Statusofthe& Experimentvipres/MIPP-Data-for-NuMI.pdf · 2012. 12. 8. · Statusofthe& Experiment...

Status of the Experiment

Jonathan M. PaleyNeutrino Flux Workshop, Pi9sburgh

December 8, 2012

Jonathan Paley, ANL HEP Division

2

Main Injector Par7cleProduc7on (MIPP) Experiment

• Goal: collect comprehensive hadron produc5on cross-‐sec5on data set with par5cle id using various beams and targets (thick and thin).

TargetJGG

Magnet

CkovDetector

RICHDetector

RosieMagnet

ToFDetector

WireChambers

TPC

BeamCkov

25 mEM &

HadronicCalorimeters

• Full acceptance spectrometer

• Two analysis magnets deflect in opposite direcLons

• TPC + 4 DriP Chambers + 2 PWCs

• Designed for excellent parLcle ID (PID) separaLon (2-‐3σ)

Nearly all detectors used in MIPP were taken from previous experiments.

1 100

0.5

1

TPC ToF Ckov RICH1 10 80Momentum (GeV/c)

Experiment located in MC7 at FNAL

~7 x 106 π’s, K’s and p’s at 85, 60, 20 and 5 GeV/c on 1% λL LH2 target.

~4 x 106 π’s, K’s and p’s at 35, 60 and 120 GeV/c on Bi and U targets.


3

The 2005-‐06 Data Run

• MIPP began its physics run in December 2004 and ran un5l February 2006.

• DAQ rate was ~25 Hz, with MIPP receiving ~5% of MI beam.

• Data collected:

• ~1.6 x 106 events of Main Injector 120 GeV/c protons on a spare NuMI target.

• ~3.2 x 106 π’s, K’s and p’s at 120, 60, 35 and 20 GeV/c on 1-‐2% λL C and Be targets.

Be

Bi

U

Cu

C Ag

MTMIPP

Rotating Target Wheel

Al

LH2 Target

Time Projec7on Chamber (TPC)

Centerpiece of MIPP, originally built for the EOS experiment and used in several other prior experiments.

Measures track trajectory in 3D: (x,z) posi5on → pad loca5ons, y posi5on → drif 5me.

Ac5ve volume of ~1 m3 and a resolu5on of ~0.5 cm3.

PID via <dE/dx> below ~1 GeV/c.

Pad Plane

E-Field

BeamDirection

4

TPC Calibra7onsRaw Data, 120 GeV/cbeam track

Corrected data using Magboltz simulation

~5 cm

Before correction After correction

Inhomogeneous magne5c field causes drif electrons to deviate from straight-‐line path to pad plane on bojom on TPC. Devia5ons of up to ~5 cm are observed!

Using a map of the magne5c field and the Magboltz simula5on, we correct these ExB drif effects to the level of ~90% (~2 mm worst case).

Electron drif velocity is found to be run/5me dependent: sensi5ve to the water contamina5on in the P10 gas!

Time-‐dependent correc5ons to the drif velocity are made and δvD /vD ~ 1%.

+×

5

Global Track Reconstruc7on

TPC track segments are matched to downstream drif chamber hits, momentum is determined from bend in both magnets.

MIPP (FNAL E907) Mom.: 120 GeV/cTarget: NuMIRun: 15118SubRun: 0Event: 33 Sun Jul 24 200513:10:30.411046 *** Trigger ***BeamWord: 0080Bits: 80D7


-900 -800 -700

50

0

50

0 1000

-50

0

50

-50 0 50

c

50

0

50

-900 -800 -700

50

0

50

0 1000

-50

0

50

x/y a/b v/su/t

DC123 Full










TPC DriP Chamber Hits

6


7

(p (GeV/c))10

log-1 0 1 2

Res

olut

ion

(%)

0

2

4

(p)10

Resolution vs. log (p)10

Resolution vs. log

Momentum Resolu7on and Bias

Black points determined by fipng central peaks of slices of dp/p to Gaussian.

Momentum resolu5on is < ~5%

Bias < ~2%. Correc5on is applied and has a very small uncertainty.

