Light Stop Notes

Light Stop Notes
by Flip Tanedo
Abstract
These are notes on light stops. Usually I hand write these notes, but since there are multiple
references it’s easier to keep track of them in a single document.
1 Motivation
We would like to explore the phenomenology of models with light stops. Why?
1. Experimental constraints on degenerate squarks. Assuming degenerate squarks
(e.g.invoking flavor-universal soft terms to avoid flavor-changing neutral current bounds)
leads to a squark mass scale of over 1 TeV (in fact, maybe more like 1.5 TeV; see 1102.5290).
[Image: see H. Bachacou talk at Lepton-Photon ’11.] However, if we only assume flavor
universality in the first two generations, we can still avoid FCNC bounds while
2. Natural SUSY. The third generation is special. It’s the stop loop that cancels the top
loop that is the dominant contribution to the quadratic cutoff-dependence of the Higgs mass.
As discussed below, there are only a few sparticles which you really need to be accessible
at the early LHC, and the stop is the only (and most important) squark on this list.
3. Little Hierarchy problem. This is again saying the same thing slightly differently: naive
application of experimental bounds suggest that the SUSY scale must be above the 1 TeV
scale; if this is so, then why is the Higgs mass below the TeV scale? As explained above,
the point is that while the rest of the SUSY spectrum may be heavy, a light stop can
accommodate a light-ish Higgs without too much tuning. The stop is special: it has O(1)
couplings to the Higgs (Yukawa) as well as strong couplings to the color sector.
• 1% tuning implies stops below 1 TeV. Other sfermions can be five times heavier
without costing any fine-tuning.
4. Light stops can be generated from RG running alone. See Stephen Martin. This is
due to the negative contribution to the running coming from the Yukawa interaction.
• e.g. light stops through gauge mediation. Usually we say gauge mediation is flavor
blind, but recall Yael’s seminar talk 1103.0292, can use messenger–matter couplings
to give flavor-dependence. General gauge mediation gives a tool for considering a
stop NLSP (gravitino LSP since this is a low scale mediation).
5. EW-scale baryogenesis. The matter–antimatter asymmetry (“baryon asymmetry of the
unvierse”) can be explained through the CP phases in the MSSM that aren’t in the SM.
Requires baryon number violating processes to be out of equilibrium at nucleation tem-
perature, which in turn requires the (sphaleron energy/critical temp) to be smaller than
the expansion rate of the universe. The leading contribution to this ratio depends on the
coupligns of light bosons to the Higgs. 0809.3760.
• Cultural note: EW baryogenesis is one popular option for the baryon asymmetry.
Others include leptogenesis and Affleck-Dine.
• However, DC et al. seem to have done a lot to exclude this story: 1203.2932. (See
their intro for references and reviews.) Since the Higgs is 125 GeV, EW baryogenesis
in the MSSM would require (as per the Chicago analysis) a TeV scale left-handed
stop and a 100 GeV right-handed stop. [By the way, DC continues to use italics to
emphasize sentences in his papers.]
6. Interesting phenomenology that is different form other squark searches.
Other ways out:
• Some other protection mechanism is operative at the TeV scale and SUSY may live at the
10 TeV scale. e.g. “Schlittle Higgs” (SUSY little Higgs).
1

• 125 GeV “Higgs” is actually a (techni-)dilaton.
2 EffectiveNaturalMinimal SUSY
The main idea is to consider the ‘minimal’ SUSY spectrum that should be accessible to the early
LHC in order to solve the Hierarchy problem. The main ingredients are:
• Light third generation squarks (light-ish stops) at a few hundered GeV. These run in the
Higgs loop and cancel the third generation Standard Model fermion loops.
• A not-too-heavy gluino (less than a couple of TeV) since these contribute to the Higgs mass
at two-loop level. Note that the gluino tends to want to pull the stop mass up due to its
large RG coefficient.
• Once you set the gluino mass, unification fixes the electroweak-ino masses to be in the few
hundred GeV range (since the spectrum is given by the ratio of gauge couplings).
• The Higgsinos are also in the same light-ish ballpark since they get their masses from the
µ term, the same source as the Higgses. What about soft breaking terms? You can’t write
any by symmetries.
2.1 Main References
• Papucci, Ruderman, Weiler, 1110.6926. Natural SUSY Endures.
• Brust, Katz, Lawrence, Sundrum, 1110.6670. SUSY, the 3rd Gen, and the LHC.
• Kats, Meade, Reece, Shih, 1110.6444. The Status of GMSB After 1/fb at the LHC.
• Older papers with the same basic ideas:
◦ 1007.3897. Effective SUSY at the LHC
◦ hep-ph/9607394. The More Minimal Supersymmetric SM. (Cohen, Kaplan, Nelson),
seems to be the original reference.
2.2 Review: Higgs in the MSSM
∆mh
2
(stops) =
3
4π2
cos2
αyt
2
mt
2
ln
mt1˜ mt2˜
mt
2 . (small mixing α)
Alternate derivation: integrate out stops completely and check their contribution to the Higgs
quartic interaction. In the decoupling limit mA0 ≫ mZ, then the Higgs mass can saturate bounds:
mh
2
= mZ
2
cos2
2β+(loops).
Including the effect of stop mixing, we have
mh
2
= mZ
2
cos2
2β+
3
4π2
sin2
βyt
2

mt
2
ln
mt1˜ mt2˜
mt
2 +ct˜
2
st˜
2
mt2˜
2
− mt1˜
2
ln
mt2˜
mt1˜
2


+
3
4π2
sin2
βyt
ct˜
2
st˜
2
mt
2

 mt2˜
2
− mt1˜
2 2
−
1
2
mt2˜
4
− mt1˜
4
ln
mt2˜
mt1˜
2


Note that since yt ∼mt, the prefactor yt
2
mt
2
∼mt
4
, which is usually how this dependence is written.
2.2.1 Comments on the calculation, effective potential method
It may be helpful to understand the origin of the terms above. From the Sparticles book (section
10.6) we note that there are three ways to calculate these corrections.
1. Direct calculation from diagrams
2 Section 2

