Openturns

=== Library ===

==== Major changes ====
* HSIC sensitivity indices
* RandomWalkMetropolisHastings now updates all components at a time, previously updated componentwise
* Iterative statistics

==== New classes ====
* RandomVectorMetropolisHastings
* IndependentMetropolisHastings
* Gibbs
* HSICEstimatorConditionalSensitivity
* HSICEstimatorGlobalSensitivity
* HSICEstimatorTargetSensitivity
* HSICUStat
* HSICVStat
* IterativeExtrema
* IterativeMoments
* IterativeThresholdExceedance
* TensorProductExperiment
* NAIS
* Pagmo
* GalambosCopula

==== API changes ====
* Removed deprecated Hanning alias
* Removed deprecated AdaptiveDirectionalSampling alias
* Removed deprecated VisualTest::DrawCobWeb method
* Removed deprecated shims module
* Removed deprecated TBB.SetNumberOfThreads/GetNumberOfThreads
* Removed deprecated MultiFORM.setMaximumNumberOfDesignPoints/getMaximumNumberOfDesignPoints
* Removed deprecated SubsetSampling.getNumberOfSteps
* coupling_tools.execute: deprecated get_stderr,get_stdout for capture_output, returns CompletedProcess
* Deprecated Nlopt|Ceres|Bonmin|CMinpack|Ipopt|TBB|HMatrixFactory::IsAvailable in favor of PlatformInfo::HasFeature, see also GetFeatures
* RandomWalkMetropolisHastings API breaks to split the likelihood definition
* Deprecated CalibrationStrategy class

=== Documentation ===

=== Python module ===
* Hide C++ getImplementation methods and add Python-specific getImplementation methods with automatic dynamic typecasting
* Add Sample.asDataFrame/BuildFromDataFrame pandas conversion methods

=== Miscellaneous ===
* Add Collection::select(Indices) method
* Parallelized SymbolicFunction evaluation (enabled from a sample size>100 with exprtk)
* Sample csv export: allow one to set precision and format
* Sample csv import: handle nan/inf values
* Add EnumerateFunction::getBasisSizeFromTotalDegree(maximumDegree)

=== Bug fixes ===
* 1260 (RandomWalkMetropolisHastings does not handle non-symmetric proposals properly)
* 1263 (Static functions are implemented in C++ classes instead of namespaces)
* 1334 (Multi-objective optimization)
* 1444 (The calibration examples are unclear)
* 1830 (Incompatible documentation for CauchyModel)
* 1914 (Cobyla & NLopt should accept LeastSquaresProblem)
* 1921 (Adaptive Directional Stratification: the coefficient of variation of the probability is inconsistent with repeated probability results)
* 1922 (Adaptive Directional Stratification: the coefficient of variation cannot be used as a stopping criterion)
* 1926 (Examples have broken graphics)
* 1931 (Interface class getImplementation methods are useless in Python)
* 1939 (Contour curves outside graph window)
* 1940 (getMaximumDegreeStrataIndex has a bug with hyperbolic rule)
* 1943 (TriangularMatrix * Point --> ScalarCollection)
* 1946 (Doc search issues)
* 1947 (The constructor of FunctionalChaosResult is undocumented)
* 1949 (The comparison operator of ComposedDistribution is wrong)
* 1958 (Performance of OrthogonalUniVariatePolynomial (precision, speed))
* 1959 (Cobyla run causes segfault when empty ctor is used)
* 1993 (IsotropicCovarianceModel has no default ctor)
* 1994 (default GaussProductExperiment is invalid)
* 1997 (KLSVD spectrum cutoff error)
* 2012 (Cannot sample a ConditionedGaussianProcess on a mesh containing conditioning points)
* 2014 (Precision issue with the Python KrigingAlgorithm test)
* 2019 (Parametric stationary covariance models doc section needs refresh)
* 2020 (Problem with the UserDefined distribution)
* 2025 (The lack of Bonmin generates an error during example compilation)
* 2031 (Multi-objective optimization example question)
* 2032 (Documentation of NAIS)

