Releases: OstapHEP/ostap
Releases · OstapHEP/ostap
v1.13.7.0
v1.13.7.0
New features
- add
RooDataSet
->TTree
transformation - add generalized Beta' distribution
- add
isfinite
method forSE
&WSE
counters data_statistics
: add the check for the finite counters- add
isfinite
method for matrices and vectors - add
isfinite
method for(W)Covariance
objects - add
Ostap::MoreRooFit::Hypot
andvar_hypot
- add Fizher's Z-distribution
- explicitely delete all created styles, see here
- add missing methods for
Ostap::Math::Bessel
- fix occasional overflows in
twosamples.py
- add new variant of toys using generic actions
Backward incompatible
Bug fixes
- fix
ostap-check-dependencies
- fix bug in
Ostap::Math::Benini
v1.13.6.0
v1.13.6.0
New features
- add more methods to
Ostap::Math::ECDF
to get the ranking - optimise
Ostap::Math::ECDF
and skip sorting when possible - add
ostap.stats.twosamples.py
to make Two Sampel Tests - add the test for Two Sample Test
- reduce code duplication between GoF-1D and Two-Sample Tests
- Add Anderson-Darling and Cramer-von-Mises Two Sample Tests
- Add progress bar to methdos dealing with unique/duplicated entries in dataset
- Add Hilbert transform
- Add Laplace transform
Backward incompatible
Bug fixes
v1.13.5.0
New features
- Add
operator+
forOstap::Math.ECDF
- Add correlation coefficient to
Ostap::Math::pow
function - Slight improvements in convolution for fitting
- Add the first Goodness-of-Fit method for multidimensional case
- Add test for Goddness-of-Fit method for multidimensional case
- Add very simple "efficiency-counter"
ostap.stats.counters.EffCounter
- suppress
ostap.core.config.config_goodby
prints for non-interactive sessions - add the most primitive splitter
ostap.utils.utils.splitter
- modify
point-to-point-dissimilarity
GoF method: split into chubnks for large datasets, use parallel processing for permutations - add keyword arguments to
WorkManager.iexecute
method - add an option to run parallel permutations using
WorkManager
parallelisation instead ofjoblib
- extend Point-to-Point Dissimilarity GoF test for 1D case
- extend 1D GoF test to include Point-to-Point Dissimilarity GoF test
- add argument
description
toWorkManager.iexecute
methods - add parallelisaiton for GoF permutations and toys
- implement tests for USTAT & DNN
- prepend the default progress-bar for trees/datasets/frames with
Entries:
- add a kind of replacement of
ROOT.RooAbsCollection.assign
for old versions of ROOT - add meaningful
description
argument to allprogress_bar
instance - extend
gof1d
andgofnd
tests - Add
RooAddPdf::fixCoefNormalization( vars )
for all appearences ofRooAddPdf
- Add (fictive) context mnager protocol to WorkManagers
Backward incompatible
Bug fixes
- fix the bug in
Ostap::UStat
v1.13.4.0
v1.13.4.0
New features
- Add
cuts
andcut_range
argumenst fords2numpy
function - Add empirical cumulative distribtion function
Ostap::Math::ECDF
- Add
cuts
andcut_range
arguments fords2numpy
method - Add method to
RooDataSet
to get empirical cumulative distrivution functions - Add
PDF1.cdf
method to get CDF from 1D-PDF - Add new module
ostap.stats.gof_1d
for goodness-of-fit for 1D - Add new test
test_stats+_gof1d
for goodness-of-fit for 1D ostap.math.math_ve
: addsignificace
,nsigmas
&nsigma
functions to calcualte significabce from p-values- Make use of
ostap.math.math_ve.significane
function - Rename
gof_1d
togof1d
- Add method
weight
toOstap::PyIterator
to access th weigth of the current event - add weight to
tree.withCut
and other tree-looping methods - Tiny tweak for treatment of drawing ranges for
f1_draw/f2_draw/f3_draw
fnuctions: form now the explicit setting has a precedence. ostap.stats.gof1d
: improve drawing methods- Add Kuiper's Goodness-of-Fit estimator to
ostap.stats.gof1d
- Add
values
iterator firnctoonfor all hjistogram classes to iterate over th ebin values - Add methods
table
andsummaary
for 1D-histogram to dump a sumaary of the histogram in the table form - Add methdod
all_positive
,all_negative
,all_nonpositive
,all_nonnegative
andall_zero
for hisgoragm clases to check if
all bins are positive/negative/non-positive/non-begrtaive/zero. - Add methods
table
andsummmary
for 2D&3D-hisograms
Backward incompatible
- add
weight
totree.withCut
and other tree-looping methods
Bug fixes
- fix typo in
ostap.math.bnase.axis_range
v1.13.3.0
v1.13.3.0
New features
- some updtae for catcing warnins
- ROOT warnings -> Python warning now used RuntimeWarning
- add new test
test_math_convolution
- add
filter='data'
argument fortarfile
(TMVA&Chopping) - set
cmake_policy
to fic nuilding for virtual environemnts (conda) - more improvements and fixes for
ostap.io.dbase
module - add test for availabe DB backends
- improve
ostap.histos.histos.book_histo
and make use of it infitting
Backward incompatible
- From now
PDF3.fitTo
always returns the tuple of(fit-result, frame)
or(fit-result,None)
to be coherent with all otherfitTo
methods.
