- Fix pip dependencies
- First public Python package release
- NLO DVEM works now (rho0 only)
- Removed (temporarily) NNLO and neural net stuff
- First beta with proper distributable Python package structure
- Added GNU AGPL-3.0 license.
- ADD: Merged pype into gepard
- CHG: Moved to git and gitlab.com
- ADD: DVEM coefficient functions (just starting, DVEM not usable yet)
- ADD: GPD E
- FIX: All test for "big" NPB paper (2008) fixed
- ADD: Evaluation of gamma function is also controlled by
- SPEED, saving significant time in fitting
- ADD: Compiles also with Intel ifort, NAG f95 and GNU gfortran compilers,
- including OpenMP parallelization.
- ADD: python extension pygepard.so created
- ADD: third SO(3) partial wave to 'FIT' model
- ADD: Evolution of GPDs at eta=0 and eta=x
- ADD: Possibility of eta-dependent ansatz
- ADD: Mathematica interface completely redone so both
- calculation of CFFs and fitting is possible from
- within Mathematica
- ADD: Evolved Wilson coefficients can be now calculated in
- the initialization phase which leads to much faster
- execution
- FIX: comparison to PEGASUS is OK at NNLO as well
- FIX: SPEED=3 should be used instead of SPEED=4 now
- ADD: non-singlet evolution
- ADD: non-singlet NNLO
- ADD: MSbar non-diagonal evolution
- ADD: intelligent choice of y-range for drawing of PDF(x)
- ADD: many example programs in ex subdirectory which produce
- data for plots in a big paper
- ADD: automatic plotting and psfragging of plots
- FIX: dependence on factorization scale corrected now
- BUG: SPEED=4 is much more unreliable than before
- FIX: charge-factor bugs (it was not NF-dependent, and was
- altogher missing from DVCS)
- ADD: PDF(x), resulting from fit, are also drawn now
- ADD: Additional experimental data points for fitting
- Fixed F2(P) bug that forced P=1 in DIS case
- Added possibility of Gaussian integration on 2**ACC points
- with new parameter ACC=1..6.
- Added test routines 'houches' and 'accuracy' for checking
- accuracy w.r.t. Les Houches benchmark and w.r.t referent most precise gepard result
- Wilson coefficients calculated in initialization phase and
- together with anomalous dimensions put into common block
- Integration moved from slatec to 8-point Gaussian
- Adacf part integrated into source
- PGPLOT support added
- Compiles cleanly on both Linux and Cygwin
- Additional user-friendly documentation in gepard.pdf
- Initial version
- Imported into subversion versioning system
- Target gepard.exe is broken (so calling from Mathematica
- doesn't work) but "pure Fortran" targets work and produce data equivalent with Figs 1 and 2 of Letter
- Numerical integration improved by subdivision of path:
- LAM..-1/2..1/2..LAM
- Some old release
- Some old release