Transverse momentum resolu5on is < 0.02 GeV

)Fit

(p10

Log-1 0 1 2

Fit

)/p

Fit

-pT

rue

(p

-0.2

-0.1

0

0.1

0.2

1

10

210

310

410

)Fit

(p10

Momentum Bias vs Log

/ ndf 2! 2814 / 88p0 0.000013± 0.008593 p1 0.00005± 0.02162 p2 0.00018± -0.02424 p3 0.00027± -0.02609 p4 0.00059± 0.02744 p5 0.00046± 0.01312 p6 0.00046± -0.02453 p7 0.00042± 0.01058 p8 0.000101± -0.001564

/ ndf 2! 239.682 / 133

Constant 5.012± 1331.387

Mean 0.023± 118.547

Sigma 0.032± 6.078

(GeV/c)Fit

p100 120 140

0

500

1000

1500 / ndf 2! 239.682 / 133

Constant 5.012± 1331.387

Mean 0.023± 118.547

Sigma 0.032± 6.078

NuMI MC

/ ndf 2! 239.682 / 133

Constant 5.012± 1331.387

Mean 0.023± 118.547

Sigma 0.032± 6.078

/ ndf 2! 204.992 / 133

Constant 4.931± 1302.650

Mean 0.025± 120.111

Sigma 0.035± 6.319

/ ndf 2! 204.992 / 133

Constant 4.931± 1302.650

Mean 0.025± 120.111

Sigma 0.035± 6.319

/ ndf 2! 204.992 / 133

Constant 4.931± 1302.650

Mean 0.025± 120.111

Sigma 0.035± 6.319

Uncorrected

Corrected

/ ndf 2! 79.658 / 133

Constant 4.737± 1209.609

Mean 0.028± 117.974

Sigma 0.041± 6.574

(GeV/c)Fit

p100 120 140

0

500

1000

/ ndf 2! 79.658 / 133

Constant 4.737± 1209.609

Mean 0.028± 117.974

Sigma 0.041± 6.574

NuMI Data

/ ndf 2! 79.658 / 133

Constant 4.737± 1209.609

Mean 0.028± 117.974

Sigma 0.041± 6.574

/ ndf 2! 70.101 / 133

Constant 4.638± 1181.475

Mean 0.031± 119.558

Sigma 0.044± 6.759

/ ndf 2! 70.101 / 133

Constant 4.638± 1181.475

Mean 0.031± 119.558

Sigma 0.044± 6.759

/ ndf 2! 70.101 / 133

Constant 4.638± 1181.475

Mean 0.031± 119.558

Sigma 0.044± 6.759

Uncorrected

Corrected


8

Absolute Momentum Scale

Afer momentum bias correc5on, single proton beam data and MC agree.

Reconstructed K0 invariant mass using tracks with p < 2 GeV/c indicates systema5c offset of ~-‐1%.

/ ndf 2r 202.1 / 96A 7.96± 31.48

0Km 0.0002± 0.4999 0Km 0.00020± 0.00665

DC Offset 1.6831± -0.6538

)2Inv. Mass (GeV/c0.45 0.5 0.55

PO

T/1M

eV6

Num

./1x1

0

0

20

40 / ndf 2r 202.1 / 96

A 7.96± 31.48 0Km 0.0002± 0.4999 0Km 0.00020± 0.00665

DC Offset 1.6831± -0.6538

Bkg-Subtracted Inv. Mass Distribution, NuMI MC, dz Cut

/ ndf = 100.9 / 962rA 1.8± 11.1

0Km 0.001± 0.496 0Km 0.000738± 0.007229

DC Offset 0.464± 0.351

/ ndf = 100.9 / 962rA 1.8± 11.1

0Km 0.001± 0.496 0Km 0.000738± 0.007229

DC Offset 0.464± 0.351

Preliminary

Preliminary


9

TPC PID Performance

TPC data are calibrated such that <dE/dx>(π) is 1 for p = 0.4 GeV/c and give expected Bethe-‐Bloch func5onal form.

<dE/dx> resolu5on ~10%.

