v1.6.0.0
VanyaBelyaev
released this
15 Mar 10:06
·
2160 commits
to master
since this release
v1.6.0.0
New features
- Add
FlattePS_pdf
- similar toBWPS_pdf
- Make few important steps towards ROOT 6.23/01 ("New PYROOT"). Full adaptation is not yet achieved, there are some pending problems with effective inheritance from C++ classes (namely
TSelector
,PyPDF
,PyVAR
, ...). There are also some puzzling crashes... - rename tests, make test selection more transparent and easy to navigate back
- rename
ostap/fitting/selectors.py
toostap/fitting/pyselectors.py
to avoid the name clash forpython3
- add helper script
pplaunch
to launch remote pp-servers via ssh tunnels - update
PyVar
,PyVar2
,PyPdf
,PyPdf2
- (almost) complete update for new PyROOT
- fix
test_fitting_minuit_weighted
- thanks to Dima Golubkov - make more coherent treatment of ROOT issues
- fix
minuit
for new PyROOT (signature ofFCN
is different!) - minor update for
minuit
: from now allow access by parameter name:
minuit = ...
minuit['p2'] = 10
minuit.minos ('p1','p2',...)
minuit.release ('p3')
- update
pptunnel
+pplaunch
with better and more informative output - add
Ostap::Math::BW3L
,Ostap::Models::BW3L
,BW3L_pdf
and extend testostap/fitting/tests/tests_fitting_breiwigner.py
- resurrected version of the Breit-Wigner profile from 3-body decays - add check for
more_itertools
, provide replacement forchunked
whenmore_itertools
is not available - update
Ostap::Math::NSphere
andOstap::Math::Positive
such that for null-parameters the reusltin poisitve function is a constant. The trick is based on properties of Chebyshev polynomials of 1st,. 2nd, 3rd and 4th kind. ostap.logger.table
add parameteralignment
that specifies the column alignment.ostap.parallel.task
add parameterbatch
, that allows to execute the tasks inbatch
mode- re-add generic Breit-wigner channel
Ostap::Math::ChannelGeneric
(for ROOT>=6.23/01 only) - add method
amplitude
forOstap::Math::ChannelBW
- re-remove generic Breit-wigner channel
Ostap::Math::ChannelGeneric
- change the default
sample
argument forPDF.generate
fromFalse
toTrue
- tiny fix for the table column alignment
- add
#include <string>
forNSphere.h
- for certain configuration it prevents compile error (thanks to Abdul-Kerim Gusseinov) - Add parameter
accept_fun
fortoys
- that allows to (re)define the accepance criteria, the default corresponds toaccept_fit
function fromostap.fitting.toys
module, that checks the fit status (0) and covariance matrix status ( 3 or -1) - Add parameter
fit_fun
fortoys
- that allows to (re)define the default "fit"-policy - Add parameter
gen_fun
fortoys
- that allows to (re)define the default "generation"-policy test_plotting_summary_graph.py
: add call forROOT.gPad.RedrawAxis
- thanks to Tom Blake- add
**kwargs
for allparallel
-methods, arguments are used forWorkManager
- remove
evaluateBatch
form all PDFs (folloiimng evolution of ROOT). We need to gradually introduceevaluateSpan
Ostap::Math::GammaBW3
: use 1/s factor instead of 1/s^3/2. thanks to Misha Mikhasenko!- Add symmetic Sinh-Asinh resolution model
ResoSinhAsinh
- Add symmetic Jonhson's SU resolution model
ResoJohnsonSU
- fix for the new signature of
TDirectory::CurrentDirectory()
method - fix/rewrite/improve
ostap.histos.compare
module - improve reweighting machinery: make it more tunable (and less automatic) and more suitable for multidimensional reweighting.
- Add new context manager
SETPARS
and use it inPDF.wilks
,PDF.wilks2
,PDF.draw_nll
,PDF.graph_nll
,PDF.graph_profile
- Add logistic/"sech-squared" resolution model
ResoLogistic
- Improve
PDG.graph_profile/PDF.graph_nll
: adddraw
argument to draw the graph in progress - Better (but not perfect yet) treatment/assignement of the unique names for many intermediate objects
Backward incompatible changes:
- Reweighting machinery: different signature of
makeWeights
function - new argumentmake_plots
, different meaning of argumentpower
, different return value
Bug fixes:
- fix bugs in
Ostap::Math::BWPS