#include "AliTMinuitToolkit.h"
#include "TVectorD.h"
#include "TMatrixD.h"
#include "TTreeStream.h"
#include "TCanvas.h"
#include "TH1.h"
#include "TF1.h"
#include "TRandom.h"
#include "TStatToolkit.h"
#include "TLegend.h"
#include "TStyle.h"
Go to the source code of this file.
| void AliTMinuitToolkitTestLinear |
( |
Int_t |
nIter, |
|
|
Int_t |
nDraw |
|
) |
| |
test of AliTMinuiToolkit class - linear fits
/
- For the moment: Test only that code does not crash
- Example to run (300 times)
1 .x $AliRoot_SRC/STAT/test/AliTMinuitToolkitTestLinear.C+(300,3)
- Test of linear fit for data with outliers
- Compare fits:
- 0.) Standard linear pol1 fit
- 1.) "robust" pol1 fit
- 2.) Standard pol1 minuit fit
- 3.) AliTMinuitToolkit with gaus+cauchy likelihood
- 4.) "robust" AliTMinuitToolkit with Huber cost function
- 5.) Bootstrap with 20 iteration
- 6.) Bootstrap with 100 iteration
- 7.) MISAC with 200 iteration
- Input data: pol1 + gaussian noise with tails (random fraction: 20 %, 20 % wider)
1 pol1+ ((rmdm>x) ? rndmG :20*rndmG)
- Test example result:
- example canvas showing fit quality for few random samples
- Canvas with residual histograms for each fit option
- misac option - smallest fraction of outlier
- robust option - smallest sigma (but some outlier)
- standard chi2 minimization (~factor 5-10 times wider than other "robust" method
- Example usage in the $AliRoot_SRC/STAT/test/
- TODO - save example images in the doxygen/html directory (STEERING part ?) - currently saved by hand - to check with OFFLINE
- $AliRoot_SRC/STAT/test/AliTMinuitToolkitTest.C+
- $AliRoot_SRC/STAT/test/AliTMinuitToolkitTestLinear.C+
- compare fits example with different likelihood/resp fits strategies for data with outliers
- compare performance fits example with different likelihood/resp fits strategies for data with outliers
Definition at line 47 of file AliTMinuitToolkitTestLinear.C.
1.8.11