Bug fixes
v1.13.2.0
v1.13.2.0
New features
- Further improvemets for nice printout of linear algebra objects
- Add
lnorm
andmnorm
methods for matrices to get L-norm and max-norm - Add
gdiagonal
method to get diagonal matrice - Add
abs
method forSVectors
- Improve Linear Algebra tests
- Improve the machinery for eigenvalues and eigenvectors
- Add possibility to avoid coloring of the header row in the tables
- Add protection for double cpoulmn (
::
) for the expression strings - Resurrect and improve
ostap.stats.corr2d
module with simple 2D-decorrelation utility - Make use of
terminaltables3
instead ofterminaltables
where/when possible - Add new method
TH1.min_positive
to get minimal positive entry (or negative infinity). - make a try to relace
distutilts
tosysconfig
inCMakeList.txt
- some reshuffling of
CMakeList.txt
- make it a bit more readable ostap.math.base
: switch for C++ version offrext10
(better treatment ofalmost-zero-numbers
test_*_toys
: make use of newTH1.mini_positive
method
Backward incompatible
Bug fixes
- Fix tiny bug in
histo_book
- Fix tiny bug in printout of linear algebra objects
v1.13.1.0
v1.13.1.0
New features
- improve all toys machinery: toys, jackknife, bootstrapping and significance
- extend all tests for toys
- add brute-force way to delete
RooDataSet
- needed for JAckknife and Bootstrap - add argument
delete=False
fordataset.bootstrap
anddataset.jackknife
method to delete the dataset - add new test
test_fitting_dataset2.py
to test interference of memory and jackknife/bootstraping - extend and improve machinery for toys
- add
std.string
andstd.string_view
intostring_types
- improve
attention
printout fromdataset/tree
methodproject
Backward incompatible
Bug fixes
- fix the typo in
ostap/tools/tests/test_tools_reweight3.py
- tiny fix for
progress_bar
- tiny fix for
ostap.utils.basic.isatty
v1.13.0.2
v1.13.0.2
New features
- improve pretty-print for matrices
- add
pretty_array
function for nice print of arrays/sequences - add some auxillary methods for matrices:
(min/max/minabs/maxabs)_element
and(min/max)_diagonal
and(min/max/minabs/maxabs)_element_index
- improve printout of
SVectorwithError
- add
pos_error/neg_error/errors
properteis toOstap::Math::ValueWithError
object ` - add column
@limit?
for printput ofRooFitResult
object to shwo the distance to the limits (if any). Distanc eof <3sigma, an d<5 sigma ar ecolored. Distances > 10sigma are omitted.
Backward incompatible
Bug fixes
- fix the typo in
Ostap/MatrixUtilsT.h
- fix
ostap.utils.basic.mtime
v.1.13.0.0.
v1.13.0.0
New features
- sight improvements for
bernstein
- disable Thiele rational interpolation from the tests. Sometnjug wrong with the code.
- extend a bit functionality of asymmetric errors (needed for graphs&plots)
- collect pretty-print functions into new module
ostap.logger.pretty
- extend and improve pretty-print stuff
- make the dbase.Item a bit lighter
- For shelve-like databases: allow to specify the preferred backend as
dbtype = ...
- reimplement
SqliteShelf
asZipShelf
withdbtype='sqlite'
- eliminate messy stuff with extensions for `XXXShelf``-classes
Backward incompatible
Bug fixes
- fix a minor bug in
bernstein.solve
- fix couple of recent bugs in `histos
v1.12.0.0
v1.12.0.0
New features
- Add estimators for harmonic, geometric, power & Lehmer means and their weighted analogues
- Reduce code duplication
- Large redesign of staistics/projection& othe rmethids for RooAbdData/TTree/DataFrame
- Large redesign if
statvars.py
module - Add
roc_curve
for making ROC curves, and corrresponsing test module - Add
eff_graph
for 1D historgams for creation of the efficiency graph
from the 1D-distribution. - Some tweaks for moments & counters
- Activate a new
draw
method (viatree_draw
) forROOT.TTree
- add
progress
andreport
optioal argumens for (almost) all Frame-related functions - Some tweaks for style configuration
- update
ostap.utils.valerrors
& and new test - allow to use
width
keyword whenline_width
is not specified forXXX.draw
method - add
loop
methdod forRooAbsData
and implementrows
in terms ofloop
- allow more recusion in
vars_and_cuts
function - add new test
- make progress bar silent if
not isatty()
unless explicitely setsilent=False
Backward incompatibl
project
(&draw
) for 2 and 3-dimession now follows the natural order of varibales:
XX.project ( target , 'x,y,z' , ...)
- For
eff
& effic' andefficinecy
methods fo r1D histograms
the confusing optional argumentincreasing=True
is replced by (less-confusin)
cut_low
and the argument is not optionl anymore - From now for weighted datasets
dataset[i]
returns(entry,weight)
tuple - from now iteration over weighted dataset gives
(entry,weight)
tuple - change sinature of
dataset.loop
,dataset.rows
methods to return tripletsindex, entry, weight
Bug fixes
- fix a typo in
ostap.ploting.canvas