2. Use the RG flow from the SUSY scale down to to the weak scale. See hep-ph/9308335.
3. Effective potential.
The effective potential method is not as numerically accurate, but is pedagogically more clear. The
loop level effective potential is
V
(1)
(Q) = V (0)
(Q) +
1
64π2
STr M4
(h) ln
M2
(h)
Q2
−
3
2
.
Here V (0)
is the tree level potential with renormalized couplings at the scale Q. Note that the
supertrace vanishes in the supersymmetric limit, so as expected the potential is protected by SUSY
and is corrected only to the extent to which SUSY is broken. The factor of 3/2 is a relic of MS
renormalization, see Peskin (11.78); e.g. it is associated with the freedom to add a finite counter
term that is a quartic polynomial in h (see Hugh Osborn’s AQFT notes).
We now focus on the contributions from the stops to the Higgs potential. In the absence of
stop mixing (which we address below) and assuming equal soft masses m˜ for the stops, we have
the following field-dependent masses:
mt
2
(h) = yt
2
|hu|2
mt˜
2
= m˜2 + yt
2
|hu|2
,
where we drop O(gEW
2
) contributions from D-terms. The point now is to consider the contributions
to the supertrace, remembering the weighting by the degree of freedom for a particle of a given
spin. The top is a Dirac fermion with four degrees of freedom, each stop is a complex scalar with
two degrees of freedom each. There is also a color factor of three. The loop-level correction is thus
∆V (1)
=
3
16π2
mt˜
4
ln
mt˜
2
Q2
−
3
2
− mt
4
ln
mt
2
Q2
−
3
2
.
The next step is to minimize the loop corrected potential. Since ∆V (1)
only involves hu, the tree-
level constraints on mHd are unchanged (see the Appendix):
mHd
2
+ |µ|2
− b tan β+
mZ
2
2
cos 2b = 0.
On the other hand, the (2,2) element of the mass matrix for the CP even Higgses is changed. You
can go ahead and take second derivatives using Mathematica (and be careful with factors of 3
which weren’t working out in my code), and you end up with a correction to the CP even Higgs
mass matrix (on the 2–2 element) of
3
4π2
yt
2
mt
2
ln
mt˜
2
mt
2,
where it is also common to restore sin β =vu/v dependence through mt = ytvu. You can go through
the diagonalization, but the punchline is that now the upper bound on the Higgs mass is
mh
2
< mZ
2
cos2
2β +
3
4π2
yt
2
mt
2
ln
mt˜
2
mt
2 .
This alleviates the tree-level bound that the Higgs shouldbe lighter than the Z.
One curiosity is that the correction to the Higgs mass is only logarithmically dependent on
the stop mass, whereas one might naively expect a power law dependence, e.g. from explicit mass
insertions of the soft terms required to bypass the vanishing supertrace. This is resolved by the
observation that the shift in the tree-level parameter mH2
2
is indeed proportional to mt˜
2
− mt
2
.
One way to see this is to consider ∂V /∂Hu with the loop level effective potential. This gives an
expression for mHu
2
, analogous to the unchanged tree-level mHd
2
expression above. Now, however,
there is a correction of the form
−
3yt
2
16π2
2mt˜
2
ln
mt˜
2
Q2
− 1 − 2mt
2
ln
mt
2
Q2
− 1 .
EffectiveNaturalMinimal SUSY 3

Thus even though the shift in mh appears to only be logarithmically dependent on SUSY breaking,
a large splitting between the stop and top would lead to a fine-tuning in mH2
2
. (At least I think
this is the logic? See the Sparticle book section 10.6.)
Another remark on the scale dependence: our correction to mh
2
is independent of ln Q. This
dependence cancels against the Q dependence of running quantities in the tree-level potential, as
is generally true in effective potential calculations. The Q dependence of the loop-level piece is
∂∆V (1)
∂ ln Q
= −
3
8π2
m˜2 yt
2
|H2|2
+
1
2
m˜ 2 .
The first term cancels the Q dependence in mH2
2
|H2|2
and the second term is a field-independent
contribution to the cosmological constant.
2.2.2 Stop mixing
Recall that the off-diagonal terms in the stop mass matrix take the form mt(A + µ∗
cot β). The A
term is a soft-breaking trilinear term analogous to the Yukawa term. (Indeed, it is often normalized
so that A is the rescaling of the coefficient relative to the Yukawa, hence the mt dependence.) The
µ-dependent term comes from the F-term contribution to the potential, where you have |∂W/∂H|2
which contains a mixing term between the µ-term and the Yukawa term.
The additional contributions to the Higgs mass from stop mixing go like products of sines and
cosines of the mixing angle and the difference between the two masses. Explicit formulae can be
found in your favorite references, e.g. Martin’s review (8.1.25 in version 6) or the Sparticle book
(section 10.6). For now we just won’t make a big deal about this.
2.3 The gluino
Stops produced indirectly from gluino decays lead to same-sign dilepton signatures (because the
gluinos are Majorana) or jets + MET. These constrain the gluino to be heaveir than 900-ish GeV.
This, however, is still compatible with naturalness since the gluino only contributes at one loop to
the stop and hence only at two loop to the Higgs.
3 Aspects of the light stop
3.1 The 125 GeV Higgs
The most general Lagrangian for effective Higgs couplings is given in 1002.1011. It takes the form...
L =
1
2
(∂h)2
− V (h) +
v2
4
(DΣ†
DΣ) 1+2a
h
v
+ b
h2
v2
+
−miψ¯LiΣ 1 + c
h
v
+ ψRi +h.c.
+cg
αs
4π
h
v
G2
+ cγ
α
4π
h
v
F2
V =
1
2
mh
2
h2
+ d3
1
6
3mh
2
v
h3
+ d4
1
24
3mh
2
v2
h4
This is mean to describe a general situation in which a light scalar h exists in addition to the
vectors and Goldstones of electroweak symmetry breaking. We’ve assumed flavor universality,
but in principle c is a general matrix. The cg and cγ couplings are small and often negligible. We
invoke custodial symmetry so that the Goldstones describe the coset SO(4)/SO(3) which can be
described using
Σ = eiσaπa
/v
4 Section 3

Working at a sufficiently low energies relative to the strong scale we are allowed to perform this
derivative expansion. In the Standard Model, a = b = c = d = 1. This allows us to repackage the h
into the usual nonlinear expansion of H,
U = 1 +
h
v
Σ.
As described in hep-ph/0703164, the effective couplings can be interpreted in terms of the Gold-
stone boson equivalence theorem. a controls 2 → 2 longitundinal boson scattering, A ≃
s
v2 (1 −
a2
). Pertubative unitarity is satisfied when b = a2
. The amplitude for longitudinal vector boson
scattering into fermions is
A(ππ → ψψ¯) =
mψ s
√
v2
(1 − ac),
since a = b = c = 1 in the SM, this is weakly coupled at all scales. The pseudo-Goldstone Higgs
models predict values of the effective couplings (model-independent-ish) which depend of v/f.
3.1.1 The role of h → γγ
We note that Γ(h→ γγ)/Γ(h→ γγ)SM ∼ a2
, coming from the W loop. Note, however, that there is
also a contribution from the top loop which is ∼5 times smaller and dependent on c. Note, also,
that the production mechanisms depend on these parameters, in particular gluon fusion depends
on c while vector boson fusion depends on a.
3.1.2 Calculation via low energy theorems
The loop induced h→ VV interaction can be calculated in a simple way using the Higgs low
energy theorems which relate this rate to the β-function of the relevant gauge group. This is
intuitively clear since at low energies the h can be thought of as a slowly varying excitation of the
vev v, and thus one can just look at the VV loop-level interaction (given by the β function) and
replace v → v + h.
For a pedagogical description, see “A Primer on Higgs boson Low-Energy Theorems” by Dawson
and Haber1. For references to original work and an application to the composite Higgs model, see
1206.7120.
This means that the contribution of stops to the h→ γγ rate is given by the MSSM QED β
functions, which should be readily available in, say, the SUSY primer. (But this doesn’t seem to
include threshold effects—i.e. it seems that we need the Higgs contribution to the stop mass.)
3.1.3 125 GeV Higgs status with respect to SUSY
The best global fit is (cited from 1208.2673)
cg
2
=
Γ(h→ gg)
ΓSM(h→ gg)
≈ 0.7 cγ
2
=
Γ(h→ γγ)
ΓSM(h→ γγ)
≈ 2.1.
What does this mean with respect to the MSSM?
• Stau loops can increase cγ without affecting cg. Carena et al. 1112.3336, Wang et al.
1207.0990, Giudice et al. 1207.6393. See also a more recent Chicago group paper on light
staus: 1205.5842.
• Once you increase cγ, you can then reduce both cγ and cg simultaneously with light stops.
3.2 Light stops from RG
[From Stephen Martin.] The first and second generation sfermion masses run according to
16π2 d
dt
mi
2
= −
a=1
3
8Ca(i) ga
2
|Ma|2
+
6
5
Yg1
2
Tr[Yi m2
]
1. http://www.osti.gov/bridge/servlets/purl/5960988-Dk3lmn/5960988.pdf
Aspects of the light stop 5

