User function abd NumConv function at RooFit

[babintsev]

Happy New Year all,
To the my previous question there is new one

ex-myNumConv.C (24.7 KB)
Now there are 2 problems -

1. plot of the input data is very strange - histogram reproduce
   data - 4-th gpad, but function in RooFit has strange shape for
   these data - 2-ns gpad
2. I started to use RooNumConvPdf for simple input function -
   myf3() - BW and gauss() - this time convolution works
   (unlike of FFT method) BUT very strange also -
   at the plot - gpad 1 and 3 - there is drop of points at t=4.6 - 5.6
   What does it mean !? where is the problem ?
   thanks,
   Vladimir

---

root -l
-> .L ex-myNumConv.C
-> Convolution(); >& l.l
.> …

[jalopezg]

Hi @babintsev,

Happy New Year! I am inviting @moneta to this thread as he may be able to help with this.

Cheers,
J.

[babintsev]

Dear Lorenzo,
I am trying to use RooFit methods to get convoluted crosssection
of neutron resonancies. Results:

1. FFT method does not work for me I do not know why
2. try NumConvPdf - strange result !

• I use ex-myNumConv.C : read input data file : data-1-10.txt
  (in fact I read full data: but it is too big, so use short variant)

• I draw 4 wimdows:
  w-1: data simulation using convolution fxg - you can see
  a lot of missing regions
  w-2: draw data-function which I used for convolution ->
  (in other words - it is data, which I read)
  you see that plot shows values ONLY in the region 3-5 eV
  –> compare this with full data sample at the w-4 !

  w-3: here I draw result of convolution :
  BLUE - initial BW resonance (the same as at w-2)
  RED - result of numeric convolution : cut at E= 4.6 eV !

so, questions : is it problem with plot procedure !?
OR is it problem with RooFit methods ?
OR it is my error --> where and what ?!

 2 topics are here :
            1) w-2 and w-4 :  why plot does not reproduce input data ?
	2) I use simple variant:
	    take BW resonance and gauss - as it is from examples
	                                  at Internet
	    - w-1  data sample using this convolution (fxg)
	    - w-3  convolution  fxg - you see cuts in region
              (it is the same problem, when I use my input data)

thank you,
Vladimir, my mail adress: babintsev@ihep.ru (Protvino, RUSSIA)

ex-myNumConv.C (24.7 KB) data-1-10.txt (5.8 KB)

[moneta]

Thank you Vladmir for the explanations. I will have a look, but it will take me same time, since it is not an easy answer.
Is this then a duplicate of RooFit import histogram plot ?

Cheers

Lorenzo

[babintsev]

… thank you, there are some modifications from
previous one : I worked with my own data,
and the last picture : using simple BW and Gauss
as a part of the question…

Vladimir.