PyROOT problem in minuit minimization

ROOT
,
1

Hello all!

Does anyone know what does it mean when, after running the simplex and migrad minimization , what happens if it stops without ending and the correlation coefficient matrix is all zero?

Thank you in advance!

my last lines:

     FCN=0 FROM MIGRAD    STATUS=CONVERGED      76 CALLS          98 TOTAL
                     EDM=0    STRATEGY= 1      ERROR MATRIX ACCURATE
  EXT PARAMETER                                   STEP         FIRST
  NO.   NAME      VALUE            ERROR          SIZE      DERIVATIVE
   2  AN1          3.50000e+00   3.39038e+00   4.00187e-03 -4.33187e-307
   3  B1           3.00000e-01   1.55896e+00   1.11311e-03 -1.55740e-306
   4  D01          5.00000e-01   6.99781e-01   2.02450e-03 -8.56291e-307
   5  AN2          0.00000e+00   constant
   6  B2           1.00000e-02   6.48821e+00   3.18387e-04 -5.44482e-306
   7  D02          5.00000e-01   5.55649e-01   2.12858e-02 -8.14421e-308
   8  ANg          0.00000e+00   constant
   9  Bg           0.00000e+00   constant
  10  D0g          1.00000e+00   1.49338e-02   1.00167e-02 -1.73066e-307
 EXTERNAL ERROR MATRIX.    NDIM=  25    NPAR=  6    ERR DEF=1
 2.886e+305  0.000e+00  0.000e+00  0.000e+00  0.000e+00  0.000e+00
  0.000e+00 2.885e+303  0.000e+00  0.000e+00  0.000e+00  0.000e+00
  0.000e+00  0.000e+00 2.885e+303  0.000e+00  0.000e+00  0.000e+00
  0.000e+00  0.000e+00  0.000e+00 2.892e+301  0.000e+00  0.000e+00
  0.000e+00  0.000e+00  0.000e+00  0.000e+00 2.929e+305  0.000e+00
  0.000e+00  0.000e+00  0.000e+00  0.000e+00  0.000e+00 2.894e+301
 PARAMETER  CORRELATION COEFFICIENTS
       NO.  GLOBAL      2      3      4      6      7     10
        2  0.00000   0.000  0.000  0.000  0.000  0.000  0.000
        3  0.00000   0.000  0.000  0.000  0.000  0.000  0.000
        4  0.00000   0.000  0.000  0.000  0.000  0.000  0.000
        6  0.00000   0.000  0.000  0.000  0.000  0.000  0.000
        7  0.00000   0.000  0.000  0.000  0.000  0.000  0.000
       10  0.00000   0.000  0.000  0.000  0.000  0.000  0.000

and it doesn’t show anything else!

2

The error matrix has blown up, see the diagonal entries with values > 1E300. Do you have a reproducer? And @moneta might have additional insights!

3

It looks to me that both the gradient and the second derivatives are pratically zero. This means that you have a problem in the function definition you have provided to Minuit

Lorenzo

4

I understand there is a problem at the definition of the fcn function. Do you know what should i consider more when writing in python? Because the script in c++ is running correctly. Are there any examples in PyRoot with Minuit? The only thing i found is this similar example : http://www-ekp.physik.uni-karlsruhe.de/~quast/Skripte/diving_into_ROOT.pdf ( page 50) and it doesn’t run correctly also, it doesn’t fit at all.

Thanks for the help again! I really apreciate it!

5

Hi,
You should not use directly the TMinuit class from Python, but use the Minimizer interface.
You can probably easily convert this C++ example to Python:
https://root.cern.ch/doc/master/NumericalMinimization_8C.html

If you need help with this, please let me know

Lorenzo

6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.