where Ca(i) are the Casimir terms for each gauge group. Observe that the |Ma|2
(gaugino) terms
are negative. This means that the sfermion masses mi grow in the IR; we can heavy (first and
second generation) sfermions automatically due to the gaugino contribution to the sfermion masses.
For the third generation, however, things are different due to the size of the Yukawas and the
soft trilinear couplings (A terms). These enter in particular combinations that we call
X = 2|y|2
mH
2
+ mD
2
+ mf˜3
2
+ 2|A|2
where H refers to the relevant type of Higgs (Hu for Xt, Hd for Xb and Xτ), mD
2
refers to the
relevant doublet, mf3
2
is the relevant right-handed sfermion mass, and A is the relevant trilinear
soft term. The Xt,b,τ tend to be positive.
Note that both the Yukawa and the soft trilinear term generate stop mixing upon electroweak
symmetry breaking.
3.2.1 Effect on the Higgses
The Higgs RG equations are
16π2 d
dt
mHu
2
= 3Xt − 6g2
2
|M2|2 −
6
5
g1|M1|2 +
3
5
g1
2
Tr[Yi m2]
16π2 d
dt
md
2
= 3Xb − 6g2
2
|M2|2
−
6
5
g1|M1|2
+
3
5
g1
2
Tr[Yi m2
] +Xτ.
Since the Xs are positive, they encourage the Higgs masses to decrease in the IR. Since mHu
2
is
sensitive to Xt, we would expect the up-type Higgs mass to run lighter... even negative. Indeed,
this is precisely what we need for electroweak symmetry breaking. Huzzah.
3.2.2 Effect on the third generation
The stop soft masses run according to:
16π2 d
dt
mQ3
2
= −
33
3
g3
2
|M3|2
− 6g2
2
|M2|2
−
2
15
g1
2
|M1|2
+
1
5
g1
2
Tr[Yi m2
] + Xt + Xb
16π2 d
dt
mU3
2
= −
32
3
g3
2
|M3|2
−
32
15
g1
2
|M1|2
+
4
5
g1
2
Tr[Yi m2
] + 2Xt.
First we might worry that the stop masses run negative. Note that the X terms appear with the
same sign as in the Higgs mass running, but with smaller coefficients. Further, unlike the Higgs
the squarks get a large negative contribution from gluinos. Thus its reasonable to believe that the
Higgs can get a vev while the squarks do not.
4 Phenomenology: searches for stops
The light stop scenario avoids several “typical” collider searches for SUSY (e.g. large MET).
4.1 Stop simplified model and pre-LHC analyses
Matt notes that he and Patrick were the first to advocate studying simplified top partner models
in hep-ph/0601124. Other papers:
• Ditops + MET: 0803.3820
• HEPTopTagger, an algorithm for top tagging which can be used to reconstruct stops.
1006.2833
• Boosted semileptonic tops: 1102.0557
4.2 Stop production
• Direct stop production, all other sparticles decoupled. (See Beenacker)
6 Section 4

• Production through heavier gluinos or squarks. (Cascades)
See hep-ph/9710451 (Beenakker et al) for NLO cross sections for stop production using Prospino.
4.3 Stop decay
4.3.1 Stop at a bump classification of experimental bounds (Maryland group)
RPV or no? LSP or no?
• Neutralino LSP? (CDF stop search)
• Sneutrino LSP? (D0 stop search)
• Graviitino LSP? (only recent dedicated searches, ongoing downstairs)
One constraint on models is that the stop branching ratio cannot be so small that it leads to a
displaced vertex. This is especially constraining on stop LSPs in RPV models where the stop
decays through a coupling which is small due to experimental constraints on RPV. As a good rule
of thumb, we need 1/Γ 1 mm.
4.3.2 Light Stop Signs classification (Harvard group)
Focus on t˜→ bW+
χ0
, divided into three categories:
• Three body decay. Stops are so light that its decay must go throughan off-shell top. The
mℓb and mT distributions will be very different from top decay and may be a good handle.
(See Peskin and Chou.) unfortunately, this region suffers from low acceptance, but “probably
deserves a closer look in the future.”
• Two body decay. For stops much heaver than the top, the top decay product may be on
shell and the neutralino/gravitino may carry large MET. While a simple MET cut helps,
there are more sophisticated ways to use this. MET in the top background comes from W
leptonic decays; thus one can play with transverse mass cuts on semileptonic decays or a
cut on mT 2. See references in Matt’s paper. (e.g. 1205.2696, stop searches in 2012)
• Stealth stop. For very light LSP, e.g. gravitino in low scale SUSY mediation, one can have
mt˜ just above mt so that the gravitino momentum is negligible. These are effectively “one-
body” decays that are very hard to separate from top pair production. Unlike compressed
spectra scenarios, events do not become more distinctive when recoiling against an addi-
tional hard jet. (See “Stealth SUSY Sampler.”)
4.4 Cut and count
The simplest searches are “cut-and-count”: pick a sufficiently distinct signal with cuts that suffi-
ciently reduce SM background and just count the total number of events. Usually there are so few
events that one cannot plot a kinematic distribution (which would provide an additional handle
for discriminating against the SM), so it is critical to have the background under control.
Two good handles for this are:
• Multi leptons: the SM doesn’t like to produce lots of hard leptons. 4 lepton events are
exceptionally rare.
◦ ATLAS: ATLAS-CONF-2012-001
◦ CMS: SUS-11-013
◦ In addition to real lepton production (e.g. from weak gauge bosons), backgrounds
come from fake leptons. For example, b-jets produce leptons wich can accidentally
pass the isolation efficiencies at ar ate of 1.5%. You also have to worry about internal
and external photon conversions when a photon produces a hard and soft lepton.
Phenomenology: searches for stops 7