=== Library ===

==== Major changes ====

==== New classes ====
* DistanceToDomainFunction

==== API changes ====
* Removed deprecated MultiStart::setStartingPoints/getStartingPoints
* Removed deprecated KarhunenLoeveResult::getEigenValues
* Removed deprecated coupling_tools.execute is_shell/workdir/shell_exe/check_exit_code arguments
* RandomVector::get/setParameter and getParameterDescription available to PythonRandomVector
* Implemented CovarianceModel::computeCrossCovariance
* Deprecated Hanning in favor of Hann
* Deprecated AdaptiveDirectionalSampling in favor of AdaptiveDirectionalStratification
* Deprecated VisualTest::DrawCobWeb in favor of DrawParallelCoordinates
* IndicatorFunction constructor takes a Domain as input
* Intervals and DomainUnions get new method computeDistanceToDomain(Point or Sample)
* Renamed some ResourceMap keys: cache-max-size>Cache-MaxSize, parallel-threads>TBB-ThreadsNumber
* Deprecated TBB.SetNumberOfThreads/GetNumberOfThreads
* Deprecated MultiFORM.setMaximumNumberOfDesignPoints/getMaximumNumberOfDesignPoints
* Deprecated SubsetSampling.getNumberOfSteps
* Normal and Student distributions: no need to specify the R parameter if it is the identity matrix
* KrigingResult::getConditionalMean and getConditionalMarginalVariance yield Samples instead of Points when applied to Samples
* Deprecated shims module

=== Documentation ===

=== Python module ===

=== Miscellaneous ===

=== Bug fixes ===
* 1840 (HMatrixFactory leaks)
* 1842 (libOT.so.0.0.0 doesn't have a SONAME)
* 1843 (Ali-Mikhail-Haq copula parameter value)
* 1844 (MaximumDistribution::computePDF is wrong when all the marginals are equal and independent)
* 1845 (MemoizeFunction does not propagate to finite difference gradient&hessian)
* 1847 (ParametricFunction require unnecessary function evaluations)
* 1854 (Sample indexing does not work on np.int64 type)
* 1856 (Speed up Normal computeSurvivalFunction)
* 1857 (ProductCovarianceModel fails when constructed with DiracCovarianceModel)
* 1858 (Normal::computeCDF(sample) crash for large dimensions)
* 1861 (Still some instabilities in Kriging with ot.StationaryFunctionalCovarianceModel)
* 1864 (FORM - IMPORTANCE FACTOR)
* 1868 (Not compatible with hmat-oss-1.7.1: no member named 'compressionMethod' in 'hmat_settings_t')
* 1870 (TruncatedDistribution fails to compute quantiles)
* 1874 (ProcessSample.getSampleAtVertex() may be useful as a public method)
* 1877 (How to model singular multivariate distributions?)
* 1878 (Rename Hanning filtering window)
* 1879 (Adaptive Directional Sampling Algorithm: the drawProbabilityConvergence graph may be wrong)
* 1880 (Adaptive Directional Sampling: some enhancements proposed)
* 1882 (Is Distribution::getLinearCorrelation useful ?)
* 1883 (Strange behavior of FORM ?)
* 1884 (setDesign can lead to wrong Sobol' indices)
* 1891 (Correlation of Halton sequence at high dimensions)
* 1911 (DirectionalSampling freezes python if not correctly initialized)
* 1912 (splitter: missing doc)
* 1915 (computePDFGradient(const Sample &) should rely on computePDFGradient(const Point &) in DistributionImplementation)
* 1918 (PythonDistribution does not allow one to overload the isDiscrete() method)

=== Library ===

==== Major changes ====

==== New classes ====
* VertexValuePointToFieldFunction
* KarhunenLoeveReduction
* KarhunenLoeveValidation
* IsotropicCovarianceModel
* KroneckerCovarianceModel
* VonMisesFactory
* KFoldSplitter, LeaveOneOutSplitter

