MnMigrad() return c++ seg-fault in PyROOT

ROOT
,
1

I want to use ROOT.Minuit2 for fitting data. However, when I execute MnMigrad() in PyROOT, Jupyter console displays c++ segmentation fault error. How to fix under my code ?

class MyFCN(Minuit2.FCNBase):
    def __init__(self):
        pass
    def __call__(self, params):
        return 1.

fcn = MyFCN()
param = Minuit2.MnUserParameters()
param.Add("test1",0.01, 0.01)
param.Add("test2",0.01, 0.01)
param.Add("test3",0.01, 0.01)
migrad = Minuit2.MnMigrad(fcn, param,3)
migrad()

Error massage is here.

SegmentationViolation Traceback (most recent call last)
/var/folders/gf/kt47jhy56f7ch38t9ft4hhfm0000gn/T/ipykernel_19839/3223852618.py in
9 param.Add(“test3”,0.01, 0.01)
10 migrad = Minuit2.MnMigrad(fcn, param,3)
—> 11 migrad()
SegmentationViolation: ROOT::Minuit2::FunctionMinimum ROOT::Minuit2::MnApplication::operator()(unsigned int maxfcn = 0, double tolerance = 0.10000000000000001) =>
SegmentationViolation: segfault in C++; program state was reset

ROOT Version: JupyROOT 6.22/08
Platform: macOS Catalina 10.15.7

2

@moneta do you see any problem in this code? Is this how one would do this in C++? Is there any alternative?

3

Hi,
It should work in C++, see for example root/DemoGaussSim.cxx at master · root-project/root · GitHub.

As a workaround I would use the Minimizer interface, see the tutorial
https://root.cern.ch/doc/master/NumericalMinimization_8C.html

Lorenzo

4

I think you need to at least implement the Up method, which is pure virtual in FCNBase. Not sure if there are other required methods as well.

Cheers,
Enrico