• Same-sign leptons: for example, from the pair production of Majorana particles like the
gauginos. For example, see Allanach and Gripaios: 1202.6616.
◦ CMS: 1205.6615, SUS-12-017-pas
4.5 Stop NLSP
Katz and Shih explored the phenomenology of a stop NLSP that decays to a gravitino and a top
(which may be off shell) 1106.0030. (This scenario was studied 13 years earlier by Peskin in hep-
ph/9909536) Thus one looks for t˜→ Wb G˜. The relevant couplings can be obtained by looking
at the Goldstino couplings to the supercurrent, for details, see the appendix of their paper. They
focus on prompt decays and ignore cases where the stop may have a displaced vertex or may even
be reasonably long-lived.
4.5.1 Stop–Goldstino couplings
From Giudice & Rattazzi (hep-ph/9801271) eq. (3.5)
L = −
k
F
ψLγµ
γν
∂νφ −
i
4 2
√ λ a
γµ
σµν
Fνρ
a
∂µG˜ + h.c.
where the first term is the χSF coupling and the second is the VSF coupling. Actually: the ∂νφ
should be replaced by Dµφ, this will give a t˜bW +
G˜ coupling.
∗ Actually, and even better expression comes from Kats & Shih (1106.0030), see their Appendix A:
Lt˜tG˜ =
i
F
∂νt˜∗
∂µG˜γν
γµ
(ct˜PL + st˜PR)t + h.c.
Lt˜WtG˜ =
i
2
√
F
gWν
+
t˜∗
∂µG˜γν
γµ
PLb + h.c.
Next use γµ
∂µG˜=0 (on-shell condition) so that ∂µG˜γν
γµ
→ 2∂ν
G˜ .
It’s worth noting that this has not yet been implemented in the latest FeynRules models. Matt
Reece has a UFO model on his website: http://www.physics.harvard.edu/~mreece/stopnlsp/,
and the Shih/Katz group also has their own proprietary code. I’ve also developed my own Feyn-
Rules model. This should be standardized in the near future.
The stop decay rate is:
Γ(mt˜ < mt) ∼
α
sin2 θW
(mtR − mW)7
128π2mW
2
F2
Γ(mt˜ > mt) ∼
mt˜
5
16πF2
1 −
mt
2
mt˜2
4
.
The difference comes from whether the top decays through the intermediate top diagram or through
the four-point interaction. Restricting to prompt decays for mt˜ < mt requires F to be as small as
possible, e.g. F ∼(10TeV)2
. Longer lived stops, e.g. collider-stable, are also possible, but are not the
main subject of this paper. The best limit currently comes from ATLAS, mt˜>300 GeV 1103.1984,
though there are also nontrivial bounds on this scenari from BBN (hep-ph/0701229, 0811.1119).
4.5.2 General Stop NLSP phenomenology
To leading order the stop decay amplitude depends on the stop mass and its mixing angle with
the heavy stop. The contribution from virtual charginos and sbottoms also contribute, but it turns
out that these do not strongly affect the phenomenology, hep-ph/9909536 (Chou & Peskin).
For mt˜ < mt, the four-point and three-point (with subsequent top decay) decay diagrams are
comparable, whereas for higher stop masses the intermediate top diagram dominates so that the
decay effectively becomes two body: t˜→ G˜t.
8 Section 4

4.5.3 Kinematic distributions
As before, the main background comes from ditop. Katz and Shih give kinematic distributions
for sample stop masses (from 120 to 200 GeV) relative to the ditop background for:
• Lepton pT
• b-jet pT : stops lighter than the top give softer b-jets.
• MET (all hadronic, lepton+jets, dilepton)
• mT : stops have a high mT tail in the leptons+jets channel.
• HT (scalar sum of jet pT ): somewhat lower values for HT for stops in leptonic channels
• mℓb (invariant mass of lepton and b-quark at parton level) for diﬀerent lepton pairings. Note
that this is not currently used in any exisiting analyses (as of the writing of that paper) and
can be a useful distinguishing variable.
I really can’t tell by eye if these distributions are discriminating... doesn’t look like it, and anyway
it’s clear that there are limits where the stop looks very much like a top.
For stops lighter than the top, the b-jets are much softer due to kinematics. This is due to a
Jacobian peak in pbT in the top frame when there is an on-shell top. On the other hand, for the
3-body decay (from the 4-point interaction), the distribution is diﬀerent.
4.5.4 Comments from Peskin and Chou (’99)
Here’s a plot from Peskin and Chou hep-ph/9909536 with the distribution of mℓb. Plotted are a
stop with mass 130 GeV (dotted) and 170 GeV (dashed) for a scenario where the lightest chargino
is light (m2 ∼ 200 GeV) and wino-like. The solid line is the meb spectrum for a top quark. The
distribution has good discriminating power against tops.
Unfortunately, the other scenarios that they present (Higgsino like lightest chargino, mixed
lightest chargino, heavy lightest chargino) do not give as much separation between the distribu-
tions.
Another handle they present is the longitudinal W polarization. Top decays tend to be highly
longitdinally polarized. In the SM,
r=
Γ(W0)
Γ(Wall)
=
1
1 + 2mW
2
/mt
2 ≈ 0.71.
In the light stop scenario, this ratio is a function of the chargino mixing matrices. The point,
however, is that one can variations between r = 0.4 and 0.9.