Clean π, p separa5on between 0.2 and 1.2 GeV/c.

q*(0.5+log(dE/dx))-2 0 2

1

10

210

310

410

TPC <dE/dx> for 0.30 < P < 0.33 GeV/c

log(p/GeV)-1 0 1 2

q*(1

/2+l

og(d

E/dx

))

-2

-1

0

1

2

1

10

210

310

410

TPC <dE/dx> vs. P, Full NuMI Data Set

Preliminary

Preliminary

e-


10

MIPP PreliminaryData-‐driven calibraLon improved Lming resoluLon by about a factor of 2.5

ToF PID Performance

Kaons

Pions

Protons

MIPP Preliminary

Kaon peak clearly visible in ToF m2 distribuLon!


11

Since all mirrors have a different response, each measurement of Npe is normalized to that of a β=1 par5cle.

Pion “turn-‐on” clearly visible; proton “turn-‐on” also visible in slices of momentum.

Shape of normalized response dist. in MC agrees very well with data.

Data-‐driven calibra5on of 96 mirrors found detector response gives <10 pe/β=1 track.

Ckov PID Performance

1 10 100

-1

0

1

p [GeV/c]

0 /

NPE

N×q

+π

-π

+K

-K

p

p

(a)Ckov Detector Response

0 / NPE N×q -4 -3 -2 -1 0 1 2 3 4

310

410

(b) DataMC : all pions kaons protons

0 / NPE N×q -4 -3 -2 -1 0 1 2 3 4

310

(c) DataMC : all pions kaons protons

6.3 ≤ p < 9 GeV/c 13.8 ≤ p < 18.3 GeV/c Must only consider “isolated” tracks passing through mirrors; reject ~50% of Ckov PID data.


12

Ckov light ring formed on array of ~2300 1/2” PMTs.

Ring radius ~ Ckov angle ~ velocity.

3σ π/K separa5on up to 80 GeV/c, 3σ p/K separa5on up to 120 GeV/c

x (cm)‏

x (cm)‏

RICH PID Performance

Status of the NuMI Target Analysis

13

NuMI Target Analysis

14

•MINOS adjusts their predicted ND neutrino energy spectra to agree with the measured spectra using (pz,pT)-‐dependent weights; these weights are an empirical fit, similar to the BPMT parameterizaLon.

• The goal of this analysis is to provide similar weights to adjust the hadron producLon predicLon off of the NuMI target with a direct measurement of the parLcle yields.

• The MIPP results will be parLcle yields, binned in (pz,pT).

x (cm)−2 −1 0 1 2

y (c

m)

−2

−1

0

1

2

−1

−0.5

0

0.5

1)/sumnon−intx−(x,y)intx((x,y)

Preliminary

(GeV/c)z p1 10 210

(GeV

/c)

T p

-210

-110

1

0

1

234567

8

9

10

11

12

13

141516171819

20

21

22

23

24

25

262728293031

32

33

34

35

36

37

383940414243

44

45

46

47

48

49

505152535455

56

57

58

59

60

61

626364656667

68

69

70

71

72

73

747576777879

80

81

82

83

84

85

868788899091

92

93

94

95

96

97

9899100101102103

104

105

106

107

108

109

110111112113114115

116

117

118

119

120

121

122123124125126127

128

129

130

131

132

133

134135136137138139

140

141

142

143

144

145

146147148149150151

152

153

154

155

156

157

158159160161162163

164

165

166

167

168

169

170171172173174175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

)T

,pzBin Numbers vs. (p

Where We Are With the Analysis


15

I had been making good progress with the analysis before I made the choice to focus on NOvA in early 2010:

Event selec5on criteria finalized (begin with 2.11x106 POT, end with 1.45x106 POT)

Momentum bias and scale correc5ons finalized. Es5mates on uncertain5es on these are quite small compared to our proposed (pz,pT) bin sizes.

(pz,pT) bins determined based on σ(N(π+)) < 3%. There are 133 bins. In most cases (at low momenta), σ(N(π+)) < 1%. Probably don’t need that many bins, so to expedite the analysis, one could easily decrease the number of bins by factor of 2-‐5 and s5ll have a big impact on flux constraints.

Large MC data set generated (but needs further tuning)

Analysis (fipng) strategy outlined and mock data tests had begun on the approach. This needs to be revisited, in par5cular there may be bejer/quicker ways of doing this using modern/standard tools (eg, Root TMVA).

Some further progress has been made on the NuMI target analysis by other MIPP colleagues since I “disappeared”, eg, data-‐driven calibra5on of TCP <dE/dx> all the way to 120 GeV/c (there is informa5on beyond ~1.5 GeV/c).

