==== Extrapolation and reducing the ndof ====
Finished my studies on reducing the number of fit parameters. Found the best function as 2 gaussians + landau.
=== 3 gaussians + landau ===
Expression: {{ :users:danyyl:dijet:screen_shot_2018-05-18_at_15.14.43.png?nolink |}}
== 300 GeV ==
{{ :users:danyyl:dijet:300_4f.png?nolink |}}
''chi_square =  202.423391874/188.0  =  1.07672016954''
== 400 GeV ==
{{ :users:danyyl:dijet:400_4f.png?nolink |}} 
''chi_square =  254.599910266/188.0  =  1.35425484184''
== 500 GeV == 
(Actually, I forced all functions to have non-zero area. You can see that the third gaussian (orange) has very small sigma and non-zero area and describes only 2 points, that has no sense. However, if I free the lower bound of the normalisation constant (area) to zero, third gaussian disappear. So, the 4th function is redundant. This was one of the reason why I started to find the way to fit the signal with at most 3 functions.)
{{ :users:danyyl:dijet:500_4f.png?nolink |}} 
''chi_square =  455.035302783/188.0  =  2.42040054672''
== 600 GeV == 
{{ :users:danyyl:dijet:600_4f.png?nolink |}}
''chi_square =  606.474750337/188.0  =  3.22592952307''

=== 2 gausians + landau ===
Expression:{{ :users:danyyl:dijet:screen_shot_2018-05-18_at_15.17.49.png?nolink |}}
== 300 GeV ==
{{ :users:danyyl:dijet:300_3f.png?nolink |}}
''chi_square =  342.1863381/191.0  =  1.79155150838''
== 400 GeV ==
{{ :users:danyyl:dijet:400_3f.png?nolink |}}
''chi_square =  349.860731707/191.0  =  1.83173157962''
== 500 GeV ==
{{ :users:danyyl:dijet:500_3f.png?nolink |}}
''chi_square =  457.792421761/191.0  =  2.3968189621''
== 600 GeV ==
{{ :users:danyyl:dijet:600_3f.png?nolink |}}
''chi_square =  657.261829338/191.0  =  3.44116141014''

The difference in the chi square lies within 1. So 4 function doesn't improve the fit too much wrt the 3 functions.

=== Extrapolation ===
Using the extrapolation formula {{ :users:danyyl:dijet:form.jpg?nolink |}}I obtained the following results.
== Mass of the 1st resonance is 300 GeV, mass of the 2nd is 500 GeV ==
Extrapolation for 300 GeV (should be the same as fit). On X axis ''mjj/Mres'', where Mres is the mass of the resonance (300 GeV). On Y axis - ''Events / 10 Gev''.
{{ :users:danyyl:dijet:300_tocheck_m1res.png?nolink&nolink }}
Extrapolation for 400 GeV. On X axis ''mjj/Mres'', where Mres is the mass of the resonance (400 GeV). On Y axis - ''Events / 10 Gev''. Looks not so bad as I expected.
{{ :users:danyyl:dijet:400_exrapol_m12res_300_500.png?nolink&nolink }}
Extrapolation for 500 GeV (should be the same as fit). On X axis ''mjj/Mres'', where Mres is the mass of the resonance (500 GeV). On Y axis - ''Events / 10 Gev''.
{{ :users:danyyl:dijet:500_tocheck_m2res.png?nolink&nolink }}

== Mass of the 1st resonance is 400 GeV, mass of the 2nd is 600 GeV ==
Extrapolation for 500 GeV using fit functions for 400 and 600 GeV:{{ :users:danyyl:dijet:500_extrapol_m12res_400_600.jpg?nolink |}}

=== Next steps ===
  - Automise the code to produce the extrapolation function for mass from 400 to 600 GeV with step of 10 GeV.
  - Have a look to the README Javier sent us about how to perform the fit