0. 
   Verify your background model (is it reasonably flat??)
   root -l <inputfile.root>
   .L tricks.C
   background(<range>,lowerp,upperp) - try several momentum intervals choose the 
   fit-range as large as possible and make sure that the bg is actually reasonably flat

1. 
   Determine the width of the broad component in the signal from simulation.
   For this you need to fit the signal in increasingly large fit-ranges to
   see the monotonous rise in the background-width result. 

   Use the macro tricks.C:
   root -l <inputfile.root>
   .L tricks.C
   loop();

   move the output to a directory of your choice
   Have a look at the output files named "LargeWidthVsRange_<xxx>_to_<yyy>.txt".
   These should show a monotonous rise of the wide component of the correlation
   with fit range and almost no dependence on momentum. The fit is done sector 
   independent to allow for better statistics

 2.
   Pick the mean value for the wide range and fix this parameter for the macro
   tricks.C when fitting the next time. This time you fit the signal width assuming a
   fixed background width - again this is not done sector-wise
   
   root -l <inputfile.root>
   .L tricks.C
   loop(<fixed width>);

   move the result to a directory of your choice
   Verify that the result of the fit for the narrow component of the distribution
   is not affected by the fit range anymore after the width of the broad component
   is fixed! The result can be found in:"SmallWidthVsRange_<xxx>_to_<yyy>.txt".
   Of course the result will become more narrow with increasing momentum!


2.b
   Check whether the free (non-constrained) result for the large width
   determined from the correlated background agrees reasonably with the one 
   from the wide signal component

   root -l <inputfile.root>
   .L tricks.C
   loop(0.0,2)


2c.
   Look at the small component  of the correlated background. Find out what 
   the variation with fit range is after fixing the broad component. Check
   that the narrow components of signal and correlated background are 
   comparable

   root -l <inputfile.root>
   .L tricks.C
   loop(<fixedwidth>,2)


 3. 
   Fit signal+background together 
   
   root -l <inputfile.root>
   .L tricks.C
   loop(<fixedwidth>,1)

   move the result to a directory of your choice
   Verify that the signal widths reproduced for the small component are as close as
   you need to the ones from the fit to the pure signal.
   If this is the case the paramtrization of signal and background is OK and you can 
   proceed to the production of parameters in experiment and simulation using the tools 
   forseen for this purpose.

4.
   Test whether the result of the fit of the experimental data is also robust against variation 
   of the fit range (especially the width of the narrow signal-component). Restrict the experimental
   analysis to one sector like this:

   root -l <exp-inutfile.root>      
   .L tricks.C
   loop(<fixedwidth>,1,<sector>)

5. Verify that the simulated widths and the experimentally determined widths agree reasonably.