Software Tools in High Energy Physics
Matthew Low
Institute for Advanced Study
Collider Physics in Mexico
September 19, 2015
Lecture 1 Review
I Introduced Monte Carlo method of computing cross-sections
I Monte Carlo generatesevents(lists of four-vectors)
I Allows arbitrary phase space cuts
I Introduced Madgraph
I Downloaded and installed
I Made process (Feynman diagrams)
I Computed cross-sections
I Played withrun card settings
Lecture 2 Review
I Moved fromparton-level todetector-level
I Interfaced Pythia and Delphes with Madgraph
I Looked at differential distributions using ROOT
I Performed simple analysis using ROOT
Lecture 3 Overview
1 Beyond the Standard Model
I Exploring other models in Madgraph
2 Implementing your own model
I Introduction to FeynRules
Beyond the Standard Model
I Madgraph has a number of common BSM models implemented
I They are stored in themodelsdirectory
> ls models/
sm mssm nmssm heft 2HDM 4Gen RS ...
I Standard model
I Minimal supersymmetric standard model
I Next-to-Minimal supersymmetric standard model
I IncludeshGG and hγγ effective vertices
I Two Higgs doublet model
I Adds 4th generation of fermions
I Adds RS gravitons
Beyond the Standard Model
I Models are all stored inUFO format
I UFO = Universal FeynRules Output
I Paper: http://arxiv.org/abs/1108.2040
I Written in Python, relevant files are human readable
Beyond the Standard Model
I What defines a UFO?
I particles.pydefines fields
I parameters.pydefines parameters (by values or by formula)
I vertices.pylists the interaction vertices and their Lorentz structure
I couplings.pylists the coupling formulas at each vertex
I lorentz.pylists the Lorentz structures used in the vertices
I couplings orders.pylists the orderof each interaction
(to save time, Madgraph only generates leading order diagrams unless specified)
Beyond the Standard Model
I The order is specified viaQED=Nand QCD=Mcommands whereN,M counts vertices of that type
I Example: p p > t t∼only generatestt¯from QCD production corresponds toQCD=2 QED=0
Beyond the Standard Model
I Example: p p > t t∼
I There are also (subleading) diagrams withs-channelγ’s andZ’s
I To get these, specifyQCD=0 QED=2
Beyond the Standard Model
I Example: p p > t t∼
I There are also (subleading) diagrams withs-channelγ’s andZ’s
I To get these, specifyQCD=0 QED=2
Beyond the Standard Model
I Example: p p > t t∼
I Can get both sets of diagrams withQCD=2 QED=2
I Can read these asless than or equal to
I Can modify orders to select only certain classes of diagrams(to speed up calculations)
Beyond the Standard Model
I Returning to BSM models
I Can check particles in a given model
MG5 aMC> import model mssm ...
MG5 aMC> display particles
Current model contains 48 particles:
w+/w- x1+/x1- x2+/x2- h+/h- ve/ve∼ vm/vm∼ vt/vt∼
e-/e+ mu-/mu+ ta-/ta+ u/u∼ c/c∼ t/t∼ d/d∼ s/s∼ b/b∼
sve/sve∼ svm/svm∼ svt/svt∼ el-/el+ mul-/mul+ ta1-/ta1+ er-/er+
mur-/mur+ ta2-/ta2+
ul/ul∼ cl/cl∼ t1/t1∼ ur/ur∼ cr/cr∼ t2/t2∼
dl/dl∼ sl/sl∼ b1/b1∼ dr/dr∼ sr/sr∼ b2/b2∼
Supersymmetry
I Adds new symmetry that exchanges bosons and fermions Q|fermioni=|bosoni
Q|bosoni=|fermioni
I Every particle has a “s”-partner
Supersymmetry
I Broad phenomenology
I Colored partnerssquarks andgluinos are strongly produced
I Decay to SM particles and the LSP(lightest supersymmetry particle) I Signal is many SM particles andE/T
I Also can have various resonances
I No observations so far (mg˜ &1.3 TeV)
Supersymmetry
I “Natural” realization
I Only sparticles requires for Hierarchy problem are light
I The others are heavy and not seen in colliders
H˜
˜tL
˜bL
˜tR
˜ g
natural SUSY decoupled SUSY
W˜ B˜
L˜i,˜ei
˜bR Q˜1,2,˜u1,2,d˜1,2
I Discussed inhttp://arxiv.org/abs/1110.6926
Beyond the Standard Model
I e.g. MSSM particles
particle name
χ01,2,3,4 n1,n2,n3,n4 χ±1,2 x1+/x1-,x2+/x2-
˜
g go
ν˜ sve/sve∼, . . .