==== API changes ====
* Removed deprecated Pairs alias
* Removed deprecated SORMResult::getEventProbabilityHohenBichler/getGeneralisedReliabilityIndexHohenBichler
* Removed deprecated OptimizationResult::getLagrangeMultipliers, OptimizationAlgorithm::computeLagrangeMultipliers
* Removed deprecated FittingTest::Kolmogorov, FittingTest::BestModelKolmogorov (DistributionFactory argument only)
* Deprecated Sample::computeStandardDeviationPerComponent, use computeStandardDeviation
* Deprecated KarhunenLoeveResult::getEigenValues, use getEigenvalues
* Removed StationaryCovarianceModel
* CovarianceModel.computeAsScalar allows scalars
* Swapped SimulatedAnnealingLHS constructor SpaceFilling/TemperatureProfile arguments
* Added EfficientGlobalOptimization::getKrigingResult, gets the updated version of the KrigingResult passed to the constructor
* Deprecated MultiStart::setStartingPoints/getStartingPoints, use MultiStart::setStartingSample/getStartingSample instead
* MultiStart::setStartingPoint/getStartingPoint throws (previously did nothing)

=== Documentation ===
* Fixed example and plot of Kolmogorov statistics.
* Added a new example showing how to combine RandomWalkMetropolisHastings and PythonDistribution

=== Python module ===
* Serialize Python wrapper objects using dill (PythonDistribution, PythonFunction ...)

=== Miscellaneous ===

=== Bug fixes ===
* 1010 (The doc for ExpectationSimulationAlgorithm is confusing)
* 1052 (The Dirichlet and Normal distributions only have 1D graphics in the doc)
* 1224 (The examples in the doc of the DomainEvent class are unclear)
* 1229 (Better encapsulation of optional 3rd-party headers)
* 1240 (There is no theory documentation for ExpectationSimulationAlgorithm)
* 1257 (Cannot create a SimulatedAnnealingLHS without specifying the temperature profile)
* 1287 (The constructors of KernelSmoothing have no doc)
* 1418 (The Contour example does not present the second constructor)
* 1425 (There is no method to create a train / test pair)
* 1431 (The figure for LHSExperiment is not accurate)
* 1459 (There is no example which shows how to set a column of a Sample)
* 1497 (There is no example to create a multivariate Normal distribution)
* 1506 (The Brent class has no example)
* 1570 (Wrong formula for MauntzKucherenkoSensitivityAlgorithm)
* 1650 (There is no example of a parametric StationaryFunctionalCovarianceModel)
* 1656 (The help page of ExpectationSimulationAlgorithm is unclear)
* 1661 (XMLH5StorageManager does not store IndicesCollection into HDF5 files)
* 1662 (Operator() of a CovarianceModel with multidimensional output should yield object of type SquareMatrix)
* 1680 (QuadraticBasisFactory multiplies cross products by 2)
* 1710 (Misleading y labels in VisualTest.DrawPairsMarginals())
* 1713 (The doc of the ResourceMap does not match the content of the ResourceMap)
* 1714 (The log-PDF of the Pareto is wrong)
* 1721 (The `draw` method of `SobolSimulationAlgorithm` does not reuse the descriptions)
* 1723 (There is no interesting example of the ExprTk feature of SymbolicFunction)
* 1725 (The BetaFactory help page has wrong equations)
* 1729 (LinearModelAnalysis sometimes fail)
* 1731 (The Extreme value example may be improved)
* 1737 (Low rank tensor doc issues)
* 1742 (The example which shows how to set the figure size is wrong.)
* 1751 (DrawCorrelationCoefficients has a wrong Text height)
* 1752 (Error while estimating the reduced log-likelihood when using a StationaryFunctionalCovarianceModel)
* 1758 (The LinearModelStepwiseAlgorithm has no example)
* 1759 (Kriging model with StationaryFunctionalCovarianceModel might provide bad results)
* 1768 (Bug in ExponentiallyDampedCosineModel & SphericalModel)
* 1771 (GridLayout hides the titles in the graph)
* 1772 (MinimumVolumeClassifier cannot draw 1D samples)
* 1774 (ExponentialModel::partialGradient is wrong)
* 1775 (ExponentialModel does not account correlation with Covariance ctor)
* 1776 (Typo in the Branin use case implementation)
* 1781 (The link_to_an_external_code example has small bugs)
* 1784 (The drawPDF method of Histogram sometimes fail)
* 1787 (CovarianceMatrix from SymmetricMatrix raises InvalidArgumentException)
* 1790 (LinearModelResult.getFormula() method is not updated by Stepwise regression)
* 1794 (Some doc examples seem to behave with new infrastructure)
* 1796 (VonMisesFactory is missing)
* 1803 (The API example of Experiment has a format issue)
* 1805 (Brent's implementation has a stability issue)
* 1807 (Beta::computeCharacteristicFunction is wrong if the lower bound is not zero)
* 1815 (1.16 fails with dlib-cpp-19.22)
* 1818 (Problem with wilksNumber)
* 1820 (Incoherent results in LinearModelAnalysis)
* 1835 (RegularizedIncompleteBeta returns nan)
* 1836 (Hypergeometric results differ without boost)
* 1837 (Poor performance when using pickle on OT objects)