4.5.5 Searching for stop NLSP, part I: Ditop analyses
The ﬁrst class of searches are those where the stop events are hidden in the ditop data. The hope
is that we would just have to see a deviation from what we expect. In the next sub-sub-section we
discuss explicit stop searches. The hidden-in-the-top-data searches can be categorized based on
the W decays:
• Dileptonic (5%): 2 leptons, 2 jets, MET. Cleanest, S/B∼2. However, very small branching
ratio so that the statistical uncertainty is roughtly of the same size as the systematic.
• Lepton + jets (35%): Larger branching ratio, but large W+jets background—require b-
tag to beat this down.
• Jets + MET (10%): zero lepton events where at least one W decays to τν with τ →hadrons.
QCD background is hard. Most stops fall in this category due to gravitino. Cut and count
is no good against the QCD background, but more sophisticated neural network techniques
may be helpful. The CMS group is working on something along these lines with a giant
maximum likelihood calculator.
• Fully hadronic (50%): QCD background is hard.
A typical set of requirements (lepton, 4+ jets, MET) lead to S/B ∼4 with systematic uncertainty
2 × statistical. Katz and Shih explain that many additioanl top quark properties are not as useful
as one would hope since they use matrix elements or neural networks to kill background and
are thus likely to discriminate against stop events as well. Other properties require an accurate
reconstruction of the ditop event so that the stop events (with missing momentum due to the
gravitino) tend to be rejected as background. More practically, since the experimentalists’ codes
are proprietary, it is inaccessible to phenomenologists.
4.5.6 Searching for stop NLSP, part II: explicit stop searches
The GMSB-inspired stop NLSP scenario presents some opportunities for novel experimental
searches. Some of these searches can be reinterpreted from existing ditop-like signatures.
• Stop decay via virtual chargino, t˜→bℓ+
ν˜. 1009.0266, 0811.0459, D0 Note 5937-CONF,
1009.5950. The neutrino can either be the LSP or can decay invisibly the a neutrino and
a neutralino/gravitino. Signatures are similar to the t˜→ W +
b G˜ decay dilepton channel.
• Another ditop-like signature is t˜→ b χ˜1
+
→ W (∗)
νχ˜1
0 .
• A similar signature is t˜→b(χ˜1
+
→ℓνχ˜1
0). 0912.1308, CDF Note 9439, FERMILAB-THESIS-
2010-16. CDF has an algorithm to reconstruct the stop mass which helps discriminate
against top events. However, there are four invisible particles and two particles (chargino
and neutralino) with unknown mass, the reconstruction is imperfect. With an assumption
for the chargino mass, there is a stop-mass-like quantity that they can construct which has
a distribution which is peaked roughly at the stop mass.
• Other searches: b-jets + MET 1103.4344, search for heavy top partner ATLAS-CONF-2011-
036.
4.5.7 Searching for stop NLSP, part III: other searches
We now consider some additional regions of stop NLSP parameter space.
• Displaced decays. The stop can be naturally long lived. For decays between 100 µm and
0.5 m from the interaction point, a displaced vertex can be tagged. One can avoid heavy
ﬂavor displaced vertices (background) by looking at distances above 1 cm, for which this
may be uased as a discriminator against the Standard Model. Stops hadronize, possibly into
charged particles which leave tracks with large amounts of energy lost through ionization.
• Stoponium. One can form a near-threshold stop bound state. The annihilation rate into
gg can be of order Γ−1
∼ 10−13
m, for nearly degenerate top and stop. Stopnoium would
give a diphoton resonance that could allow a precise stop mass measurement in a few years
at the LHC.
10 Section 4

• Flavor violation, SS2L. Stop to charm and neutralino/Gravitino may become a dominant
decay if there are new sources of flavor violation. In this case, the Tevatron could exclude
stops up to 180 GeV if mt˜ − mχ
0
50 GeV. With even small flavor violation, mesino
oscillations hep-ph/9909349 can convert stops into antistops and lead to same-sign dilepton
events with jets and MET. See also 1207.6794 for t→ ch, which may be relevant.
4.6 Bump hunting
Brust, Katz, and Sundrum presented a bump search for RPV models 1206.2353. In these models,
the stop is indeed the LSP, but this decays into two jets via the RPV coupling ∆W ⊃ tRdRdR.
(These RPV couplings individually have to be so small to avoid proton decay and flavor bounds
that it’s sufficient just to turn on one coupling at a time.) This means that you can get signatures
which are devoid of MET and leptons. Existing cut and count searches aren’t optimized for this
sort of signal since they miss out on the dijet decay.
The BKS paper presented an alternate search based on reconstructing the resonances of the
dijet decay of the stops. They consider cascade decays of heavier squarks into the light stops. They
focus on sbottom decays rather than heavy stop decays since the latter decays are more model-
dependent: the heavy stop to light stop-Z and light stop-h decays depend on the left-right stop
mixing. Meanwhile, they focus on: b˜→ W (∗)
t˜. The strategy is:
1. Tag events with two isolated leptons and “moderate” MET (to kill the Drell-Yan dilepon
contribution)
2. Cluster the jets with a large radius. If you miss jet components then you can’t reconstruct
the resonance.
3. If the event contains 4+ jets, then we have to play combinatoric games. Since we assume that
these jets come from the light stops, we can bias our combinatoric searches by minimizing
the difference between reconstructed invariant masses.
Background: top pair production with dileptonic decays (100 × signal), but one can cut away
most of the other backgrounds. So one has to use an additional discriminator: the hardness of the
event. In ditop events, the hardness of the event correlates with the hardness of the leptons and
MET. In the signal, however, there may only be a mild splitting between the stop and sbottom
(by assumption in the model) so that typically one cannot emit an on shell W. This means there’s
relatively low lepton pT and MET even if the overall event is hard. So the simplest thing to do is to
cut the high MET and high lepton pT tail. One can do better, however, by defining dimensionless
variables in where these quantities are scaled by the total event hardness.
4.7 Light [Stealth] Stop Signs 1205.5808
Harvard group. Focuses on the stealth stop window, mt mt˜ 250 GeV, look for new handles on
this difficult region. The difficulty, recall, comes mainly from the fact that these stops decay into
stops and a relatively soft neutralino/gravitino so that these effectively look like top decays.
Note that at 7 TeV, the stop cross section is only a sixth of the top cross section for degenerate
stop/top. This drops rapidly for larger masses. Thus we need as many handles as we can get since
simply measuring the total rate for events passing top selection is not enough.
4.7.1 Spin semi-correlation
The most important fact of the stealthy regime is that stops are spin zero and are produced without
spin correlations. Thus the stop and antistop decays are completely uncorrelated. What does this
buy us?
Suppose we have a sample of ‘top’ events which include stealthy stops. The real top decays are
spin correlated, while the stops slightly wash out this correlation. Melnikov and Schulze 1103.2122
showed that these correlations are stable with respect to NLO corrections, which is a unique
property for our handles of stealth stops.
They focused on looking at, for example, the distribution of ∆φ(ℓ+
, ℓ−
). I’ll not go into this
further since spin correlation techniques are a whole separate bag of worms. See the paper for a
discussion and references.

4.7.2 Stop versus top production
The stop production rate is suppressed relative to the top production near threshold. There are
two reasons for this.
1. gg→tt¯ has t- and u-channel singularities which are regulated by mt. On the other hand,
di-stop production does not have such a singularity. This is because the only diagram which
looks like it could give such a singularity. This can lead to the use of a rapidity gap as a
discriminator. This explains why the gg production channel for tops and stops are different
(68 pb vs 11 pb).
2. Stop production is p-wave in the s-channel (that’s a confusing way of saying it) production
from the qq¯ initial state. This is because the stop final states must carry angular momentum;
the production goes through an s-channel gluon which is spin-1. This means the production
rate goes like β3
(where β is the velocity) near threshold. This suggests looking at higher
pT where stop production might be enhanced. This, however, is less interesting at the LHC
due to the parton PDFs. This explains why the q q¯ production channel for tops and stops
are different. (23 pb vs 1.6 pb.)
Indeed, the stop production cross section from gluinos is given to leading order by:
σ gg→t˜t∗˜ =
αs
2
π
s
β
5
48
+
31m2
24s
+
2m2
3s
+
m4
6s2
log
1 − β
1 + β
where β2
= 1 − 4m2
/s is the subprocess velocity.
4.7.3 Rapidity Gaps
The gluon to (s)top production cross sections are, in the massless limit:
σ(stop) =
5αS
2
π
48s
σ(top) =
αS
2
24s
(t2
+ u2
)
1
tu
−
9
4s2
.
The top singularities are regulated by the top mass so that it ends up being log s/mt
2
enhanced.
You can also see this using MHV amplitudes,
M(1+
, 2−
, 3ϕ, 4ϕ) = ie2 [13] 23
13 [23]
= e2
× phase
M(1+, 2−, 3−, 4+) = ie2 [14] 23
13 [23]
= e2 u
t
× phase.
They note that the top production case corresponds to the “familiar splitting function ameliorating
the ple in a t-channel diagram to the squareroof of a pole in the amplitude.” The two results are
related by a SUSY Ward identity.
They then look at the effect of this on the rapidity distribution ∆y between the top and anti-top.
They noted that the effect of the shape can be mimicked by increasing the renormalization scale.
4.8 Kinematic variables with endpoints for background
“Stop the top background,” Bai, Cheng, Callicchio, and Gu. 1203.4813.
4.9 Tagging boosted tops from stop decay
• 1205.5816: Stolarski, Kaplan, and Rehermann.
12 Section 4

