CKMfitter  
About CKMfitter The Paper The Code Contact
C K M    F I T T E R
Mirror Site for Europe: http://ckmfitter.in2p3.fr
Contents

Home
What is CKMfitter
Fit Inputs
Plots & Results
Talks & Writeups
The CKMfitter Working GroupWho is CKMfitter
Contact: CKMfitter




Visitors since June 11, 2001

 

Using the CkmFitter Framework Package
Last update - 03 March 2006

OVERVIEW


The Ckmfitter software is publicly available at this location (latest update: 20 may 2006). Winter 2007 update of the code is available Here (18 June 2007) (untar it after downloading the main code).

IMPORTANT NOTICE: In order to distinguish between (official) plots done by the CKMfitter group and unoffical plots done with the package available above, it is asked to label "package" under the CKMfitter logo on the unoffical plots. Thanks.

This page intends to provide help in producing standard plots with the package using the inputs |Vub|, |Vcb|, epsilonK, DeltaMd, DeltaMs and sin2beta (as can be found here ). Other plots can be produced modifying the default datacards provided on this web page, and help can be obtained by contacting the authors.

The minimal system configuration required to run the software is:

  • FORTRAN 77 and C++ compilers
  • HBOOK / PAW
[ Everything you wanted to know about PAW (Tutorial, FAQ, etc.): HERE (PAW is part of the CERNLIB package available HERE)]

IMPORTANT: CERNLIB and RFIO problem The default version of the "packlib" library since 2004 in the CERNLIB is compiled with the "shift" library. It exists a version compiled without the "shift" library: "packlib_noshift". If you encounter problems during compilation of the CKMfitter package, this is certainly related to thisCERNLIB problem. To solve it, edit the Makfile and change the line:
CERNLIBBAS = -lmathlib -lpacklib
with
CERNLIBBAS = -lmathlib -lpacklib_noshift
and then the usual "gmake".

SETUP AND RUN THE CODE

  • Copy CkmFitter.uu in a working directory, and execute:
    "./CkmFitter.uu".
    (if the file doesn't execute automatically, type "chmod u+x CkmFitter.uu", and try again)

  • Compile the code (in the "src" directory):
    gmake (or "make -f Makefile" if the first command doesn't work)
    followed by
    gmake install
    to install the binary in the "bin" directory.

  • To produce the global fit in the (rho-eta) plane (region around the apex of the triangle in the (rho-eta) plot), type (in the root directory):
    ./bin/runckm datacards/ckm_all.data datacards/ckm_flags.data

    [Each constraint in the (rho-eta) plane corresponds to a dedicated datacard. See the "datacard" directory.]

    The script "runckm" calls the CKMfitter binary (CkmFitterApp). It is driven by two types of datacards (located in the "datacards" directory): the first type (like "ckm_all.data") contains the fit ingredients, and the second type (like "ckm_flags.data") contains the fit options. More details on the content of these datacards are given below.

  • Once the code stops running, you can plot the results in PAW by typing in PAW the command (the file "ckmglobal.kumac" is in the "kumac" directory):
    exec ckmglobal#plot_rhoeta ,

    Change its content to your needs (especially the location of your hbook files)


    FIT INGREDIENTS

  • The fit ingredients are specified in the datacard ckm_all.data (containing all ingredients of the standard fit). The name of the output HBOOK file is also specified in this datacard.

  • A standard (rho-eta) plot is composed by various constraints :
    - the global fit using all constraints (obtained by "ckm_all.data")
    - individual constraints obtained by the files "ckm_vub.data", "ckm_epsk.data", etc.

  • Each ingredient is characterized by 4 quantities:
    $Param
    name = 'sin2beta' --> gives the name of the ingredient
    value = 0.731, 0.056, 0.0 --> value +- Gaussian error +- flat error
    TakeMeIn = T --> whether it should be used in the fit
    Free = F --> whether it should be let floated in the fit
    $end

    The measurement should have the values "TakeMeIn = T" and "Free = F", whereas the theoretical parameters should have the values "TakeMeIn = T" and "Free = T".

  • There is one special setting to choose in this datacard, called "UseOneNDof", which is used to specify the number of degree of freedom of the fit. Two cases are possible:
    - you use at least 2 ingredients both depending on (rho-eta): UseOneNDof=F
    - you use only 1 ingredient depending on (rho-eta): UseOneNDof=T

  • The other setting in this datacard is called "$constrainMeasurements". It is used when performing 1-D fits, or 2-D fit in other space than the standard (rho-eta) space. Set the "constrUse" to 1 for the variables that should be used in these non-standard fits, and to 0 for the other variables.

    FIT OPTIONS

    The default flag datacard, ckm_flags_all.data, allows to realize a frequentist fit in the 2-D (rho-eta) plane, using a rough binning of 50 x 50 (less CPU-consuming for a first test of the fit, but which should be increased to 200 x 200 for final plots). To realize most of the standard fits, only a few switches need to be modified, as specified below. For more complicated fits (for example, related to charmless decays), other switches not mentioned below should be modified. Please contact the authors for more details. The basic switches to know are the following:

  • UseImprove and UseMinos are used to improve the convergence of the fit. They are CPU consuming (especially the second one), but allows to obtain completely converged results (not-well converged results show irregularities).
  • FitLambda, FitA, FitRhoBar, FitEtaBar: to let the Wolfenstein parameters floating or not.
  • ScanType2D: to choose between the four 2-D standard fits in the (rho,eta), (sin2alpha, sin2beta), (sin2alpha, gamma), (sin2beta, gamma) planes.
  • Granularity: granularity of the scan (good quality is 200 x 200)
  • ScanDimensions: gives the minima and maxima of the 2D fits.
  • NumberOfFits: used when UseImprove=T: number of fits performed at each point to improve the convergence. The default is 1, and you can go up to 6 when the convergence is bad (3 is a good compromise).
  • DoTraceParameters: perform a 2-D frequentist standard fit
  • DoConstrainMeas1D: perform a 1-D frequentist fit
  • DoConstrainMeas2D: perform a 2-D frequentist non-standard fit (other than the choices in ScanType2D)
  • DoTraceChi2Min: set to true to evaluate the CL of the fit
    		•  
     
    Webmaster of this site: Stephane T'Jampens
    Last update: March 3, 2006