=== Library ===

==== Major changes ====
* Drop normalization in KrigingAlgorithm
* Drop normalization & transformation handling in GeneralLinearModelAlgorithm
* XML/H5 storage (hdf5 library)
* C++11 requirement

==== New classes ====
* BlockIndependentDistribution
* FejerAlgorithm
* GridLayout
* MinimumVolumeClassifier
* StationaryFunctionalCovarianceModel
* XMLH5StorageManager

==== API changes ====
* Removed deprecated Weibull, WeibullFactory, WeibullMuSigma classes
* Removed deprecated GumbelAB class
* Removed deprecated Event class
* Removed deprecated EnumerateFunction constructors
* Removed various deprecated distribution accessors
* Deprecated Pairs class, see VisualTest.DrawPairs
* Deprecated SORMResult::getEventProbabilityHohenBichler, use SORMResult::getEventProbabilityHohenbichler instead
* Deprecated SORMResult::getGeneralisedReliabilityIndexHohenBichler, use SORMResult::getGeneralisedReliabilityIndexHohenbichler instead
* Renamed SobolSequence::MaximumNumberOfDimension as SobolSequence::MaximumDimension
* Added VisualTest::DrawLinearModel(linearModelResult), useful if the test is performed on the training samples
* Added VisualTest::DrawLinearModelResidual(linearModelResult), useful if the test is performed on the training samples
* Deprecated OptimizationResult::getLagrangeMultipliers
* Moved OptimizationAlgorithm::computeLagrangeMultipliers to OptimizationResult
* Added AIC & BestModelAIC static methods in FittingTest
* Added AICC & BestModelAICC static methods in FittingTest
* Moved BuildDistribution from FunctionalChaosAlgorithm to MetaModelAlgorithm
* Added Drawable::BuildRainbowPalette(size)
* Added Drawable::BuildTableauPalette(size), which is now the default palette.
* Added Drawable::ConvertFromRGBIntoHSV
* Added FittingTest::Lilliefors, BestModelLilliefors
* Deprecated FittingTest::BestModelKolmogorov(Sample, DistributionFactoryCollection, TestResult), use BestModelLilliefors
* Deprecated FittingTest::Kolmogorov(Sample, DistributionFactory, TestResult, level), use Lilliefors
* MetamodelValidation: now computePredictivityFactor returns Point, drawValidation return GridLayout
* Deprecated coupling_tools.execute is_shell/workdir/shell_exe/check_exit_code arguments

=== Documentation ===
* Sphinx-gallery used to render examples

==== API documentation ====
* Clarified SobolIndicesExperiment page, notations now consistent with SobolIndicesAlgorithm page
* Clarified SobolIndicesAlgorithm and (Saltellli|Martinez|MauntzKucherenko|Jansen)SensitivityAlgorithm pages, corrected formulas
* Documented how to turn warnings off or write them on a file

=== Python module ===
* Renamed Viewer *_kwargs arguments to *_kw (matplotlib convention)
* Add ProcessSample Field accessors

=== Miscellaneous ===
* Do not compute Lagrange multipliers by default during an optimization
* Add ResourceMap::FindKeys
* Allow computeLogPDF methods to output values lower than SpecFunc::LogMinScalar

