Pyroot: passing function to minimizer class

[afylot]

---

ROOT Version: 6.16.00
Platform: CentOS Linux release 7.6.1810
Compiler: gcc (GCC) 4.8.5 20150623

---

I want to use the Class ROOT::Math::Minimizer in Python. I would need to define in python the Functor class in order to pass the function to the minimizer. I have written some code taking testpyfit.py from How to pass fcn func defined on python side to ROOT.Fitter? as inspiration

Unfortunately the ROOT::TPyMultiGenFunction is quite different from ROOT::Math::Functor and my code fails with

     32     myFunctor_instance = MyFunctor()
---> 33     minimum.SetFunction(myFunctor_instance)
     34     minimum.SetVariable(0, "x0", x0_p, error_x0_p)
     35     minimum.SetVariable(0, "x1", x1_p, error_x1_p)

TypeError: none of the 2 overloaded methods succeeded. Full details:
  attempt to access a null-pointer
  void ROOT::Math::Minimizer::SetFunction(const ROOT::Math::IGradientFunctionMultiDimTempl<double>& func) =>
    could not convert argument 1

import ROOT
def main():
    npar=2
    x0_p = -2.0
    x1_p = 1.0
    x2_p = 0.5
    error_x0_p = 0.05
    error_x1_p = 0.2
    error_x2_p = 0.1

    minimum = ROOT.Math.Factory.CreateMinimizer("Minuit2", "")
    # minimum.SetMaxFunctionCalls(1000000)
    # minimum.SetMaxIterations(10000)
    # minimum.SetTolerance(0.001)
    # minimum.SetPrintLevel(1)

    class MyFunctor( ROOT.Math.Functor ):
        def __init__( self ):
            ROOT.Math.Functor.__init__( self )
            pass

        def NDim( self ):
            print 'PYTHON MyFunctor::NDim called'
            return npar

        def DoEval( self, x ):
            #ret = chi2func(x[0], x[1], x[2])
            ret = x[0] * x[0] + x[1] * x[1] + x[2] * x[2]
            print 'PYTHON Functor::DoEval val=', ret
            return ret
    
    myFunctor_instance = MyFunctor()
    minimum.SetFunction(myFunctor_instance)
    minimum.SetVariable(0, "x0", x0_p, error_x0_p)
    minimum.SetVariable(0, "x1", x1_p, error_x1_p)
    minimum.SetVariable(0, "x2", x2_p, error_x2_p)
    minimum.Minimize()

    return