1

Dec 7/22 PionLT/KaonLT Analysis Meeting Notes

2



3

(Notes by GH and SJDK)

4


5

Today: KaonLT will be discussed first

6


7

Please remember to post your slides at:

8

https://redmine.jlab.org/projects/kltexp/wiki/Kaon_LT_Meetings

9


10

Present

11



12

Regina  Garth Huber, Stephen Kay, Ali Usman, Alicia Postuma, Nathan Heinrich

13

CSULA  Konrad Aniol, Erika Gwin

14

CUA  Richard Trotta

15

JLab  Dave Gaskell

16

Ohio  Julie Roche, Jacob Murphy

17

FIU  Pete Markowitz

18


19

Richard Updates

20



21

 LT separation code

22

 it's slow getting information back from Bill

23

 Heep analysis

24

 SIMC/hcana calculate variables differently

25

 Adding new variables that calculate it in the same way as hcana into SIMC

26

 DG  In complete recon event or a new subroutine?

27

 In a new one

28

 DG  Should be fine

29

 Maybe call it "recon_hcana"?

30

 Don't do it as a separate branch!

31

 Will be annoying to integrate, also everyone will want this

32

 New submodule is good, will be modular then

33

 Total Efficiencies per run for each setting

34

 Needed input into Bill's LTsep code

35

 Haven't incorporated Ali's updated efficiencies

36

 Error bars are off as a consequence

37

 Some are grabbing the wrong calculations too

38

 t binning plots/demonstrations shown for Runs 49655009 LeftCentRight

39

 ~2% downtrend seen in Effs for Right setting, probably following real

40

changes in running conditions

41

 Actual tbinning and diamond cuts will be done in Bill's LTsep code

42

 Need to use Jacob's code to apply diamond cuts

43

 GH  For MMK plot, should zoom in on Lambda peak and see if it's

44

stable versus run number. Apply finer binning too

45

 RT  Will add it, will do 2D phi dist too

46

 GH  also need to add a slide of what Effs and typical values are used in

47

the TotEff calc, so we can see exactly what was done

48


49

 *Need to do a HMS calorimeter efficiency study*, can't rely on what

50

the report file gives

51

 DG: this needs to be for all PID detectors

52

 need to select some runs which are sufficiently clean and then apply

53

the obtained values globally, preferably vs. spectrometer momentum

54

 often, can never get rid of enough pi (for HMS), so might even

55

have to resport to Heep data, but then have to worry about not

56

illuminating the full focal plane

57

 the studies are simple, but this needs to be completed before starting

58

LTseparations

59


60

 iFarm issues

61

 Tape silo had a robotics failure yesterday, can't get stuff from the tape

62

silo currently

63

 can only replay data already cached to disk

64


65

Ali Updates

66



67

 Updates to Error Calculations

68

 Heep coin data from 6.2/8.2/10.6 GeV

69

 Error bars were too big

70

 New calculation using binomial errors

71

 need to *add* a slide showing old and new calculations

72

 New calculation shrinks error bars quite drastically

73

 For 8.2 Gev, maybe even too small

74

 10.6 GeV looks good with one outlier

75

 All efficiencies seem to be lower for this point

76

 Has lowest beam on time?

77


78

 DG  Need to *correct* EDTM LT for time the beam is off

79

 the issue is that the EDTM keeps counting when beam off, need to

80

correct the scalers via an approximate formula derived by PBosted

81

 Richard's Lumi script should do this?

82

 DG  *Current off correction* from PBosted study:

83

 LiveTime=(Rmeas(1f))/f

84

 Rmeas = edtm_accepted/edtm(nominal edtm live time)

85

 f = current(no cut)/current(with threshold)

86

 this formula is an *approximation*, assumes the beam ramp up time is

87

negligible, which may or may not be true, depending on the running

88

conditions, probably this approximation will be worse for early 10.6

89

GeV running

90

 AU  probably tnis is NOT implemented yet for Heep, physics settings,

91

will discuss with RT

92


93

 DeltaPi

94

 Ali has received an invitation from Stefan Diehl to present some DeltaPi

95

results at a conference at ECT* in Trento in August

96

 will discuss JLab conference travel funds application with RT

97


98

Stephen Updates

99



100

 No new progress on RF class

101

 Will update hcana in group directory to use new RFtime variables

102

 Stephen will try to do this today, let him know if there are any issues

103

with the new hcana version, should only affect the CT/RF classes

104

 Should not need a DEF file change

105

 RFTime.* is in there currently, this will pick up new variables

106


107

Alicia Updates

108



109

 BSA analysis

110

 Edited report files to calculate effs separately for each helicity

111

 Can't separate charge by helicity

112

 Will cut charge in half to normalise dummy subtraction

113

 DG  Should contact Steve Wood wrt to helicity scalers

114

 Might've been added for A1n/d2n?

115

 SK  Would explain why helicity scalers didn't work for KaonLT but were

116

added later (but still had issues)

117

 AP Should CC Carlos Yero when asking Steve on this

118

 shown runs have same TotEffs for +, helicity to 3rd decimal point

119

but LH2, Dummy do have a different Effs

120

 DG  HGC pion efficiencies are way too low

121

 Just set them to 1 until RT/AU complete their study

122


123

 Weighted average of BSA for LeftCentRight

124

 Experimented w/Complete fit (A=sinphi,B=cosphi,C=cos2phi) and

125

Approximated fit (A=sinphi only)

126

 complete fit is overfitting the data, ending up with B,C>A for some

127

fits

128

 DG  can constrain B,C based on world data

129

 we have LT,TT from Fpi2 but only up to Q2=2.45

130

 also have CLAS data

131

 SK  Glancing back at old code from his PhD analysis @ Mainz:

132

he ran an *initial fit* with just one Sin/Cos term (dashed red for

133

Alicia), and then fed this into the second fit

134

 Can adjust the second fit in a few ways then, can constraint the

135

asymmetry A to vary in some range, such as permitting A to only vary

136

within 12sigma of initial fit

137

 B/C can also be restricted too

138

 Julie discusses some of the difficulties of interpreting the results

139

that were used when presenting DVCS BSA data

140

 Suppressing B,C terms can make things look "more precise", but

141

actually shrinks the error bars in an unrealistic manner

142

 in the end, the DVCS paper presented both sets of results in a paper

143

and gave a caution that the result and obtained error bar depends on

144

how the analysis is done

145

 GH  Could quote weighted everage of complete/approx fits?

146


147

 DG  What about *radiative corrections* to the BSA? This is an issue that

148

just occurred to me now

149

 GH  Unlike the LTsep results, we aren't quoting a cross section. Not

150

using SIMC to get asymmetry, while we use SIMC in the ratio method to

151

get the cross sections

152

 DG  Yes, Acceptances should cancel in the BSA

153

 GH  From SIMC studies for LT data, can *estimate* how large radiative

154

corrections are for a t bin by turning on/off the radcorr in SIMC and and

155

how large the effect is, no easy way to see how this corrects the

156

asymmetry

157

 DG  Iterative fit? Put SSA into SIMC and see how it changes things when

158

RadCorr On/OFF

159

 maybe can also use SIMC to do tbin centering for BSA, just like we

160

do for the cross sections via the Ratio Method

161

 DG  Interesting question, could be an advantage over CLAS analysis in

162

that we can actually study and estimate the RadCorr effects. Will think

163

about this further

164


165

Next Meeting

166



167

Wed Dec 14 @ 9:00 Eastern/8:00 Regina/6:00 Pacific

168

 PionLT will go first

169

