can we determine what?
So even if discovered right away, making a theory for Z’ will take years…
Lesson: expect to do model-building armed with fragmentary information
Everywhere at LHC: Jets, Jets, Jets
Not all LHC events make
*hard* (pT > 100 GeV) jets
Still the probability of a
pT ~ 20 GeV jet is very high
But what *are* jets?
Naïvely: quarks, antiquarks, gluons
produced in scattering turn into jets because of confinement, hadronization
D0 Dijet Event
Everywhere at LHC: Jets, Jets, Jets
Not all LHC events make
*hard* (pT > 100 GeV) jets
Still the probability of a
pT ~ 20 GeV jet is very high
But what *are* jets?
Naïvely: quarks, antiquarks, gluons
produced in scattering turn into jets because of confinement, hadronization
D0 Dijet Event
WRONG
ZZ
quarks
e+e- quark-antiquark
ZZ
gluons
e+e- quark-antiquark
Lack of particle states in an interacting QFT Once produced, a quark or gluon immediately begins to radiate
Nearly massless quarks Spin-one radiation patterns
The radiation is dominantly collinear (along direction of motion) Or soft (low energy, and subject to destructive interference)
Approximate conformal invariance The process is scale invariant and forms a fractal pattern
Weak (but nonzero!) ‘t Hooft coupling (sNc) The angular width of the fractal is small if the ‘t Hooft coupling is small
Jet, pre-confinement: a narrow fractal distribution of (mostly) gluons a fundamental object in an interacting gauge theory
Note this requires resummed perturbation theory
ZZ
gluons
e+e- quark-antiquark
ZZ
e+e- quark-antiquark
flux confined
Jets: role of confinement
Obviously, confinement has a role: turn the fractal pattern of gluons into a jet of mostly mesons, a
few baryons
But in fact confinement in QCD has very little effect and that this is critical for the phenomenon of jets
How are these statements consistent?
To understand this, consider string theory: Suppose I set an open string in motion in a particular state In what circumstances might you directly observe the state at infinity?
If open string coupling go = 0, closed string coupling gc << 1,
then typically string will oscillate, twist off closed strings –
“gravitational radiation”
Initial state scrambled
If open string coupling go << 1, closed string coupling gc = go
2,
then typically string will oscillate, snap into few open strings
Initial state scrambled
If open string coupling goND ~ 1, closed string coupling gc = go
2,
then typically string will Instantly breaks into many pieces
Initial state preserved
ZZ
gluons
e+e- quark-antiquark
ZZ
e+e- quark-antiquark
flux confined
hadrons
ZZ
e+e- quark-antiquark
Strings vs. QCD Flux Tubes Closed string coupling: gc vs. 1 / N2
Open string coupling: go vs. 1 / N Effective open string coupling: gcND vs. F / N
QCD has F = N = 3
If F = 0, no jets!
If 0 < F << N , no jets? Maybe not… Or quasi-jets, but jet momentum not ~ quark/gluon momentum
Light flavors!
Summary of Jets and Confinement N >> 1 and F << N good for non-perturbative aspects of QCD
But the failure of these same conditions allows parton-hadron dualitywhich allows us to precisely test Short-distance QCD scattering, decays of heavy particles (e.g. top), etc. The semi-perturbative process of jet formation
In short: We should not take the hadronic jets of QCD for granted!!
Too few flavors or many colors, confinement ruins jets Too many flavors, no confinement and no hadrons
This is one of the reasons why jets are so ill-defined theoretically [which means there’s more work to do!]
Standard Model Backgrounds Almost every new physics signal has
a large standard model background or a large detector background or both
Experimentalists spend much of their time Measuring backgrounds in data Predicting backgrounds in advance of an analysis Checking backgrounds in course of an analysis
A lot of theoretical calculation and simulation goes into this effort
Backgrounds are huge – though fortunately they are smaller at high energy
Quick Review: Why do backgrounds fall?Backgrounds fall with energy
Quick Review: Why do backgrounds fall?Backgrounds fall with energy
Cross-section formulas - example:
All parton distribution functions fall like a power of x Parton-parton c.m. energy ~ (x1 x2)1/2 (14 TeV) Most parton-parton cross sections ~ 1/Energy^2
Quick Review: Why do backgrounds fall?Backgrounds fall with energy
Cross-section formulas - example:
All parton distribution functions fall like a power of x Parton-parton c.m. energy ~ (x1 x2)1/2 (14 TeV) Most parton-parton cross sections ~ 1/Energy^2
What’s this?
The debris from the proton-proton collision! Unavoidably produced, always there.
The “Underlying Event”!
CMS experiment: Simulated g g Higgs Z Z e+e-+- + underlying event!
CMS experiment: Simulated g g Higgs Z Z e+e-+- + underlying event!
CMS experiment: Simulated g g Higgs Z Z e+e-+- + underlying event!
