Often oen can add soft Gaussian constraint for soem fit parameters, e.g. one can constraing the rsignal resolution:
sigma_MC = VE( 0.015 , 0.001**2 ) ##
sigma_cnt = model.sigma.constrainTo ( sigma_MC , 'sigma_constraint')
my_constraints = ROOT.RooFit.ExternalConstraints ( ROOT.RooArgSet ( sigma_cnt ) )
dataset = ...
model.fitTo ( dataset , ... , constraints = my_constraints , .... ) Clearly several constraints can be combined togather
sigma_cnt = model.sigma.constrainTo ( sigma_MC , 'sigma_constraint')
peak_cnt = model.mean .constrainTo ( VE(3.096,0.001**2) , 'mass_constraint' )
my_constraints = ROOT.RooFit.ExternalConstraints ( ROOT.RooArgSet ( sigma_cnt , peak_cnt ) ) For the next version of ostap, one will be able to avoid the explicit creation of
ROOT.RooFit.ExternalConstraint and ROOT.RooArgSet
sigma_cnt = model.sigma.constrainTo ( sigma_MC , 'sigma_constraint')
peak_cnt = model.mean .constrainTo ( VE(3.096,0.001**2) , 'mass_constraint' )
model.fitTo ( dataset , ... , constraints = ( sigma_cnt , peak_cnt ) , .... )