1

Meeting to discuss BSA systematic uncertainties calculation

2

24Apr02

3

(Notes by GH)

4


5

Present

6



7

Regina  Garth Huber, Nacer Hamdi, Ali Usman, Alicia Postuma, Muhammad Junaid,

8

Nathan Heinrich, Vijay Kumar

9

JLab  Dave Gaskell

10

Ohio  Julie Roche

11

CUA  Tanja Horn

12

York  Stephen Kay

13


14

Alicia has prepared some slides to guide our discussion.

15


16

At present, Alicia calculates the BSA 10 different ways:

17

FullFit SineFit

18

Nominal Cuts A A'

19

Narrow CT CN CN'

20

Wide CT CW CW'

21

Narrow MM MN MN'

22

Wide MM MW MW'

23


24

For each one, obtain a fitting uncertainty for Sin(phi) term, yielding 10

25

different values of the BSA and 10 corresponding uncertainties.

26


27

The question is how should we calculate A +/stat +/syst ?

28

 right now, taking the errorweighted average of A and A' as the final BSA

29

result

30


31

Q1: at what stage do we take the weighted average of the Full and Sine fits?

32


33

Option 1: calculate error weighted average for every set of cuts, relative to

34

the (A and A' average), and use this to calculate delta(CT), delta(MM)

35


36

Option 2: average the cut dependences first, not relative to (A and A' average)

37


38

Dave: likes Option1 (weighted averages first). Want to find the systematic

39

relative to the final results, not the intermediate results

40


41

Julie: you should take the fullfit A as the answer, and use the (AA')

42

difference from sinefit as a systematic

43

 Tanja prefers this too

44


45

Garth: since each of the cut study uncertainties include the statistical

46

uncertainties resulting in the respective fits, we will likely have to remove

47

the raw statistical uncertainty in quadrature from the systematic coming from

48

the fits

49

 stats do not enter in (AA'), but it does in the others

50


51

Plotting a systematic error band, or double error bars (as Hall B does)?

52


53

Dave: prefers double error bars

54

 a systematic error band implies that all points move togther, while they

55

probably move independently

56


57

Plot points at mean t or central t?

58

 agree that mean t is better, since the BSA results are not bin centered

59


60

Q2: How to calculate error due to cut dependence?

61

delta(C)= [ ACN+ACW ]/2

62

delta(M)= [ AMN+AMW ]/2

63

 Dave: this is how he does it

64

 Nacer: many experiments give the RMS of 2,3 variations

65

 Garth: likes delta(C), delta(M) separately for errors discussion, but we

66

need to give the community an easily understandable way to use our BSA

67

data. For this, suggest to take RMS of (AA', CN, ..., MW results) as the

68

systematic.

69


70

Beam Polarization Error:

71

 all points should move together

72

 since this is not the dominant uncertainty (smaller than above systematic),

73

just add it in quadrature with the others

74


75

Good Consensus:

76

 show A as result

77

 include RMS of all different variations as a systematic

78

 show as a double error bar

79


80


81


82


83