• 1205.2696: Plehn, Spannowski, Takeuchi.
4.10 MET and MT shapes
1205.5805. Alves, Buckley, Fox, Lykken, Yu (Tien-Tien).
5 Review of Work: UV theories
Common theme: the third generation is special through compositeness. Otherwise you have to
figure out a way to explain flavor structure; compositeness does this ‘automatically’ e.g. through
a small splitting in anomalous dimensions at the high scale.
5.1 Supersymmetric warped extra dimension
Uses XD localization and ‘emergent’ (See below) or ‘partial’ supersymmetry hep-ph/0302001.
• General problem: gauginos are external to the strong dynamics (compare to McSSM) and
need to be both light and have supersymmetric couplings to avoid dangerous quadratic
divergences in the Higgs mass. However, the gauginos are part of the elementary sector
(which feels SUSY breaking strongly) and neither assumption holds automatically, requiring
some model building.
• Automaticatically inherits nice features of RS: e.g. hierarchical Yukawas
•
5.1.1 Emergent SUSY 0909.5430
Stringily-motivated; suppose SUSY is high scale but “vestiges” of supersymmetry are ’accidentally’
redshifted down to the TeV scale. In the dual sense, this is emergent/accidental supersymmetry
in the strong sector. SUSY no longer solves the big Hierarchy problem (which is now solved by
warping), but it does address the Little Hierarchy problem.
• Goes on to extend this idea: replace non-SUSY weakly coupled sector with the weakly-cou-
pled sector of split supersymmetry: light gauginos. Suppresses dominant Higgs radiative
corrections in the little Hierarchy.
• Higgs is a composite state whose mass happens to be light compared to the dynamical scale.
(The usual problem in composite models.) Introduces a ‘little’ µ-problem, despite being
technically natural.
• Also became a prototype for his effective SUSY work with Katz et al.
5.1.2 Partially supersymmetric, pseudo-Goldstone Higgs 1004.5114
By Redi and Gripaios. Supersymmetric Randall-Sundrum. Understood to be SO(5)/SO(4) pseudo-
Goldstone Higgs. Maps onto the ‘More Minimal SUSY’ic SM.’ Different from Raman’s model in
that it’s really supersymmetrizing the minimcal composite Higgs model. Thie Higgs only couples
derivatively to the strong sector, Yukawas obtained through mixing with elementary sector.
• Regarding gaugino problem: assume SUSY breaking in elementary sector is large com-
pared to the compositeness scale but soft, so couplings are approximately supersymmetric.
Gaugino masses then require an approximate R-symmetry that is respected by SUSY
breaking.
• Generation of Higgs potential is different from non-supersymmetric case. In minimal com-
posite Higgs, required Higgs potential to come from loop-level couplings to fermions (e.g.
top). However, for the SUSY MCH, a quartic of the correct order of magnitude is auto-
matically generated and mass terms are only generated at loop level when SUSY is broken.
This avoids problems composite Higgs models that have a high compositeness scale.
Review of Work: UV theories 13

5.2 Deconstructed third generation 1103.3708
See Jack’s talk last year. The quiver diagram can be understood as a deconstruction of the general
framework above.
5.3 Composite MSSM 1201.1293
Another handle on strong coupling is Seiberg duality. This is again a variation of the same theme
above.
See my separate notes on McSUSY. For references see arXiv:1106.3074 (Csaki, Shirman,
Terning, “A Seiberg Dual for the MSSM”) and arXiv:1201.1293 (Csaki, Randall, Terning, “Light
Stops from Seiberg Duality”). For background, see my Seibergology notes.
5.4 Flavor mediation 1203.1622
Explicit flavor structure introduced in messengers. For related ideas, see Yael’s seminar talk last
week.
6 Experimental bounds
6.1 General squarks
Searches for first and second generation squarks have been very restrictive. However, as we
explained above, these bounds have essentially no impact on fine-tuning in the MSSM.
6.2 Stops after ICHEP 2012
14 Section 6

6.2.1 ATLAS-CONF-2012-059: 7 TeV 4.7/fb
Search for light stops.
Signal: 2 leptons + jets + MET
Bounds: assuming t˜→ bχ+
with mχ+ = 106 GeV, fixes mt˜ > 130 GeV for mχ0 < 65 GeV.
6.2.2 ATLAS-CONF-2012-070: 7 TeV 4.7/fb
Search for light stops with mass around (or lighter) than the top mass. Look for one or two leptons,
large MET, light flavor jets, and b-jets.
Signal: 1 or 2 leptons + 1 or 2 b-jets + jets + MET
Bounds: Assuming t˜→ bχ+
. Excludes 120 GeV <mt˜ < mt for mχ0 ∼ 55 GeV.
6.2.3 ATLAS-CONF-2012-071: 7 TeV 4.7/fb
Search for medium-mass stop partners decaying to t and neutral non-interacting particle. Sensitive
up to stop masses up to 200 GeV. Direct pair production of top partners (stop or spin-1/2 partner)
with t˜→ tχ0
. Look for two leptons in final state.
Signal: 2 leptons + jets + MET.
Bounds: Rules out a mt˜≈ 300 GeV decaying to a massless neutralino, assuming a signal that
is 1σ below the central value.
6.2.4 ATLAS-CONF-2012-073: 7 TeV 4.7/fb
Search for direct heavy stop production; events with an isolated lepton, jets, and MET, with
t˜→ t LSP.
Signal: 1 lepton + jets + MET
Bounds: Rules out 230 GeV <mt˜ < 440 GeV for massless LSP.
6.2.5 ATLAS-CONF-2012-074:
Heavy stop assuming t˜→ t LSP with both quarks decaying hadronically.
Signal: 0 lepton + jets + MET
Bound: excludes 370 GeV <mt˜ < 465 GeV for massless LSP.
6.3 Stop NLSP
See the Stop NLSP section above for the Katz and Shih analysis of past Tevatron and LHC bounds.
Here we review their section 4, which describes some of the current (at the time) bounds on the
stop NLSP scenario.
The searches which are most sensitive are those which:
• are more accepting of soft jets, or
• use more discriminating variables such as mT or some kind of reconstructed stop mass.
6.3.1 Cut and count
The expected number of stop events is
Nt˜t˜∗ =
εt˜t˜∗
εtt¯
×
σt˜t˜∗
σtt¯
× Nt t¯.
The first factor is the relative acceptance for stop pair production. The second factor is the relative
production cross section. Armed with this information, one can calculate exclusion confidence
levels. Here are the results:
Experimental bounds 15