`˜ el+/el-,er+/er-, . . .
˜
q ul/ul∼,ur/ur∼, . . . t˜ t1/t1∼,t2/t2∼, . . . H,H±,A h2,h+/h-,h3
Beyond the Standard Model
I Can check interactions in a model
MG5 aMC> display interactions
Current model contains 812 interactions 1:h1 h1 h2 QED=1
2:h2 h2 h2 QED=1 3:h1 h1 h1 QED=1 4:h1 h2 h2 QED=1 5:a a h- h+ QED=2 ...
Beyond the Standard Model
I Example: Generatepp→g˜g˜
MG5 aMC> import model mssm MG5 aMC> generate p p > go go MG5 aMC> output MyGluinoProcess MG5 aMC> launch
MG5 aMC> generate events
Beyond the Standard Model
I Example: Generatepp→g˜g˜
I How does this depend on the model parameters?
I e.g. mg˜,mq˜, tanβ, etc.
I Stored inparam card.dat
I There are masses, widths, mixing angles, etc.
Beyond the Standard Model
I Example: Generatepp→g˜g˜
> cat param card.dat BLOCK MASS #
5 4.889917e+00 # mb 6 1.750000e+02 # mt 15 1.777000e+00 # mta 1000021 6.077137e+02 # mgo 1000022 9.668807e+01 # mneu1
SUSY Parameters
I Note in older supersymmetry studies, often one chooses a model with fewer parameters
I e.g. mSUGRA has only: M0,M1/2,A0, tanβ, and sign(µ)
I Means all parameters in supersymmetry card are not independent
I Specify high energy parameters, usespectrum generators to RG evolve down to the weak scale
I Options for how many loops to include in the running
2 4 6 8 10 12 14 16 18
Log10(Q/1 GeV) 0
500 1000 1500
Mass [GeV]
m0 m1/2 (µ2+m02)1/2
squarks sleptons M1 M2 M3 Hd Hu
SUSY Parameters
I At the LHC, supersymmetry searches are based onsimplified models
I Described in this writeuphttp://arxiv.org/abs/1105.2838
I Not tied to adhoc assumptions in the high-scale models
I e.g. consider looking for stop squarks
I ˜t→tχ1 →t+E/T
I ˜t→bχ±→b+W +E/T
I ˜t→tχ2 →t+Z +E/T
I Can apply to your specific SUSY model
SUSY Parameters
I Spectrum generators are still useful to make MSSM parameter cards
I Two common programs are:
I SUSPECT2
(http://www.coulomb.univ-montp2.fr/perso/jean-loic.kneur/Suspect/) I SoftSUSY
(http://softsusy.hepforge.org/)
I For low energy spectra, can disable all RG running
MSSM
I In the spirit of simplied models, studies often use simplified spectra
I Keep a few particles at low energies, decouple the rest
I Event generation is faster if you exclude irrelevant diagrams
I e.g. studying electroweakino pair production
I Relevant in split SUSY models
I Electroweakino production has lower rates, weaker limits
MSSM
I The process is: generate p p > x1+ n1 I Mediated by SM gauge bosons
I Also mediated by squarks
I But whenmq˜ is large, we know this diagram will go to zero
I Can speed up generation using special Madgraph syntax
MSSM
I The process is: generate p p > x1+ n1
I Recall theslashcommand to exclude particles
I And themultiparticlecommand to group particles
define squark = ul ul∼ dl dl∼ ur ur∼ dr dr∼
generate p p > x1+ n1 / squark
I Now only gauge boson mediated diagrams are computed
FeynRules
I What if we want to study something other than the MSSM?
