Filewatcher File Search File Search
Catalog
Content Search
» » » » » libopenturns-dev_1.0-4_i386.deb » Content »
pkg://libopenturns-dev_1.0-4_i386.deb:648138/usr/share/doc/libopenturns-dev/  info  control  downloads

libopenturns-dev - headers and development libraries for OpenTURNS…  more info»

changelog.gz

2012-04-17  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

		* [1.0] Changed the behaviour of the analytical functions. Now, the ability to perform symbolic differentiation is checked only at the first call to the gradient() or hessian() methods, with an exception if the formula cannot be differentiated analytically. It solves partially ticket #411.
				Removed the deprecated methods in the following classes:
						* DistributionImplementationFactory
						* LinearModelTest
						* VisualTest
						* KernelSmoothing
						* NumericalMathFunction
						* LinearModelFactory
						* Graph
						* GraphImplementation
						* Interval
				Improved the use of the Log class in the following classes:
						* MarginalTransformationEvaluation
						* FittingTest
						* RandomMixture
						* NumericalMathFunctionImplementation
				Added the AggregatedNumericalMathEvaluationImplementation, AnalyticalNumericalMathEvaluationImplementation and DatabaseNumericalMathEvaluationImplementation classes to the TUI. It also contributes to solve ticket #411.
				Fixed ticket #417 by applying Sofiane's patch.
				Fixed ticket #418 by applying Sofiane's patch.
				Fixed #420
				Fixed #414. distcheck ok.
				Added the ability to specify the separator in the BuildFromCSVFile() and exportToCSVFile() methods of the NumericalSample class.
				Applied Sofiane's patch for #432. distcheck ok.
				Fixed header installed twice. distcheck ok.
				Fixed #434
				Fixed python test of ProcessSample related to #434
				Fixed #423
				Fixed #425. distcheck ok.
				Optimized matrix conversion related to #425. distcheck ok.
				Removed useless references to the OT namespace.
				Unified the construction of the result object for the different matrix operators.
				Removed useless calls to the checkSymmetry() and checkHermitian() methods.
				Simplified several methods of the HermitianMatrix class.
				Added the real() and imag() methods to the ComplexMatrix and HermitianMatrix classes.
				Changed the internal storage of the SymmetricMatrix, the SymmetricTensor and the HermitianMatrix classes. Now, only the lower triangle is stored and addressed, excepted for reading values and for pretty-printing.
				Fixed a bug in the matrix multiplication between a square matrix and a symmetric matrix.
				Fixed a bug in the computation of the determinant for symmetric matrices.
				Added the computation of right (ie classical) eigen vectors to the SquareMatrix and SymmetricMatrix classes.
				Fixed a bug in the element accessors for the HermitianMatrix class. Now, the elements are properly conjugated on reading, and only the lower triangle is accessible for writing.
				Fixed a bug in the returned type for the symmetric matrix/symmetric matrix multiplication and the hermitian matrix/hermitian matrix multiplication.
				Fixed a bug in the aggregation mechanism for normal atoms in the RandomMixture class.
				Fixed typecheck typemaps of NumericalSample, NumericalPoint. Added NumericalSample copy constructor to properly copy all attributes (name, description, ...). Closes #445.
				Reverted changes on lib/src/Base/Common/Object.hxx -r2428:2427. distcheck ok.
				Fixed a minor bug in the CleaningStrategy class.
				Fixed several normalization bugs in the SpectralNormalProcess class.
				Changed the definition of the UserDefinedSpectralModel class. Now, it accepts only nonnegative frequencies and the associated frequency grid, but it can be evaluated on negative frequencies thanks to the hermitian symmetry.
				Renamed the evaluation() method into an evaluation operator for the several filtering window implementations.
				Changed the default behaviour of the FilteringWindow class. Now, it is equal to the Hamming window by default.
				Improved the performance of the Hanning class.
				Fixed numerous normalization bugs in the WelchFactory class.
				Fixed a normalization bug in the Hamming class.
				Changed the way the title of the importance factors graph is built in the QuadraticCumul class.
				Added a constructor based on a numerical point with description to the ConstantRandomVector class.
				Added a missing copyOnWrite() in many interface classes.
				Removed whitespace in Pairs::__repr__ to conform to other drawables outputs.
				Fixed a bug in the symmetrize() method in ComplexMatrixImplementation class.
				Fixed roundoff errors in some tests.
				mingw: fix incorrect prefix path when cross compiling.
				Applied Ivan's patch to update wrapper templates (see #421).
				Applied Ivan's patch for #436.
				Fixed warning in Path::CreateTemporaryDirectory where the result of mkdtemp was not checked. Added a test in wrapper templates to check against sample execution (see #436).
				Fixed error in commit 2441.
				Fixed WIN32 compilation related to r2440.
				Fixed rounding error in UserDefinedSpectralModel check test. distcheck ok.
				Enabled missing LHSExperiment python check test. distcheck ok.
				Removed useless Exception test files ; all Exceptions are tested in t_Exception_std.
				Fixed a bug with the blockNumber parameter and a wrong error message in the WelchFactory class.