Remarks on the self-consistency of their simulations:
• The use of these ratios and Ntt¯ cause many systematic errors in the simulations to cancel.
(This is a sanity check for the results.)
• Raw number Ntt¯ agrees with experimental references at 30%.
• The stop cross section has a power law dependence on the stop mass, but the stop limits
are not appreciably changed by 10% changes in the acceptance in either direciton.
Results, very light stops ∼120 GeV
• Acceptance is aﬀected signiﬁcantly by the requirements on number and ET of jets. Light
stops tend to decay into soft b-jets.
16 Section 6

• The acceptance is close to 1 only for the D0 stop search since it does not impose a require-
ment on the number of jets. This gives it the best cut-and-count exclusion for a 120 GeV
stop.
• The b-tagged sample in the CDF search is relatively high, while other analyses have stricter
jet ET requirements leading to lower acceptences.
Results, ∼150 GeV intermediate mass stops
• b-jets are harder and are more efficient in passing selections, but reduced cross sections
weaken the limits.
• ATLAS top partner search for leptons + jets has high acceptance because it cuts on mT >
120 GeV, which elminates the top backgroun.
• O(100)/pb data should be able to reach 95% CL exclusion for stops up to 180 GeV, but
beyond this one is limited by systematic errors.
6.3.2 Stop mass reconstruction
Based on 0912.1308, CDF Note 9439, FERMILAB-THESIS-2010-16, which uses a dilepton stop
mass reconstruction algorithm at CDF. The algorithm assumes a stop decay t˜→ bχ+ → ℓνbχ0,
but turns out to work also for the stop NLSP scenario by giving a much sharper mass distribution
compared to the top that is centered at a different value. It seems like Katz and Shih mapped on
stop NLSP parameter points to similar gravity mediated parameter points so that they could use
the CDF analysis.
The result of this analysis gives a stop NLSP exclusion for mt˜<150 GeV. This is the best limit
in their paper. “This illustrates the power of using more discriminating variables in searching fo
rnew physics in the ditop sample.” Future directions include using the D0 stop search (which they
could not reinterpret because it would require access to their data).
6.3.3 Other types of measurements
• Invariant mass of the lepton-b system. (See hep-ph/9909536) Can reduce ditop background.
Ambiguities in pairing lepton and b-jet, but even incorrect pairing contributes in a similar
way to the differences between distributions. This distrubtion has not been published (as
of the Katz-Shih paper) since the 700/pb data from the Tevatron.
• b-jet pT distribution. Is there a way to use the softness of b-jets from stop decay to separate
the ditop background? Limiting factors: pT 12GeV for proper reconstruction of b-jet pT
and efficiency decreases with pT .
• Displaced decays. No dedicated searches as of the time of the Katz/Shih paper, though
you can find some ideas from searches for hidden valley scenarios. Current ATLAS and CMS
studies for charged stop-hadrons are not constraining due to the requirements of each search.
6.4 The gluino
Current bounds (from Han, Katz, Krohn, Reece) push the gluino above ∼900 GeV.
• ATLAS-CONF-2012-004: two same-sign leptons, jets and missing transverse momentum
• 1205.3933: CMS, same-sign dileptons and b-tagged jets
• CMS-PAS-SUS-11-027: single lepton and jets using templates
• ATLAS-CONF-12-037: large jet multiplicities and missing transverse momentum
6.5 Flavor/Naturalness bounds on stop mixing
6.5.1 1206.5303: IAS Higgs couplings from natural SUSY
Defines ratio of Higgs couplings to SM values:
ri =
ghii
ghii
SM
.
Experimental bounds 17

Writing Xt = At − µ/tan β, we have rG = rtrG
t˜
with
rG
t˜
− 1 ≈
1
4
mt
2
mt˜1
2 +
mt
2
mt˜2
2 −
mt
2
Xt
2
mt˜1
2
mt˜12
2 + (D − term contribution)
This can give a substantial effect for reasonably light (≈250GeV) stops: rG =1.24. This would give
a 53% increase in gluon fusion Higgs production. A large stop mixing, however, could reduce this
effect. This is limited by naturalness: a large Xt contributes to the weak scale tuning directly and
through the requirement of larger diagonal soft masses.
Stronger constraints for stop mixing come from B → Xsγ, which only allows new physics to
contribute at about 30% of the loop-suppressed SM contribution. To avoid large tree-level SUSY
contributions, it is typical to assume parameters in which such contributions cancel and measure
the fine tuning. The result of the IAS group analysis is that to avoid cancellations to a level in one
part per ten, we must have Atµ tan β/mt˜2 < few.
6.6 Fits to Higgs decay channels
6.6.1 1207.1445: Fermilab h → γγ vs h → gg rates
1, 2, and 3σ contours for the global fit of the ICHEP 2012 Higgs data compared to the range of
decay widths that can results from the addition of new charged particles or new charged-and-
colored particles. Star marks the best fit for a top partner corresponding to Rt = yt/yt
SM
= −0.84.
6.6.2 Isn’t there also an Italian paper with a similar discussion?
Appendix A Review of the MSSM
Particle content, following Terning’s conventions. Content based on Csaba’s review.
χSF SU(3) SU(2) U(1) B L
L 1 −
1
2
0 1
E¯ 1 1 1 0 −1
Q
1
6
1
3
0
U ¯ 1 −
2
3
−
1
3
0
D ¯ 1
1
3
−
1
3
0
H1 = Hd 1 −
1
2
0 0
H2 = Hu 1
1
2
0 0
The superpotential includes Yukawa terms and the µ term (flavor indices suppressed):
W = λu QHu U¯ + λdQHd D¯ + λLLHdE¯ + µHuHd.
18 Appendix A