=== Bug fixes ===
* 1001 (Add method SobolSimulationResult::draw)
* 1259 (The diagonal of a scatter plot matrix should have the histograms)
* 1267 (Some CSV files cannot be imported)
* 1377 (The `setKnownParameter` method is not compatible with `buildEstimator`)
* 1407 (GeneralLinearModelAlgorithm mishandles user-specified scale parameter when normalize is True)
* 1415 (The BuildDistribution static method should not use the KS-test)
* 1421 (UserDefinedStationaryCovarianceModel doc suggests input dimension can be >1)
* 1436 (The style of the curves is unpleasing to my eyes)
* 1447 (Highly inaccurate result in reliability model when using subset of RandomVector)
* 1465 (The Sample constructor based on a list and an integer should not exist)
* 1470 (setNbModes is sometimes ignored)
* 1474 (optimization defaults)
* 1507 (Leak in Collection typemaps)
* 1510 (ot.Ceres('LEVENBERG_MARQUARDT') and ot.Ceres('DOGLEG') do not handle bound constraints)
* 1515 (KernelSmoothing build failure)
* 1520 (The NLopt test is dubious)
* 1521 (Basis of MonomialFunction)
* 1529 (The error of the NonLinearLeastSquaresCalibration and GaussianNonLinearCalibration are different)
* 1540 (SubsetSampling: incorrect event sample)
* 1547 (Mesh does not check the simplices indices)
* 1549 (Doc of evaluation operator of KrigingResult)
* 1553 (Optimization algorithms ignore MaxEvaluationNumber parameter in SORM)
* 1556 (WeibullMin::computePDFGradient yields the partial derivatives in the wrong order)
* 1558 (Example estimate_multivariate_normal: FittingTest::BestModelBIC fails to compute the BIC)
* 1564 (Set a Point makes OT crash)
* 1567 (The API doc of SobolIndicesExperiment has a format issue)
* 1573 (LinearModelAnalysis::drawQQPlot line is not the first bisector)
* 1578 (Option to suppressing and/or save warnings?)
* 1581 (LinearModelAlgorithm run() fails to parse Sample description)
* 1586 (Documentation: description error in the API for the FittingTest_BestModelKolmogorov and FittingTest_BestModelChiSquaredclasses)
* 1590 (The equation of the Fejer quadrature rule is triplicated)
* 1592 (SubsetSampling returns an error if Pf=1)
* 1594 (LinearLeastSquaresCalibration and CalibrationResult)
* 1599 (FieldToPointConnection-BlockSize is missing)
* 1603 (FieldToPointConnection generates an invalid exception)
* 1605 (MaximumLikelihoodFactory cannot be used with FittingTest.Kolmogorov)
* 1624 (The graphs_loglikelihood_contour example has a bug)
* 1642 (Big white space at the beginning of examples)
* 1643 (Problem in MaximumDistribution PDF)
* 1647 (MCMC::computeLogLikelihood does not compute the log-likelihood)
* 1651 (Cobyla freezes in 0T1.16rc1)
* 1658 (TimeSeries accessor)
* 1660 (Cannot extract continuous modes from KLResult when dimension>1)
* 1668 (LevelSetMesher does not take into account the comparison operator)

=== Library ===

==== Major changes ====
* New EV solver for KarhunenLoeveP1Algorithm (Spectra), with sparse matrix and HMatrix support
* Enable HMat AcaRandom compression method

==== New classes ====
* Ipopt optimization solver

==== Documentation ====

==== API changes ====
* Removed deprecated OptimizationAlgorithm::GetLeastSquaresAlgorithmNames
* Removed deprecated GaussianNonLinearCalibration,NonLinearLeastSquaresCalibration::set,getAlgorithm
* Removed deprecated MethodOfMomentsFactory::set,getOptimizationProblem
* ResourceMap::Set* methods no longer add new keys, the new Add* methods must be used instead
* Removed OPTpp
* Renamed HistogramFactory::computeSilvermanBandwidth into HistogramFactory::computeBandwidth.

