Dear experts,
I’m trying the Offset=‘initial’ option in my fit, but it seems the output FNC value does not change. The attached script includes a simple Gaussian fit to show the effect.
The fit results with Offset=False:
RooFitResult: minimized FCN value: 7454.93, estimated distance to minimum: 0.000293861
covariance matrix quality: Full, accurate covariance matrix
Status : MINIMIZE=0 HESSE=0
Floating Parameter FinalValue +/- Error
-------------------- --------------------------
mean 1.0030e+02 +/- 2.25e-01
sigma 1.0064e+01 +/- 1.59e-01
with Offset=‘initial’:
RooFitResult: minimized FCN value: 7454.93, estimated distance to minimum: 0.000293861
covariance matrix quality: Full, accurate covariance matrix
Status : MINIMIZE=0 HESSE=0
Floating Parameter FinalValue +/- Error
-------------------- --------------------------
mean 1.0030e+02 +/- 2.25e-01
sigma 1.0064e+01 +/- 1.59e-01
with Offset=‘bin’:
RooFitResult: minimized FCN value: 36.277, estimated distance to minimum: 1.30573e-10
covariance matrix quality: Full, accurate covariance matrix
Status : MINIMIZE=0 HESSE=0
Floating Parameter FinalValue +/- Error
-------------------- --------------------------
mean 1.0030e+02 +/- 2.25e-01
sigma 1.0060e+01 +/- 1.59e-01
My questions are:
- Is this expected? Is the
Offset=’initialtaking effect? - Is the
binoption for binned fit? Or could it be used for unbinned fit?
I’m using ROOT 6.36.04 on CentOS Stream 9. Thanks!
BTW, in an old version of ROOT (6.24.06/x86_64-centos7-gcc11-opt), it seems to be working as I expect:
RooFitResult: minimized FCN value: -0.990198, estimated distance to minimum: 7.02851e-07
covariance matrix quality: Full, accurate covariance matrix
Status : MINIMIZE=0 HESSE=0
Floating Parameter FinalValue +/- Error
-------------------- --------------------------
mean 1.0030e+02 +/- 2.25e-01
sigma 1.0060e+01 +/- 1.59e-01
test1.py (359 Bytes)