P2VVPhysFit

1
P2VV/PhysFit
8 months later

• Gerhard Raven
• Juan Palacios
• Tristan du Pree
• Nikhef
• ßs and mixing WG

2
P?VV reminder
Bs?J/?f Bd?J/?K(Kp-) Bs?ff

• 4 observables 3 angles (O) time
• Normalized sum of 24 terms
• Coefficients cij cij(A,d,fs,q,w)
• Functions hi(t) hi(t,?Gs,?ms, Res(t,t))

3
WANTED
From talk at software week 2 Oct 2007
now have

• We would like a
• Fast (analytically integratable)
• Generally applicable
• Publicly usable
• Fit model, including
• Angular acceptance correction
• Proper time resolution
• Tagging
• Reusing existing code as much as possible
• Sharing new code and methods for all angular
  analyses

Difficulty was normalized sum over
not-normalized, possibly negative terms
4
PSVV sum pdf RooRealSumPdf(coeffs,funcs)
Use RooRealSumPdf!
Coefficients cijcij(A,d,fs,w,q)
Functions RooProducts fi(O)hj(t)
Time dependence hj(t)Res.convolution(exp)
Angle dependence AnglesAbsFunc fi(O)
Angle bases -TransAngles -HelAngles
Resolution RooGaussModel RooGexpModel RooAddModel
Exponents RooFormulaVars expsinh, expcos
5
The new things we madeand which are in CVS

• AnglesAbsFunc
• Abstract base class for angular functions
• Specific goal angular acceptance correction
• Concrete implementation of formulas
• TransFi(ThreeAngles)
• HelFi(ThreeAngles)
• Some classes to organize the package
• ThreeAngles, TwoAngles, OneAngle
• AngularAcceptanceCorrectionSet
• Similarly want to group Ai, and strong phases
• Finally PdfMakerFunctions

• In P2VVTHREEANGLESFFSS
• Bs ? J/? f
• Bd ? J/? K(Kp-)

6
How to build my own P2VV pdf

• Example 1
• B ? KK
• B ? f f

• In P2VVTHREEANGLESSSSS
• Make new HelTransFi classes
• Write down the values and integrals of the
  functions, just like in FFSS
• evaluate()
• analyticalIntegral()
• Copy where possible the functions from FFSS
• Find a way to reuse existing ones
• Build a pdf similar as in PdfMakerFunctions()
• Writing down all the correct parameters/products

7
How to build my own P2VV pdf

• Example 2
• B ? K µ µ

• In P2VVTHREEANGLESFFSS
• In transversity basis same angular functions as
  in Bs?J/?f
• Dont have to write any new angle classes
• Write out amplitudes Ai(s) analytically
• See basis functions sfj(s)?
• If not in RooFit write new classes
• RooProduct(transfi(O)sfj(s))
• Multiply all these products with the correct
  coefficients and sum these

hep-ph/0205254, p.21
s dependence instead of t
Have to determine angular acceptance corrections
as a function of s
8
How to play

• Open fresh shell
• unsetenv User_release_area
• mkdir cmtP2VV
• cd cmtP2VV
• setenv User_release_area PWD
• setenv CMTPROJECTPATH PWDLHCBPROJECTPATH
• setenvLHCb v23r6

• cd /cmtP2VV/LHCb_v23r6
• getpack PhysFit/P2VVPython head
• cd PhysFit/P2VVPython/v1r0/cmt
• gmake

• getpack PhysFit/P2VV head
• cd PhysFit/P2VV/v1r1/cmt
• gmake

• cd /cmtP2VV/LHCb_v23r6/PhysFit/P2VVPython/v1r0/py
  thon
• SetupProject LHCb v23r6
• python i testMakePdf.py

9
Some details on Bs?J/?f

• For flexibility of the code
• One has to put in general CPV parameters Df, Sf
• User can define e.g. Sf(sin(fs),), Df(cos(fs),)
  prior to passing to pdfmaker
• The maker function also does not fix SiAi21
• Will do something similar for the strong phases
• Might want to put in variables sinphi_par and
  cosphi_perp instead of the phases itself
• There might be a more stable fit parametrization
  than the one the maker of the pdf (here me)
  comes up with!
• and phases are in general not so stable

10
Warning

• The example works
• if you set read0 in the script
• Butit is just an example
• Not the definite version for J/?f J/?K
• Remarks/suggestions are very welcome
• tdupree_at_nikhef.nl
• Buta lot of things already on to do list

11
Checks to do

• Check angular functions
• Plot projections
• Calculate integrals
• Compare with numerical
• Check pdf
• Compare with Edinburgh/Marseille/CERN/
• Generate with one pdf, fit with another
• Can already be done now!
• By anybody

12
Some things to add

• Use same pdf maker for Bs?J/?f and Bd?J/?K(Kp-)
  with different configurations
• Functions to
• Determine angular acceptance correction
• Make acceptance corrected plots
• Bugless RooEffProd available from 25 June!
• Add resolution/acceptance to toy data
• Convert helicity?transversity amplitudes
• Would be nice to
• Plot CP-even/odd components
• Generate toy data faster
• Documentation!

13
Some possible extensions

• Pdfs
• J/?K(Ksp0), KK, ff, Kµµ, Kf
• Pdf J/?f bad EvtGen DC06 (see backup)
• Background
• Early measurements
• ONEANGLE
• J/?(nS) ?(nS) polarization (namespace V2FF?)
• Effects
• Angular resolution
• Proper time acceptance

14
Conclusions

• We have a method to build general P2VV angular
  pdfs, including angular acceptance corrections
• The most important code is in CVS now and there
  is a simple method to use it
• With the python script you can just start to play
• Everything is there to extend it to other
  channels
• KK ff relatively straightforward
• Kµµ less trivial
• Many functionalities are known how to add, but
  work needs to be done

Remarks/suggestions/requests welcome
15
BACKUP
16
Howto according to J.A.

• Open fresh shell
• unsetenv User_release_area
• mkdir cmtP2VV
• cd cmtP2VV
• setenv User_release_area PWD
• setenv CMTPROJECTPATH User_release_areaLHCb_
  release_areaGaudi_release_areaLCG_release_a
  rea
• setenvLHCb v23r6

• cd /cmtP2VV
• getpack PhysFit/P2VV head
• getpack PhysFit/P2VVPython head
• cd PhysFit/P2VVPython/v1r0/cmt
• Source setup.csh
• Cmt br selectUser_release_area global gmake

• cd /cmtP2VV/LHCB_v23r6PhysFit/P2VVPython/v1r0/pyt
  hon
• SetupProject LHCb v23r6
• Python i testMakePdf.py

17
Comparing old new EvtGen

• New
• Old
• So sign beta wrong
• Underestimated half of the amplitudes by factor
  ei?Gt
• Bottom line
• old version correct for leading orders of ?G and
  fs

18
Decay rates in old EvtGenwrong amplitudes used
for DC06

• Same as for J/?K(Ksp0) except for GL, GH
• Will perform toy study to see systematic effect

a.o. sign beta_s wrong
19
Good formulas Jpsiphi(for comparison)