=== Python module ===
* ProcessSample __getitem__ returns Sample instead of Field
* Implement list indexing

=== Miscellaneous ===
* Add Sample::getMarginal(Description)
* Fixed TBB performance when used together with OpenBLAS

=== Bug fixes ===
* 1124 (DistributionFactory::buildAsXXX methods not documented)
* 1213 (The legend of the graphics in MetaModelValidation is wrong)
* 1222 (There is no kriging example based on HMAT)
* 1331 (FittingTest_BestModelBIC sometimes fail)
* 1335 (The return of the unsafe ResourceMap)
* 1337 (The rDiscrete function has no help page)
* 1349 (Problem in the graph of Histogram)
* 1351 (Text has a zero size)
* 1354 (The doc of GeneralizedParetoFactory does not reflect the implementation)
* 1371 (Memory leak in ot.TruncatedDistribution)
* 1372 (wrap MultiStart::OptimizationResultCollection)
* 1374 (The setKnownParameter of the factories are not documented enough)
* 1376 (The View class has no example)
* 1378 (Kriging-related covariance model weirdness)
* 1383 (The ExprTk engine for SymbolicFunction does not document the "var" keyword)
* 1384 (The ExprTk engine is not case-sensitive)
* 1388 (Kolmogorov fails on a PythonDistribution)
* 1390 (The doc for the `computeQuantile` method does not describe the optional `tail` argument)
* 1393 (SobolSimulationAlgorithm should be simpler)
* 1395 (Indexing Sample improvement)
* 1403 (Python import otagrum throws an error)
* 1404 (Bug in KrigingAlgorithm+hmat-oss)
* 1405 (Most of the simulation algorithms for rare event fail on a coronavirus example)
* 1416 (MethodOfMomentsFactory has no setOptimizationBounds method)
* 1419 (BoundingVolumeHierarchy segaults/hangs)
* 1423 (The computeSilvermanBandwidth of the HistogramFactory has no help)
* 1432 (Expected improvement-based EfficientGlobalOptimization stopping criterion could be improved)
* 1437 (OrderStatisticsMarginalChecker bound message)
* 1438 (Normal distribution: computeComplementaryCDF)
* 1443 (Not all distributions have a getRoughness() method)
* 1448 (Memory consumption leads to crash)
* 1449 (P1LagrangeInterpolation sometimes fails)
* 1455 (GLM::setCovarianceModel could lead to unexpected behavior of parameter optimization in KrigingAlgorithm)
* 1456 (truncation of distribution)
* 1461 (ComparisonOperator().getImplementation().getClassName() segfault)
* 1471 (The graphics of KarhunenLoeveQuadratureAlgorithm has no axes)
* 1485 (The Normal().getRoughness() method is wrong)
* 1495 (Wrong formula for Expected Improvement evaluation)

=== Library ===

==== IMPORTANT: Distributions parametrization changes ====
* New argument ordering in Frechet ctor: scale(beta), shape(alpha), location(gamma) (swapped alpha and beta)
* New parametrization in Gumbel: scale(beta), position(gamma) (beta=1/alpha for first argument)
* New parametrization in Beta: shape(alpha), shape(beta), location(a), location(b) (beta=t-r for second argument)
* New parametrization in InverseGamma: rate(lambda), shape(k) (swapped arguments)
* New parametrization in InverseNormal: location(mu), rate(lambda) (swapped arguments)
* New parametrization in Laplace: mean(mu), rate(lambda) (swapped arguments)
=> One can use "import openturns.shims as ot" to maintain compatibility with older scripts

==== Major changes ====
* New optimization solvers (Dlib, Bonmin for mixed integer optimization problems)
* New distributions (SquaredNormal, WeibullMax, Pareto, DiscreteCompoundDistribution, MixedHistogramUserDefined)
* New estimators (ParetoFactory, GeneralizedExtremeValueFactory, LeastSquaresDistributionFactory, PlackettCopulaFactory)
* New copulas (JoeCopula, MarshallOlkinCopula, PlackettCopula)
* Linear model learner (LinearModelStepwiseAlgorithm)
* System events (IntersectionEvent, UnionEvent, SystemFORM, MultiFORM)