Can QCD theory of proton structure predict properties of underlying event?!?!?! A challenge to formal theorists!
Since we cannot currently model it, must measure it!
One of first measurements this year (at 10 TeV) and next year (at 14 TeV): the properties of the average underlying event: how many particles? What pT distribution?
Fluctuations in the underlying event are hard to measure – and can mask new physics
All LHC predictions are affected by the underlying event; if underlying events are more accurate than is guessed, it would cause some problems for the experiments
(Also every interesting proton-proton collision will be muddied by 4 – 20 simultaneous and boring ones )
What should theorists calculate?Tree-calculation solved – faster automation current goalBut trees are always ambiguous, so need first quantum correction
Dominant backgrounds are QCD multi-jet events, so most important calculation a theorist can do is pure QCD…? No!
For fixed # jets, many processes contribute # jets often does not equal # external legs Multijet events are poorly measured
Most measurements require at least one lepton or photon – they are “easy” So jets + lepton (i.e., W or Z) or jets + photon are most important
State of art: W + 3 jets [4 in reach?] But note: top-pairs = W + 4 jets already Lots of signals are lepton + 4 jets [SUSY!] Lots of signals are leptons + 6 jets [SUSY!]So there’s a long way to go – HELP!
What (not) to Compute Calculating total cross-sections is easier for theorists
But measuring total cross-sections is all but impossible
Therefore theorists must provide differential cross-sections to allow these effects to be properly modeled Harder for theorists; Analytic answers rare Need to produce a computer program which can compute value of
differential cross-section for a particular final state
Otherwise, experimentalists can only adjust the normalization of the tree-level calculation; shape still tree-level Hope d ~ dtree * (loop / tree)
This can fail badly when looking at tails of distributions…
Unfortunately theory still has a long way to go here
Aside from the fact that calculations are hard, there are deep conceptual problems (not new though, so not easy)
Fundamental problems of perturbation theory: an asymptotic series in a running coupling – essential ambiguities
Radiation of multiple gluons; breakdown of fixed-order perturbation theory resummations in branching processes in initial, final state
Inability to quantify theoretical errors on any given calculation
Example: g g Higgs boson (150 GeV) LO: 15 pb NLO: 25 pb NNLO: 30 pb
There are important, challenging, understudied formal problems in quantum field theory here; they deserve more attention!
Finally -- the Signals of New Physics!
???!!!???
Let’s talk a little about Supersymmetry (SUSY)
a possible solution to the hierarchy problem, yes… a favorite of string theorists popular even 25 years ago, so the detectors were optimized to find it
(along with the Higgs and “technicolor”)
Instructive LHC lessons even if SUSY isn’t found at the TeV scale
Now we have all heard SUSY Missing Energy i.e. Missing Transverse Momentum!and let’s recall why it is true […!...]
SUSY Missing Energy+Jets+LeptonsTrue in the simplest of the minimal SUSY models where SUSY particles are odd under a new Z2 symmetry (“R-parity”)
always produced in pairs decays of SUSY particles always have SUSY particles in final state the lightest one (“LSP”) can’t decay
The lightest one is neutral, colorless, and lives forever
LHC makes colored objects easily, colorless objects not Therefore LHC makes gluinos and squarks most often
if they are not too heavy.
Decaying colored objects must dump color into the final state But the LSP is neutral So the color must exit as quarks or gluons
JETS! Typically high pT
Often a partner of a Z or W is produced This often allows for a lepton or two to be produced as well
SUSY Missing Energy+Jets+LeptonsTrue in the simplest of the minimal SUSY models where SUSY particles are odd under a new Z2 symmetry (“R-parity”)
always produced in pairs decays of SUSY particles always have SUSY particles in final state the lightest one (“LSP”) can’t decay
The lightest one is neutral, colorless, and lives forever
LHC makes colored objects easily, colorless objects not Therefore LHC makes gluinos and squarks most often
if they are not too heavy.
Decaying colored objects must dump color into the final state But the LSP is neutral So the color must exit as quarks or gluons
JETS! Typically high pT
Often a partner of a Z or W is produced This often allows for a lepton or two to be produced as well
Many Models have similar signatures.
…or at least produce MET, high-pT jets and leptons in different combinations.
ATLAS detector: Supersymmetric event with jets, muons and MET – and U.E.
Missing Energy Especially Vague
Almost a useless discovery by itself… and not even clear cut…
If event has missing transverse momentum, only can conclude Something visible was mismeasured, or Something visible went into a crack or near the beam, or Something invisible was created and not observed
But maybe just neutrinos?
If it’s a new neutral particle, that’s great! But hardly SUSY.
We don’t yet know if The particle is a fermion or boson The particle is produced in pairs The particle is stable; lifetime > 10-8 sec, or decays to neutrinos?
Expect long time from first claim of SUSY to convincing evidence!!
