powheg in herwig++ for susy - durpowheg in herwig++ for susy alix wilcock ippp, durham university...

Powheg in Herwig++ for SUSY

Alix Wilcock

IPPP, Durham University

27/04/2015

Based on work done with P. Richardson, S. Platzer and B. Fuks

Acronyms

SUSY = Supersymmetry

Want SUSY to solve the hierarchy problem

Supersymmetry

- it’s not dead

Require SUSY partner of the top quark to be light(mt1

. few TeV)

Still possible in compressed spectra scenarios

Supersymmetry - it’s not dead

Require SUSY partner of the top quark to be light(mt1

. few TeV)

Still possible in compressed spectra scenarios

Compressed spectra SUSY

Mass difference between SUSY particle and the decayproducts is small

No energetic Standard Model objects

Not a lot of missing ET

Look for SUSY particles recoiling against hard initial-stateradiation

χ01

χ01

q

qq

qp p

Compressed spectra SUSY

Mass difference between SUSY particle and the decayproducts is small

No energetic Standard Model objects

Not a lot of missing ET

Look for SUSY particles recoiling against hard initial-stateradiation

χ01

qq

p p

q χ01

q

Acronyms

Herwig++

= Hard emission reactions with interfering gluons

Acronyms

Herwig++ = Hard emission reactions with interfering gluons

Monte Carlo event generator

Takes theoretical model → simulates expected experimentaldata

Monte Carlo simulations split into several stages:

Hard process → Parton shower → Hadronisation

Matrix-element matching

Parton showers resum large logarithms

Good approx. in soft/collinear limit

Doesn’t describe hard emissions well

(Remember hard emissionsimportant when studyingcompressed spectra SUSY)

j

k

iMn

θ

1(pj+pk )2 = 1

EjEk (1−cos θ)

Improve simulation of hard radiation in the shower usingNLO matrix-element matching

Combines exact matrix elements with the parton shower

We use the POsitive Weight Hardest Emission Generator(Powheg ) formalism

Matrix-element matching

Parton showers resum large logarithms

Good approx. in soft/collinear limit

Doesn’t describe hard emissions well

(Remember hard emissionsimportant when studyingcompressed spectra SUSY)

j

k

iMn

θ

1(pj+pk )2 = 1

EjEk (1−cos θ)

Improve simulation of hard radiation in the shower usingNLO matrix-element matching

Combines exact matrix elements with the parton shower

We use the POsitive Weight Hardest Emission Generator(Powheg ) formalism

Powheg formalism

For a pT ordered parton shower, cross section for the first emission:

Normal parton shower

dσPS = B(ΦB)dΦB

[∆(pmin

T , pmaxT ) + ∆(pT , p

maxT )P(z) dΦR

]∆(pT , p

maxT ) = exp

(−∫ pmax

T

pT

P(z)dΦR

)

Powheg corrected parton shower

dσPO = B(ΦB)dΦB

[∆(pmin

T , pmaxT ) + ∆(pT , p

maxT )

RB dΦR

]

∆(pT , pmaxT ) = exp

(−∫ pmax

T

pT

RB dΦR

)

B(ΦB) = B(ΦB) + V(ΦB) +

∫R(ΦB ,ΦR)dΦR

Powheg formalism

For a pT ordered parton shower, cross section for the first emission:

Normal parton shower

dσPS = B(ΦB)dΦB

[∆(pmin

T , pmaxT ) + ∆(pT , p

maxT )P(z) dΦR

]∆(pT , p

maxT ) = exp

(−∫ pmax

T

pT

P(z)dΦR

)

Powheg corrected parton shower

dσPO = B(ΦB)dΦB

[∆(pmin

T , pmaxT ) + ∆(pT , p

maxT )

RB dΦR

]

∆(pT , pmaxT ) = exp

(−∫ pmax

T

pT

RB dΦR

)

B(ΦB) = B(ΦB) + V(ΦB) +

∫R(ΦB ,ΦR)dΦR

Matrix-element corrections

Powheg correction available in Herwig++ for large number ofStandard Model processes.

For BSM processes, limited by absence of virtual matrix elements→ Powheg style matrix-element correction

Generate hardest emission using RBut local normalization is B rather than B

Implement ME correction using Matchbox and MadGraph

MadGraph 5 - used to generate B and RMatchbox - framework for NLO calculations, MC@NLO andPowheg matching to the Herwig++ angular ordered anddipole showers

Matrix-element corrections

Powheg correction available in Herwig++ for large number ofStandard Model processes.

For BSM processes, limited by absence of virtual matrix elements→ Powheg style matrix-element correction

Generate hardest emission using RBut local normalization is B rather than B

Implement ME correction using Matchbox and MadGraph

MadGraph 5 - used to generate B and RMatchbox - framework for NLO calculations, MC@NLO andPowheg matching to the Herwig++ angular ordered anddipole showers

Top squark pair production

Simulated pp → t1t∗1 , stable t1 at

√s = 14TeV, mt1

= 700 GeV.Limit simulation to hard process + full parton shower

LO

MEC

10−6

10−5

10−4

10−3

10−2

10−1

Matchbox +MadGraph 5

dσ/dpT[fb/GeV

]

0 500 1000 1500 2000

0.6

0.8

1

1.2

1.4

pT,t1t∗1[GeV]

MEC/LO

Example: e+e− → qqg [hep-ph/0310083]