I Use FeynRules(other default models also made in FeynRules) I Homepage is http://feynrules.irmp.ucl.ac.be/
FeynRules
I Below are thecategories
I Within each category are a number of implemented models
FeynRules
I e.g. the strongly interacting light Higgs
I Paper: http://arxiv.org/abs/hep-ph/0703164
I Basis of dimension-6 operators that parameterize strong dynamics
FeynRules
I If model is not already implemented, can implement it yourself
I FeynRules is a Mathematica package that allows the calculation of Feynman rules in momentum space for any QFT physics model
I The Feynman rules calculated by the code can then be used to implement the new physics model into other existing tools, such as MC generators
I Download fromhttp://feynrules.irmp.ucl.ac.be/
I Unzip into a directory
FeynRules.m
FeynRules
I Models are stored inModels
I Default include the standard model
I Models specified by 3 main types of information
I Fields
I Parameters
I Lagrangian
I Can also specify symmetries and can rotate fields
FeynRules
I We will look at some examples of how to input information
I e.g. photon field
V[1] = {
ClassName -> A, SelfConjugate -> True, Mass -> 0,
Width -> 0,
ParticleName -> ‘‘a’’, PDG -> 22,
PropagatorLabel -> ‘‘a’’, PropagatorType -> W,
Manual on the arxiv http://arxiv.org/abs/1310.1921 Table 6, Page 23-25
FeynRules
I The photon is a vector so it fills the arrayV[]
I Fermions fill the arrayF[]
I Scalars fill the arrayS[]
I Ghosts fill the arrayU[]
I Can also defined “unphysical” fields that are not mass eigenstates
V[11] = {
ClassName -> B, Unphysical -> True, SelfConjugate -> True,
Definitions -> { B[mu ] -> -sw Z[mu]+cw A[mu]}
}
FeynRules
I We can input parameters
I asexternalor internal
I with formulas or numbers
I e.g. the top mass(external, number) ymt = {
ParameterType -> External, BlockName -> YUKAWA, OrderBlock -> 6, Value -> 172,
FeynRules
I We can input parameters
I asexternalor internal
I with formulas or numbers
I e.g. sin2θw (internal, formula) sw2 = {
ParameterType -> Internal, Value -> 1-(MW/MZ)^2,
Description -> ‘‘Squared Sin of the Weinberg angle’’
}
FeynRules
I Note that standard model parameters are specified by 3 numbers
I The other parameters are computed in terms of these
I Common choices areα−1em,GF,αs.
Block sminputs
1 1.325070e+02 # aEWM1 2 1.166390e-05 # Gf 3 1.180000e-01 # aS
I Can write parameters in a new model in terms of a few inputs
FeynRules
I Once we havefieldsand parameters, we use them to write a Lagrangian
I The Lagrangian is written in Mathematica notation
I e.g. gauge kinetic terms
lag := -1/4 FS[B,mu,nu] FS[B,mu,nu]
- 1/4 FS[Wi,mu,nu,ii] FS[Wi,mu,nu,ii]
I lagis the Lagrangian
I FSis a function to represent the field strength
I B, Wiare the field variables
I mu, nu, iiare indices to be summed
FeynRules
I All of these are stored in text fileSM.fr
I Will use this file inside Mathematica
I In Mathematica, load FeynRules
FeynRules
I All of these are stored in text fileSM.fr
I Will use this file inside Mathematica
I In Mathematica, load FeynRules
FeynRules
I Now FeynRules has been loaded
I Next we need to load our model into Mathematica
I LoadModeluploads our fields, parameters, and Lagrangian
I LoadRestrictionremoves interactions considered irrelevant
I DiagonalCKM.rstremoves off-diagonal CKM interactions
I Massless.rstremoves Yukawa interactions for light quarks (huu¯,hdd¯,hss¯,hcc)¯
FeynRules
I Now the model has been uploaded
I FeynRules provides a number of functions to check your Lagrangian
I See Table 16, page 46
CheckMassSpectrum[Lag];
CheckHermiticity[Lag];
I Can construct vertices using
vertices = FeynmanRules[lag]
decays = ComputeWidths[vertices];
I Will warn when conserved currents are violated at vertices
FeynRules
I FeynRules does 2→1 computations already
I Sufficient to compute decay widths and branching ratios
PartialWidth[{φ1, φ2, φ3}, decays];
TotWidth[φ1, decays];
BranchingRatio[{φ1, φ2, φ3}, decays];
I Can be used for studies (without having to run Madgraph)
I Also returns closed form formulas for the width
FeynRules
I We have seen once we load a Lagrangian into FeynRules, we can
I perform simple checks (mass diagonal, Hermitian, etc.)
