TMVA crashing: Helge, you have another bug

TMVA
1

Hi Folks,

I am just starting (learning) how to use BDTs with TMVA. I did a test with a small sample of background and data. That seemed to work ok. So I threw much bigger trees at the code and it is crashing out with a wired error message

— TFHandler_BDT : -----------------------------------------------------------------------------------------
— TFHandler_BDT : Variable Mean RMS [ Min Max ]
— TFHandler_BDT : -----------------------------------------------------------------------------------------
— TFHandler_BDT : MLR: 0.58518 10.889 [ -99.000 11.738 ]
— TFHandler_BDT : MWR: 0.32689 10.851 [ -99.000 9.9587 ]
— TFHandler_BDT : MSCW: 0.76385 11.196 [ -99.000 90.979 ]
— TFHandler_BDT : MSCL: 0.41075 10.983 [ -99.000 79.023 ]
— TFHandler_BDT : EmissionHeight: 16.308 22.326 [ -99.000 6281.8 ]
— TFHandler_BDT : SizeSecondMax: 1107.0 4368.3 [ 150.00 1.0066e+06 ]
— TFHandler_BDT : theta2Calc: 3.2553 1553.3 [ 4.1362e-08 1.6739e+06 ]
— TFHandler_BDT : dist2Calc: 104.42 2181.6 [ 0.16345 2.1152e+06 ]
— TFHandler_BDT : -----------------------------------------------------------------------------------------
— BDT : Begin training
— BDT : Training 400 Decision Trees … patience please
— DecisionTree : Helge, you have another bug …nSelS+nSelB= 1.71779e+06 while total number of events = 1.70045e+06
***> abort program execution

Using root 5.24/00

Any ideas?

2

Hi,

I recommend you try with ROOT 5.32. If this does not work, you should consider reporting this issue in the Math and Stat forum.

Cheers,
Philippe.