On top of this, the following terms are B or L violating and are removed by R-parity which is
defined in terms of B, L, and the spin S by R = (−)3B+L+2S
,
WR =α1 QL D¯ +α2LL E¯ +α3LHU +α4 D¯D¯U¯.
Appendix B MSSM Higgs scalar potential
One of the most important parts of the MSSM is the scalar potential for the Higgs and the
constraints from the requirement of electroweak symmetry breaking. See hep-ph/9606414 for
details.
B.1 Supersymmetric contribution
Recall that the general form of the scalar potential for a renormalizable supersymmetric theory is
V (φ) = Wi
∗
W i
+
1
2
a
ga
2
(φ∗
Ta
φ)2
,
where the first term is the F-term contribution and the second is the D-term contribution. From
the superpotential we can see that only the µ-term contributes to to F-term potential since the
sfermions don’t get vevs:
∆VF (Hu, Hd) = |µ|2
|Hu|2
+ |µ|2
|Hd|2
.
The D-term potential is simpilified using σijσkl = 2δilδjk − δijδkl,
∆VF (Hu, Hd) =
1
2
g2 Hu
∗ σ
2
Hu
2
+
1
2
g′2
|Hu|4 + (u→ d)
=
1
2
g2
(2|Hu
+
|2
|Hu
0
|2
− |Hu
0
|2
|Hu
+
|2
) +
1
2
g′2
(|Hu
0
|2
+ |Hu
+
|2
) + (u→ d, +→−)
=
1
8
(g2 + g′2
)(|Hu
0
|2 + |Hu
+
|2 − |Hd
0
|2 − |Hd
−
|2)2 +
1
2
g2|Hu
+
Hd
0∗
+ Hu
0
Hd
−∗
|2.
B.2 Mass matrix in SUSY limit
Now let us write out the mass matrix for a complex scalar field φ in the basis Φ= (φ, φ∗)T so that
the mass matrix M2 appears as
1
2
Φ∗M2Φ. Sub/superscripts i, j,k refer to derivatives with respect
to φ or φ∗
respectively, while subscripts a refer to the gauge group.
(M2
)i
j =

 W¯ ik
Wkj +
1
2
Da
i
Daj +
1
2
(Da)j
i
Da W¯ ikj
Wk +
1
2
Da
i
Da
j
W¯ k
Wikj +
1
2
DaiDaj W¯ jk
Wki +
1
2
Da
i
Daj +
1
2
(Da)j
i
Da


B.3 SUSY breaking contributions
We now include soft SUSY breaking terms. These come in the form of holomorphic masses (B
terms) and soft masses. (There are no holomorphic trilinear A terms by gauge invariance.)
∆Vsoft(Hu, Hd) = mHu
2
(|Hu
0
|2
+ |Hu
+
|2
) +mHd
2
(|Hd
0
|2
+ |Hd
−
|2
) − Bµ(Hu · Hd)
= mHu
2
(|Hu
0
|2
+ |Hu
+
|2
) +mHd
2
(|Hd
0
|2
+ |Hd
−
|2
) − Bµ(Hu
+
Hd
−
− Hu
0
Hd
0
+ h.c.)
where we’ve written the B term with a factor of µ pulled out as a choice of convention. Note that
it is also standard to write b = Bµ (e.g. Stephen Martin’s review).
B.4 Total Higgs scalar potential
As written in Martin’s review (eq. 8.1.1 in version 6):
V = (|µ|2 + mHu
2
)(|Hu
0
|2 + |Hu
+
|2) + (|µ|2 + mHd
2
)(|Hd
0
|2 + |Hd
−
|2)
+[b(Hu
+
Hd
−
− Hu
0
Hd
0
) + h.c]
1
8
(g2 + g′2
)(|Hu
0
|2 + |Hu
+
|2 − |Hd
0
|2 − |Hd
−
|2)2 +
1
2
g2|Hu
+
Hd
0∗
+ Hu
0
Hd
−∗
|2
MSSM Higgs scalar potential 19

B.5 Electroweak symmetry breaking
[I’ve been sloppy with the sign of b, sorry.] Note that the supersymmetric potential is
leads to a positive definite mass matrix whose vev is Hu = Hd = 0. Note, further, that we need
Hu
+
= Hd
−
= 0; in the following we write H for H0
for simplicity. In order to break electroweak
symmetry, we need the soft SUSY breaking terms. Introducing some notation, the form of our
potential is:
V = mH1
2
|Hd|2
+ mH2
2
|Hu|2
− m12
2
(HuHd + h.c.) + (D terms).
One may worry about the negative contribution coming from the m12
2
piece. The quartic contri-
bution from the D term (given above) vanishes in the direction |Hu| = |Hd| (they are “D-flat”), so
this is a legitimate concern. In order for the potential to be bounded from below in this direction,
we thus require
mH1
2
+ mH2
2
> 2|m12
2
|.
mH1
2
= |µ|2
+ mHd
2
mH2
2
= |µ|2
+ mHu
2
m12
2
= b=Bµ
Observe that b = Bµ alsways favors electroweak symmetry breaking. The D term contributions
are given above. Requiring that we get a negative squared mass near Hu = Hd = 0 for one linear
combination gives (by taking the determinant of the mass matrix)
b2
> mH1
2
mH2
2
.
You might worry that this is only sufficient and not necessary since one might have the mass matrix
having two negative eigenvalues. However, the previous condition mH1
2
+mH2
2
>2|m12
2
| enforces that
the trace of the mass matrix is positive. Thus EWSB requires that the mass matrix has exactly
one negative eigenvalue. (Thanks to the Sparticle book for spelling this out.)
We thus have the bounds:
1. 2b < mHu
2
+ mHd
2
+ 2|µ|2
2. b2
> (|µ|2
+ mHd
2
)(|µ|2
+ mHu
2
).
These relate supersymmetric in the superpotential to soft breaking terms, parameters which a
priori are expected to be unrelated. Note that there is no solution for mHu
2
= mHd
2
, if one expects
these two to be equal at a high scale, then electroweak symmetry is unbroken at tree-level.
Comments on RG: One can consider the RG running of these parameters from the high scale
down to the weak scale. A reasonable assumption is to only include the top Yukawa coupling and
the gauge interactions. The relevant expressions are the β-functions for mH2
2
, mt˜
2
, and mQ3
2
. These
are functions of the gaugino masses, the gauge couplings, the top Yukawa, and the top A-term.
The main story is this: (see hep-ph/9606414): Loops involving the top Yukawa coupling want to
make the up-type Higgs mass negative, this is indeed what we want for EWSB.
Let us assume that the neutral component of the Higgs doublets obtain vevs vu and vd. As
usual, we define v2
= vu
2
+ vd
2
= 2mZ
2
/(g2
+ g′2
) = (174 GeV)2
and tan β = vu/vd. These vevs can be
chosen to be real and positive. Using these relations we can write ∂V /∂Hu,d
0
= 0:
∂V
∂Hu
0
= 2mH2
2
vu − 2bvd +
1
2
(g2 + g′2
)(vu
2
− vd
2
)vu
= 2vu |µ|2
+ mHu
2
−b cotβ +
mZ
2v2 (vu
2
− vd
2
)
= 2vu |µ|2
+ mHu
2
−b cotβ −
mZ
2
cos 2β .
Similarly for ∂V /∂Hd
0
, with a minus sign on the final term. We end up with the conditions:
mHu
2
+ |µ|2
− b cot β −
mZ
2
cos 2b = 0
mHd
2
+ |µ|2 − b tan β+
mZ
2
cos 2b = 0.
20 Appendix B

It is common to rearrange these into constraints on the parameters:
|µ|2
=
mHd
2
− mHu
2
tan2
β
tan2β − 1
−
1
2
mZ
2
b =
(mHu
2
+ mHd
2
+ 2|µ|2
)sin 2β
2|µ|
.
MSSM Higgs scalar potential 21

Light Stop Notes

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to Light Stop Notes (20)

More from Flip Tanedo (20)

Recently uploaded (20)

Light Stop Notes