I compute widths and branching ratios
I Our main use is to write a UFO model to be used in Madgraph
WriteUFO[lag];
I Writes a new folder containing our model
I Take the folder, copy it intomodelsin Madgraph directory, now can run the new model!
FeynRules
I Let’s do an example
I ψ colored fermion (top partner) I φ neutral scalar(dark matter) I with the interaction
L=Lkin+λφψtR + h.c.
I The parameters areλ,mΨ,mφ
I (Simplified model of top partners in composite models)
FeynRules
I In FeynRules, we open a new file MyModel.fr
I We start with the parameters
M$Parameters = { lam = {
ParameterType -> External, Value -> 1.0,
InteractionOrder -> {NP, 1}, ComplexParameter -> False }
}
I NPnames the vertex, rather than QEDor QCD
FeynRules
I Next we specify the fields
I First, we start with ψ
M$ClassesDescription = { F[100] = {
ClassName -> tp, SelfConjugate -> False, Indices -> {Index[Colour]},
QuantumNumbers -> {Y -> 2/3, Q -> 2/3}, Mass -> {Mtp, 1000},
Width -> {Wtp, 1}
},
FeynRules
I Next, the fieldφ
S[100] = {
ClassName -> phi, SelfConjugate -> True, Indices -> { }, Mass -> {Mphi, 200}, Width -> {Wphi, 1}
} }
FeynRules
I Finally we write the Lagrangian
LKin := I tpbar.Ga[mu].DC[tp,mu] - Mtp tpbar.tp
+ 1/2 del[phi,mu] del[phi,mu] - 1/2 Mphi^2 phi^2;
LInt := lam1 phi tpbar.ProjP.t;
LNew := LKin + LInt + HC[LInt];
I Load the model into FeynRules
LoadModel[‘‘SM.fr’’, ‘‘MyModel.fr’’]
LoadRestriction[‘‘DiagonalCKM.rst’’, ‘‘Massless.rst’’]
FeynRules
I Can perform cross-checks (mass spectrum)
FeynRules
I Can perform cross-checks (Hermiticity)
FeynRules
I Can compute total widths
I Then write the UFO model
FeynRules
I Can compute total widths
I Then write the UFO model
FeynRules
I Directory made: MyModel UFO
I Copy this intomodelsin Madgraph directory
I Run Madgraph, but import this model
MG5 aMC> import model MyModel UFO MG5 aMC> generate p p > tp tp MG5 aMC> output MyOwnProcess
FeynRules
I Can include the new interaction
I Notice the coupling orderNP=2
I Now can compute cross-sections and generate events
FeynRules
I Production rate forpp→ψψ¯
[TeV]
MT
2 4 6 8 10 12 14 16
TT) [pb]→(pp σ
10-6 10-5 10-4 10-3 10-2 10-1 1 10 102
)-1(3 abnumber of events
10 102 103 104 105 106 107 108 109
Fermionic Top Partner 100 TeV 14 TeV
I (Aside: as mentioned, stop searches can be recast for this channel, very useful for future collider studies)
Lecture 3 Summary
I Looked at the new physics models included in Madgraph
I Learned how to load a model and check particle content
I Noted that additional models are available on FeynRules website
I Downloaded and load FeynRules program
I Learned manipulations/checks that can be done with FeynRules
I Implemented our own model into FeynRules