				Added a missing legend position in the GraphImplementation class.
				Added a clean() method to the ComplexMatrix class.
				Removed the useless symmetrize() method of the ComplexMatrix class.
				Added a clean() method to the Matrix class.
				Added a simplification mechanism to the MarginalTransformationEvaluation class that improves significantly its performance.
				Reworked the WhittleFactory class. More tests are needed.
				Added the getMarginalProcess() method to the CompositeProcess class.
				Added the getMarginal() method to the DynamicalFunction, TemporalFunction and SpatialFunction classes.
				Fixed a memory leak when calling the getMarginal() method of DistributionImplementation children classes without cast into a Pointer or an interface class.
				Improved the const corectness in various places.
				Enabled missing NormalCopulaFactory python check test. distcheck ok.
				Enabled missing FunctionalChaos_ishigami_database python check test.
				Removed duplicate OrthogonalProductPolynomialFactory check test in favor of disabled OrthogonalBasis test. distcheck ok.
				Added missing Whittle parameters in openturns.conf.in.
				Improve parallel python exemple: start the minimum threads needed. Reindent to python style.
				Removed duplicate ComposedNumericalMathFunction test (NumericalMathFunction_composition) and LinearCombinationYYYYImplementation test (NumericalMathFunction_linear_combination).
				Removed useless file t_UserDefined_std_bug.py.
				Removed useless folder lib/src/Chain
				Enabled LinearCombinationYYYYYImplementation python tests.
				Preparing debian package update.
				Updated closed tickets log. distcheck ok.
				CMake: fixed missing examples scripts wrt autotools.
				CMake: fixed missing validation files wrt autotools.
				Fixed gcc 4.7 unqualified template function lookup behavior towards OTtestcode.hxx and other warnings.
				Fixed the bugs in the WhittleFactory class.
				Added a verbose option, an history mechanism and a selection mechanism to the WhittleFactory class.
				Added the WhittleFactoryState class.
				Improved the pretty-printing of the ARMA class.
				Changed the ARMAFactoryImplementation class in order to accept ranges of orders.
				Improved the WelchFactory class in order to accept a block number of 1.
				Updated rpm packages.
				Fixed #464
				Made OTbind shared to mimic autotools.
				Updated debian packages.
				Fixed t_WelchFactory_std rounding.
				Set version number 1.0 distcheck ok.
				Added BoxCoxFactory parameters to the ResourceMap class.
				Added a parameter to check the roots of the ARMA polynomials in the WhittleFactory class.
				Removed the useless inclusion of the csv parser in the TimeSeriesImplementation class.
				Fixed some cast warnings and type errors.
				Fixed some numerical precision troubles in the tests.
				Updated the MuParser version (2.2.2) in order to fix some bugs in the previous embedded version (2.1.0).
				Improved the compression and added some customization in the openturns_script.nsi file. Care: some OpenTURNS paths are hard-coded!
				Improved the compression and added some customization in the openturns_developer_script.nsi file. Care: some OpenTURNS paths are hard-coded!
				Improved the optimization flags: now, OpenTURNS is approximately 30% faster on a core i5 laptop while being i586 compatible. Care: some msis paths are hard-coded!
				Improved the use of the path in the GraphImplementation class.
				Added an OS specific implementation of the system() function to the Os class.
				Reverted most r2472 changes: hardcoded paths, installer customization, aggressive optimization flags, and use CreateProcess only for graphs. distcheck ok.
				Fixed invalid wrapper template header.
				Fixed a bug in the GraphImplementation class under windows when the path is specified.
				Customized the installer in order to give it an OpenTURNS look.
				Reverted the Makefile in distro/windows.
				Fixed missing extern keyword in previous commit 2474
				Removed useless swig option in windows Makefile
				Removed hardcoded paths.
				Improved icon readability
				Cleaned openturns icon and included multiple sizes. Made simple_cxx_example actually simple.
				Reactivated the call to Os::ExecuteCommand() in the HypothesisTest, FittingTest, LinearModelTest, LinearModelFactory and DistributionImplementationFactory classe.
				Added a mechanism to activate/deactivate the call to CreateProcess() in the Os class. This mechanism is compiled only in the Windows version of OpenTURNS, and the default is to use system() instead of CreateProcess. The user has to call the Os.EnableCreateProcess() method in Python in order to switch to the new implementation of the Os::ExecuteCommand() method.
				Improved the pretty-printing of some NumericalMathEvaluation classes, in order to get a reusable expression for functional chaos meta-models in other contexts.
				Improved the knowledge-based otimization in the MarginalTransformationEvaluation class.
				Fixed undefined reference to Os::UseCreateProcess_
				Fixed a conversion warning in the ZipfMandelbrot class.
				Fixed an initialization trouble in the Os class.
				Fixed bug in HyperbolicAnisotropicEnumerateFunction weight constructor. Fixed InverseNormal mean computation. Fixed LogUniform gradients. Fixed a warning in SpectralNormalProcess. distcheck ok.