Uncorrected shower:

over populates hard regions of phase space in pT . mt1region

has unpopulated dead zone for pT & mt1

Effect of the matrix element correction: exclusion bounds

ATLAS search for direct production of the top squark inevents with missing ET and two b-jets

t1 → bχ+1 → bf f ′χ0

1 with mχ+1−mχ0

1= 5GeV

Selection criterion Signal region A Signal region B

EmissT > 150 GeV > 250 GeV

Leading jet, j1 pT > 130 GeV, |η| < 2.8 pT > 150 GeV, |η| < 2.8

Subleading jet, j2 pT > 50 GeV, |η| < 2.8 pT > 30 GeV, |η| < 2.8

Third jet, j3 veto if pT > 50 GeV, |η| < 2.8 pT > 30 GeV, |η| < 2.8

∆φ(pmissT , j1) - > 2.5

b-tagged jetsj1 and j2 b-tagged with

pT > 50 GeV, |η| < 2.5

j2 and j3 b-tagged with

pT > 30 GeV, |η| < 2.5

mink (∆φ(pmissT , jk )) for k ≤ 3 > 0.4 > 0.4

EmissT /(

∑ni=1(p

jetT

)i + EmissT ) > 0.25, n = 2 > 0.25, n = 3

mCT [ref] > 150, 200, 250, 300, 350 GeV -

HT,3 =∑

(pjT

)i for all i > 3 - > 50 GeV

mbb =√

(pb,1 + pb,2)2 > 200 GeV -

Effect of matrix element correction: before

ATLAS search for direct production of the top squark inevents with missing ET and two b-jets

t1 → bχ+1 → bf f ′χ0

1 with mχ+1−mχ0

1= 5GeV

Original signal simulated with MadGraph + PYTHIA 6

200 300 400 500 600 700

mt1 [GeV]

100

200

300

400

500

600

mχ

0 1[G

eV]

ATLAS-SUSY-2013-05

∆mχ+1 −χ0

1= 5 GeV

t 1→bχ+1

forb

idden ∫

L = 20.1fb−1

√s = 8TeV

ATLAS result

Herwig++

Effect of matrix element correction: after

ATLAS search for direct production of the top squark inevents with missing ET and two b-jets

t1 → bχ+1 → bf f ′χ0

1 with mχ+1−mχ0

1= 5GeV

Original signal simulated with MadGraph + PYTHIA 6

200 300 400 500 600 700

mt1 [GeV]

100

200

300

400

500

600

mχ

0 1[G

eV]

ATLAS-SUSY-2013-05

∆mχ+1 −χ0

1= 5 GeV

t 1→bχ+1

forb

idden ∫

L = 20.1fb−1

√s = 8TeV

ATLAS result

Herwig++

And now for something a little different...

We’ve looked at SUSY searches based on a hard ISR jet + EmissT

But this in not the only option...

“monojet” = ISR jet + EmissT

“monophoton” = energetic photon + EmissT

“monotop” = top quark + EmissT

χ01

qq

p p

q χ01

q

t

Monotop search for SUSY

Study LHC sensitivity using Monte Carlo simulations of 300fb−1 of14TeV collisions

Search for pp → t1χ01t with t1 → cχ0

1

Experimental signal is t + EmissT

t1 and χ01 light

Other SUSY particles decoupled m ≈ 10 TeV

Simulation

Signal:

pp → t1χ01t with t → bqq′ simulated in MadGraph 5

t1 → cχ01, parton shower, hadronization done with Herwig++

Background:

Process Simulation details

ttHard process at NLO with

PowhegBox, matched to Herwig++

Single top As above

tW production As above

W (→ lν) + light-jetsW production at NLO matched to

LO W+ 1 or 2 jets using Sherpa

γ/Z (→ l l/νν) + jets As above

Wbb with W → lνHard process at LO with

MadGraph, matched to Herwig++

Diboson NLO using Powheg in Herwig++

Event Selection Criteria

Designed to reflect final state of signal events

Exactly zero leptons

Exactly one b-jet, pT > 30GeV

Three other jets with pT > min(pbT , 40GeV)

Impose further cuts to maximize sensitivity = S√S+B

S , B are number of signal and background events passing the cuts

EmissT > 200GeV

50GeV < mjj < 100GeV

100GeV < mbjj < 200GeV

∆φ(pmissT , pj1) > 0.6 and ∆φ(pmiss

T , pb) > 0.6

∆φ(pmissT , pt) > 1.8

Results - Scan

Scan (mt1,mχ0

1) plane

Superimpose ATLAS monojet search for pair-produced top squarks

150 200 250 300

mt1[GeV]

50

100

150

200

250

300

350

mχ

0 1[G

eV]

mt1< m

χ01

+mc

ATLAS: 95% CL, 20fb−1

Monotop: 2σ, 300fb−1

Can monotops provide competitive sensitivity to monojettechniques?

No!

Results - Scan

Scan (mt1,mχ0

1) plane

Superimpose ATLAS monojet search for pair-produced top squarks

150 200 250 300

mt1[GeV]

50

100

150

200

250

300

350

mχ

0 1[G

eV]

mt1< m

χ01

+mc

ATLAS: 95% CL, 20fb−1

Monotop: 2σ, 300fb−1

Can monotops provide competitive sensitivity to monojettechniques? No!

Summary

SUSY is not dead

Light top squarks could still exist in compressed spectrascenarios

Search for them using non-standard analysis techniques, e.g.monojet or monotop signals

Monojet searches require accurate simulation of hard radiationand work pretty well

Monotops are not competitive at LHC energies