Is the MSSM Well-Motivated… ?Minimal SUSY [“MSSM”]:
Superpartners for all known particles Two Higgs doublets, not one.
Supersymmetry is well motivated Stabilizes mW/mPl hierarchy against radiative corrections …As long as mu problem is solved…
Note – size of hierarchy NOT predicted
Minimality is not well motivated Solves nothing Makes theorists feel good – simplicity, beauty, elegance, Occham’s razor
But remember muon, 3rd generation, Z boson…
One should not give these two words equal weight!
Dangers of Minimalism
For a theorist, adding one or two new particles may Leave the main terms in the Lagrangian alone Leave the key mechanism unchanged Make the model look uglier Make the model less predictive (more parameters)So theorists always like minimal models (easier to publish!)
So do experimentalists (… why? …)
For an experimentalist, adding one or two new particles may Change the observable signatures 100 percent Contradict the “lore” as to how to discover Pose enormous challenges unrecognized in minimal model
Modifying the Higgs Sector
A light Higgs boson is a very sensitive creature
New particles in loops can dramatically alter cross-sections, photon branching fraction More scalars can generate mixing of eigenstates, new decay channels, new production
mechanisms.
Consider adding a single real scalar S to the standard model S carries no charges and couples to nothing except the Higgs, through the potential
If <S> = 0, an Invisible Decay
If, <H>=v / √2 , <S>=0,
then S2H2 (v+h)2S2 = v2 S2 + 2v hSS + hhSS
This allows h SS (if mh > 2 mS) with a width ~ 2v2 / mh
This can easily exceed decays to bottom quarks, with width ~ yb2 mh !
So Br(h SS) could be substantial, even ~1 for a light Higgs boson, depending on
But S is stable. There is an S -S symmetry. So this decay is invisible.
Therefore a light Higgs could be essentially invisible! (its existence might be inferred in VBF or diffractive Higgs production, with difficulty.)
If <S> ≠ 0, a second ‘Higgs’
If <H>=v / √2 , <S> = w / √2, S = (w+s) / √2 , then
S2H2 (v+h)2(w+s)2 = v2 s2 + w2 h2 + 4vw hs + 2v hss + 2w hhs + hhss
new mass terms and a mixing term, plus cubic, quartic couplings
Thus we have two eigenstates with masses m1 , m2
Both eigenstates couple to WW, ZZ, bb, gg, , through their h component;
If <S> ≠ 0, a second ‘Higgs’
So there are two scalar particles that can be produced in gg collisions
And both decay to usual Higgs final states, via their h component --- thus
1has same branching fractions as an SM Higgs boson of mass m1
2 has same branching fractions as an SM Higgs boson of mass m2
EXCEPTION: if m1 > 2 m2, then a new decay channel opens up:
122(bb)(bb), (bb)(), ()()
These exotic final states can occur in many models; recent interest, since a light Higgs with these decay channels can escape LEP bounds.
… it was just one little particle …Thus,
one additional particle can ruin your whole day
And it’s not even that unmotivated – in the first case it is a simple dark matter candidate.
At least we know about this one. It’s the particles we haven’t thought much about that could really
hurt us. We have to keep our eyes open.
We should always be very suspicious of potential Cultural Bias: The culture of theorists always prefers minimal models. Nature may not share this bias.
What does String Theory predict?Hard to find string models without extra matter!
“Millions of models without chiral exotics” [D.Luest’s talk] But these models typically have vector exotics, extra gauge sectors
what are their masses? how do they couple to SM fields?
Often some of vector matter is massless until a symmetry is broken This breaking scale, like weak scale, can naturally be set by SUSY breaking
Thus string theory suggests (to me, anyway) Non-minimal particle content
at or near the TeV scale, coupled to us with 1/TeV-scale interactions.
New heavy charged or colored particles (m > 100 GeV) New heavy or light neutral particles (m > 10 MeV?)
Hidden Valleys: a Subclass of Hidden
SectorsOf course, a hidden sector could be … well … hidden
Producing particles in such a sector could lead to only MET
To infer the structure of the hidden sector would require studying the distribution of MET and accompanying jets/leptons/photons
This would be exciting but very difficult and ambiguous
But it is also possible for a hidden sector NOT to be hidden at all !
