xfitter merge requestshttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests2023-06-15T17:39:58+02:00https://gitlab.cern.ch/fitters/xfitter/-/merge_requests/297Dyturbo interface v1.32023-06-15T17:39:58+02:00Stefano CamardaDyturbo interface v1.3Updated dyturbo interface for version 1.3.0Updated dyturbo interface for version 1.3.0Stefano CamardaStefano Camardahttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/292Chebyschev standalone implementation2023-06-02T11:11:14+02:00Simone AmorosoChebyschev standalone implementationhttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/295Update install-lapack to 3.11.02023-03-27T16:20:57+02:00Simone AmorosoUpdate install-lapack to 3.11.0 - Install LAPACK version in install-lapack to 3.11.0
- Support also curl to download the package - Install LAPACK version in install-lapack to 3.11.0
- Support also curl to download the packagehttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/294Update lhapdf to 6.5.42023-03-27T16:20:28+02:00Simone AmorosoUpdate lhapdf to 6.5.4- update LHAPDF version in install-xfitter to 6.5.4- update LHAPDF version in install-xfitter to 6.5.4https://gitlab.cern.ch/fitters/xfitter/-/merge_requests/276Improve ceres2022-10-05T17:21:54+02:00Stefano CamardaImprove ceresGeneral improvements of ceres minimisation:
- Log penalty terms
- Cost function scaling
- Covariance and error estimates
- Improved output
LOG PENALTY TERMS
In the current implementation, the ceres cost function does not include the l...General improvements of ceres minimisation:
- Log penalty terms
- Cost function scaling
- Covariance and error estimates
- Improved output
LOG PENALTY TERMS
In the current implementation, the ceres cost function does not include the log penalty terms of the chi-square, leading to an incorrect minimisation when the option Chi2ExtraParam = 'PoissonCorr' is enabled in steering.txt.
The problem with the log penalty terms is that they can give a negative contribution to the chi-square, which would represent a complex residual in a least squares minimisation cost function.
In order to include the log penalty terms, I have added an additional set of Ndata residuals. A constant offset is added to these residuals, so that negative log terms do not lead to complex residuals.
In order to get the correct value of the cost function, a loss function is implemented which subtracts the constant offset which was added to the residuals of the log penalty terms.
COST FUNCTION SCALING
The cost function of ceres is normalised as 1/2 * ||fi||. I have added a factor of sqrt(2) to the residuals so that the ceres cost function exactly matches the chi-square value of FCN. This should give a more correct estimate of the parameters error.
COVARIANCE MATRIX
The covariance matrix calculation and error estimates for the fitted parameters are added. The global correlation coefficient, as estimated by MINUIT, is also added.
IMPROVED OUTPUT
Standard output is improved, and results are saved to parsout_0 and ceres.out.txtStefano CamardaStefano Camardahttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/275Additive option for RT scaling of FL and F22022-09-01T15:28:41+02:00Stefano CamardaAdditive option for RT scaling of FL and F2The chi-square minimum of the default NNLO fit with RTOPT as VFN scheme shows some hyperbolic discontinuities.
The reason of such discontinuities is the scaling of the RT correction to FL from the pure gamma contribution to the sum of ga...The chi-square minimum of the default NNLO fit with RTOPT as VFN scheme shows some hyperbolic discontinuities.
The reason of such discontinuities is the scaling of the RT correction to FL from the pure gamma contribution to the sum of gamma, Z and gamma/Z interference terms. Such scaling is ill-defined when the massless FL cross zeros, and gives raise to hyperbolic discontinuities which can generate false minima and not positive-definite covariance matrix at the minimum.
As a temporary solution, I have added an option which implements and additive correction. For consistency the same additive corrections is implemented also for F2. A better solution would consist in evaluating FL and F2 separately for u and d quarks, and compute the gamma, Z and gamma/Z terms from the u and d quark contributions.
The discontinuities at minimum can be checked in the test: temp/checkmin-default/output/minuit.out.txt [default-minuit.out.txt](/uploads/ca95a3f57852ae07b9bd886fb336d83f/default-minuit.out.txt)
The smooth minimum with the additive correction is in temp/checkmin-additive/output/minuit.out.txt [additive-minuit.out.txt](/uploads/57da4864fb29a5023e65549f99b7144e/additive-minuit.out.txt)Stefano CamardaStefano Camardahttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/278Use q2 grid limits, and not q20 to cut away data2022-09-01T14:19:33+02:00Alexander GlazovUse q2 grid limits, and not q20 to cut away dataFix bug observed by @amoroso (do not skip predictions for data between q2minGrid and q20 )Fix bug observed by @amoroso (do not skip predictions for data between q2minGrid and q20 )Alexander GlazovAlexander Glazovhttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/279Update applgrid to 1.6.32, remove long tests from the CI2022-08-31T15:53:15+02:00Simone AmorosoUpdate applgrid to 1.6.32, remove long tests from the CI - Update APPLgrid to 1.6.31 in install-xfitter
- By default do not install DYTURBO
- ZPT and ZMVFNS-fit tests are now not run by default
- Remove download of unused PDFs in the CI image - Update APPLgrid to 1.6.31 in install-xfitter
- By default do not install DYTURBO
- ZPT and ZMVFNS-fit tests are now not run by default
- Remove download of unused PDFs in the CI imageSimone AmorosoSimone Amorosohttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/272Re-enable code to produce files for LHAPDF reweighting2022-08-23T14:13:57+02:00Alexander GlazovRe-enable code to produce files for LHAPDF reweightingAlexander GlazovAlexander Glazovhttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/274Update fastnlo reaction to read NNLO grids2022-08-17T15:37:06+02:00Simone AmorosoUpdate fastnlo reaction to read NNLO grids- Update fastnlo toolkit to 2.5.0-2826, supports new NNLO jet grids.- Update fastnlo toolkit to 2.5.0-2826, supports new NNLO jet grids.Simone AmorosoSimone Amorosohttps://gitlab.cern.ch/fitters/xfitter/-/merge_requests/172Interface to DYTurbo2022-08-07T12:55:07+02:00Stefano CamardaInterface to DYTurboAdded new reaction which provides interface to DYTurbo.
A CMake find script to link DYTurbo as a library is included.
The parameters of the non-perturbative form factor can be fitted or profiled
An example/test for fitting the CDF Z pT...Added new reaction which provides interface to DYTurbo.
A CMake find script to link DYTurbo as a library is included.
The parameters of the non-perturbative form factor can be fitted or profiled
An example/test for fitting the CDF Z pT measurement is added, and the install-xfitter script is updated to install dyturbo.
TODO:
- merge xfitter-datafiles zptcdf branch into master
- some cleanup (ReactionAPPLgrid.cc, install-xfitter, etc...)