Comparison of Data and MC pz Spectra

16

log10(p/GeV)-1 0 1 2

posi

tive

trac

ks

410

510

DataMC

log10(p/GeV)-1 0 1 2

nega

tive

trac

ks

210

310

410

510

DataMC

AnalysisThreshold

AnalysisThreshold

Preliminary

Preliminary

Comparison of Data and MC pT Spectra

17

p_T (GeV)0 0.5 1 1.5 2

posi

tive

trac

ks

310

410

510

DataMC

p_T (GeV)0 0.5 1 1.5 2

nega

tive

trac

ks310

410

510

DataMC

Below-thresholdelectrons

Below-thresholdelectrons

Preliminary

Preliminary

Preliminary Pion Yield Measurement(TPC-‐only)

18

<dE/dx>10

log0 1 2

Num

./PO

T

-610

-510

-410

-310

Data TPC <dE/dx> Distribution, q > 0, Bin 15

4.24e-04±) = 6.19e-02 +/N(


<dE/dx>10

log0 1 2

Num

./PO

T

-610

-510

-410

-310


2.60e-04±) = 2.50e-02 +/N(


<dE/dx>10

log0 1 2

Num

./PO

T

-610

-510

-410

-310


2.76e-04±) = 2.71e-02 +/N(


<dE/dx>10

log0 1 2

Num

./PO

T

-610

-510

-410


2.48e-04±) = 1.67e-02 +/N(


Preliminary

Preliminary

Preliminary

Preliminary

(GeV/c)z p1

)+ /)/N

(- /

N(

0

5

<= 0.05 GeV/c, +0t

0.00 < p

<= 0.15 GeV/c, +1t

0.05 < p

<= 0.19 GeV/c, +2t

0.15 < p

<= 0.25 GeV/c, +3t

0.19 < p

<= 0.33 GeV/c, +4t

0.25 < p

<= 0.45 GeV/c, +5t

0.33 < p

<= 2.00 GeV/c, +6t

0.45 < p

z) vs. p+/)/N(-/N(

Preliminary Pion Yield Measurement (TPC-‐only)

19 (GeV/c)z p1

Dou

ble

Rat

io

0

5

z) vs. p+//-/)/MC(+//-/Data(

<= 0.05 GeV/c, +0t

0.00 < p

<= 0.15 GeV/c, +1t

0.05 < p

<= 0.19 GeV/c, +2t

0.15 < p

<= 0.25 GeV/c, +3t

0.19 < p

<= 0.33 GeV/c, +4t

0.25 < p

<= 0.45 GeV/c, +5t

0.33 < p

<= 2.00 GeV/c, +6t

0.45 < p

z) vs. p+//-/)/MC(+//-/Data(

Data Data/MC

Preliminary

Preliminary

Conclusions and Thoughts on Possible Paths Forward

20

• MIPP collected several millions of events of π, K and p beams at various momenta incident on various targets, 1.6 x 106 120 GeV protons on an actual NuMI target.

• All MIPP sub-‐detector systems have been calibrated and the MC tuned to the data. MC/Data PID agreement looks reasonable, but some further fine-‐tuning is needed.

• PID response in MC can be tuned “by hand” using the above files to be “good enough”. ToF and RICH need some fine-‐tuning, TPC and Ckov should be ok now.

• Using Root tools (TMVA, TSVD) could make analysis more straigh{orward (not saying it’ll be “easy”!)

• Low-‐energy K± and K-‐short produc5on measurement is relevant for NOvA NDOS analysis, obviously other experiments would benefit from anything MIPP produces (yields or ra5os, p < 3 GeV/c, p > 20 GeV/c, etc.)

• Help from experiments that would benefit from these data would be appreciated, and it should not be difficult to get postdoc up and running. Eg, flat Root ntuple format of data and MC exists.

Statusofthe& Experimentvipres/MIPP-Data-for-NuMI.pdf · 2012. 12. 8. · Statusofthe& Experiment...

Documents

Transcript of Statusofthe& Experimentvipres/MIPP-Data-for-NuMI.pdf · 2012. 12. 8. · Statusofthe& Experiment...