Such is the case of a hidden valley
Hidden Valley Scenario (w/ K. Zurek)
A scenario: A Very Large Meta-Class of Models
Basic minimal structure
Standard ModelSU(3)xSU(2)xU(1)
Communicator
Hidden ValleyGv with v-matter
hep-ph/0604261
A Conceptual DiagramEnergy
Inaccessibility
Hidden Valleys and high multiplicity Hidden sector particles may decay visibly, producing 2 or 3 SM particles each
Hidden sectors have their own interactions which can lead to decays or other processes that multiply the number of hidden particles
Let’s look at one example [simply for illustration – other examples can work very differently]
High-multiplicity final states have been considered Pairs of top-antitop pairs Various SUSY decays to 12 particles Black holes
But with the exception of the latter these are not the discovery channels
And black holes have large cross-sections – hidden valleys often don’t
Z’Z’
v-quarks
Analogous to e+e- hadrons
q q Q Q : v-quark production
q q Q Q
qqQQ
qq QQ
Z’Z’
v-gluons
Analogous to e+e- hadrons
q q Q Qv-hadrons
Analogous to e+e- hadrons
Z’Z’
q q Q Q
Analogous to e+e- hadrons
qq QQZ’Z’
qq QQ
v-hadrons
But some v-hadrons decay in the detector to visible particles, such as bb pairs, qq pairs, leptons etc.
Some v-hadrons are stable and therefore invisible
A rare Z + many jets event?
Or an exotic decay of a heavy resonance?
And what if F is not ~ N ?
Exotic decays Above: a Z’ model with an exotic decay
Exotic decays could appear for other particles Higgs (up to 100% if light) Other neutral scalars (up to 100%) Lightest standard model superpartner (100%)
And other new dark matter candidates in other models Rare W, Z, top decays
These can be difficult to discover if they dominantly involve jets, have few leptons/photons
The phenomenology of Higgs, or SUSY, or Extra Dimensions, etc., can be altered 100%.
Or the effect could be subtle, but no less theoretically important
Hidden Valleys and long lifetimes
New neutral particles with mass m < TeV scale coupled to us by interactions at scale M ~ TeV scale have long lifetimes
~ m5/M4 [ dim 6 operator] inside detector for m > 1 GeV ~ m7/M6 [ dim 7 operator] inside detector for m > 10 GeV ~ m9/M8 [ dim 8 operator] inside detector for m > 100 GeV
Smaller masses most decay outside detector
Many hidden sectors will have several stable particles with different masses, couplings, approximate conservation laws… Consider for example QCD! Many different long lifetimes… Easy to extend lifetime by conservation law (e.g. helicity suppression)
Hidden sectors may appear along with new heavy metastable charged particles too (vectorlike exotics)
q q Q Q
Analogous to e+e- hadrons
qq QQZ’Z’
qq QQ
v-hadrons
But some v-hadrons decay in the detector to visible particles, such as bb pairs, qq pairs, leptons etc.
Some v-hadrons are stable and therefore invisible
Clearly something new!
Or…
Discovering long-lived particlesATLAS and CMS were not built to find long-lived particles
Discovering long-lived particles – especially those decaying in flight to jets – is a complicated experimental analysis
The backgrounds don’t come from the Standard Model – they come from the detector
Warning: the ATLAS, CMS trackers are not as passive as would be ideal. Every jet of pions will contain One or more pion-tracker interaction One or more photon e+e- conversion
So there are lots of things that look like sprays of particles…
LHCb (somewhat accidentally) may have a better design:a larger matter-free region near the collision pointCould they be the first to discover new physics!? Even the Higgs?!
Clearly something new!
Or is it a rare event with many pion-tracker interactions?
Summary No matter how hard you think the LHC experiment is, it’s harder
Drowning in Data Incomplete Theory Looking for Gold in Golden Sands Many Years to Determine Particles and their Properties
There is still a lot of deep theoretical work to do for the LHC Jets – can the theory be advanced? Underlying Event – can it be treated properly? Standard Model Backgrounds – techniques, mathematics
Let’s keep our eyes (and those of our colleagues) wide open Minimality is not motivated and is a truly dangerous bias Non-minimal models can and do shake the assumptions that
underly the automated and human data analysts
A Theorist’s Worldview Heaven
The essential properties of the universe are simple and logical, and within our grasp.
All particles are well-motivated by basic principles All dynamical mechanisms are minimal and elegant With enough intelligent reasoning and a few more hints,
theorists can soon deduce the structure of the laws of nature Hell
The essential properties of the universe are complex and we have not yet even begun to understand their logic, if any.
Some particles are just there; they are not motivated by any theoretical requirement.
Most dynamical mechanisms are non-minimal and baroque Theorists are far from determining the principles, if any, that
govern the laws of nature, and therefore far from guessing what they are.
An Experimentalist’s Worldview Hell
The essential properties of the universe are simple and logical, and within our grasp.
All particles are well-motivated by basic principles All dynamical mechanisms are minimal and elegant With enough intelligent reasoning and a few more hints,
theorists can soon deduce the structure of the laws of nature Heaven
The essential properties of the universe are complex and we have not yet even begun to understand their logic, if any.
Some particles are just there; they are not motivated by any theoretical requirement.
Most dynamical mechanisms are non-minimal and baroque Theorists are far from determining the principles, if any, that
govern the laws of nature, and therefore far from guessing what they are.