==== New classes ====
* Dlib
* SquaredNormal
* NullHessian
* GumbelLambdaGamma
* LogNormalMuErrorFactor
* WeibullMax
* WeibullMaxFactory
* WeibullMaxMuSigma
* GeneralizedExtremeValueFactory
* Pareto
* LeastSquaresDistributionFactory
* ParetoFactory
* DiscreteCompoundDistribution
* Bonmin
* IntersectionEvent
* UnionEvent
* SystemFORM
* MultiFORM
* JoeCopula
* MarginalEvaluation/Gradient/Hessian
* MarshallOlkinCopula
* LinearModelStepwiseAlgorithm
* MixedHistogramUserDefined
* PlackettCopula
* PlackettCopulaFactory

==== API changes ====
* FittingTest methods to return fitted distributions with factory as argument
* Removed deprecated specific RandomVector constructors
* Removed deprecated HypothesisTest::Smirnov
* Removed deprecated FittingTest::TwoSamplesKolmogorov
* Removed deprecated LinearModel, LinearModelFactory
* Removed deprecated HypothesisTest::(Partial|Full)Regression
* Removed deprecated OptimizationProblem(levelFunction, levelValue) ctor
* Removed deprecated (Linear|Quadratic)(LeastSquares|Taylor)::getResponseSurface
* Removed deprecated VisualTest::DrawEmpiricalCDF,DrawHistogram,DrawClouds
* Moved dot to Point::dot
* Deprecated Weibull in favor of WeibullMin
* Deprecated WeibullMuSigma in favor of WeibullMinMuSigma
* Deprecated WeibullFactory in favor of WeibullMinFactory, buildAsWeibull
* Deprecated GumbelAB
* Deprecated GaussianNonLinearCalibration,NonLinearLeastSquaresCalibration::set,getAlgorithm
* Added getConditionalMarginalCovariance method to KrigingResult
* Added getConditionalMarginalVariance method to KrigingResult
* Deprecated OptimizationAlgorithm::GetLeastSquaresAlgorithmNames
* Added linearity features to Function
* Added 'removeKey' method to ResourceMap
* Removed Copula class
* Deprecated Event class, use ThresholdEvent/ProcessEvent/DomainEvent classes
* Deprecated EnumerateFunction constructors
* Added a minimum probability accessor to SubsetSampling
* Added a UniVariateFunction interface to solvers and integration algorithms

=== Python module ===
* Add Domain.__contains__ operator

=== Miscellaneous ===
* Add GeneralizedPareto location parameter
* Add getSobolGroupedTotalIndex method for FunctionalChaosSobolIndices for the indice of a group of variables

=== Bug fixes ===
* 997 (Adding minimum volume set examples)
* 1004 (The doc for SobolSimulationAlgorithm has issues)
* 1006 (Text drawable does not handle size)
* 1130 (Inconsistency in FittingTest)
* 1160 (2-d GaussianProcess realization graph regression)
* 1169 (Missing key in ResourceMap)
* 1173 (There is no dot product example)
* 1185 (Bug with Normal.computeMinimumLevelSet method)
* 1190 (computeProbability clamped by Domain-SmallVolume)
* 1197 (Doc error: TrendTransform)
* 1198 (Doc error: ValueFunction)
* 1202 (Sample::sort & Sample::sortAccordingToAComponent only return new Samples)
* 1204 (sortAccordingToAComponent does not check its inputs arguments)
* 1209 (LinearModelStepwiseAlgorithm from otlm does not exist in OT)
* 1216 (The CalibrationResult doc does not match the code)
* 1247 (KrigingAlgorithm could not compute amplitude analytically with ProductCovarianceModel )
* 1264 (ProductCovarianceModel ignore active parameters of its 1d marginals)
* 1282 (The dlib example fails)
* 1283 (LogNormalFactory::buildMethodOfLeastSquares has no doc)
* 1289 (The help page of PythonFunction has formatting issues)
* 1303 (Rosenblatt transformation segfault)