2012-02-24  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

  * [1.0rc1] MERGE: trunk> svn merge --reintegrate https://.../dutka (2054:2173)
    Removed deprecated wrapper function changeDirectory().
    Removed preloading of libOT.so in Python. Complete refactoring of function resolution in wrappers. Closes #300.
    Fixed bug in wrapper method binding.
    Added new exception catch in Python.
    Fixed bug in Collection.hxx (wrong arguments for delitem method).
    Fixed bug in WrapperCommon.cxx (error in function name).
    Added workaround for bug in libstdc++ 4.4.5 (incorrect exception catched when raised by std::vector<T>::at).
    Fixed bug in WrapperCommons_macros.h (missing macro for WIN32).
    Fixed bug in WrapperInterface.h causing invalid pointer conversion.
    Cleaned various portions of code in order to eliminate most of warnings.
    Added error propagation in TBB version WrapperObject class.
    Made libOTbind static (archive).
    Added 2 mandatory macros for wrappers: WRAPPER_BEGIN and WRAPPER_END.
    Applied M. Souchaud's patch: W2K cd improvement and path fix.
    Added PIC flag to libOTbind since it will be used in DSOs.
    Renamed WrapperCommon_private.h as WrapperCommon_extern.h.
    Factorization of exported wrapper functions (extend usage of BIND_METHODS/BIND_ACTION macros).
    Made libOTbind a dynamic library since static ones encouter several portability problems.
    Added visibility pragma definitions for future usage.
    Removed deprecated class SobolIndicesResult.
    Removed deprecated class SobolIndicesParameters.
    Removed deprecated method CorrelationAnalysis::SobolIndices.
    Removed deprecated test t_CorrelationAnalysis_sobol.
    Removed related SWIG files.
    Applied D.Barbier's patches. Closes #378.
    Fixed bug in lib/src/Base/Stat/CMakeLists.txt file (useless file not removed in previous commit).
    Applied D.Barbier's patches. Closes #380.
    Changed lib and python tests in order to manage both Linux and Windows without changes.
    Changed utils/adapt_check_tests_sh in concordance.
    Fixed bug in libOTbind-0.dll installation.
    Fixed new bug in libOTbind-0.dll installation.
    Fixed bug in libOT-0.dll and libOTbind-0.dll installation.
    Added colorization to make check and make installcheck outputs.
    Added better support for Numpy ndarray in NumericalSample.
    First attempt to parallelize NumericalSample::computeKendallTau() method. :-|
    Applied M. Souchaud's patch for Windows.
    Fixed bug in NumericalSample::computeKendallTau() method.
    Removed useless file utils/adapt_check_tests.sh. Now source files are identical between Linux and Windows.
    Fixed bug due to missing file.
    Applied M. Souchaud's patch for windows distro.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (2055:2178)
    Renamed the isInside() and isNumericallyInside() methods of the Interval class into contains() and NumericallyContains(). The old methods are now obsolete.
    Added the NegativeBinomial class.
    Fixed a minor bug in the NegativeBinomial class.
    Enhanced many tests related to distributions in order to test characteristic functions and probability generating functions.
    Fixed buggy python tests.
    Renamed InverseHalton class into ReverseHalton to be more consistent with the literature.
    Added the HaselgroveSequence class, which implements a new low discrepancy sequence based on irrational translations of the nD canonical torus.
    Added missing entries in ResourceMap and openturns.conf.
    Fixed the bug described in ticket #365 in the LeastSquaresStrategy class thanks to the provided patch. The tests have been updated also.
    Fixed ticket #366: now, the input sample of the projection strategy is stored in the physical space in all circumstances (provided by the user or generated on the fly). It closes ticket #366.
    Fixed a bug in the computeCDF() method of the DistributionImplementation class. It closes the ticket #371.
    Fixed minor bugs in the OTXXX.hxx files.
    Added an history mechanism to all the NumericalMathFunction types. It is deactivated by default, and stores all the input and output values of a function when activated.
    Added a new constructor to the Mixture class in order to specify the collection of distributions and the weightds separately.
    Added new constructors to the FunctionalChaosAlgorithm class in order to use a database instead of a model to perform the computation.
    Changed the constructors based on databases and weights for the ProjectionStrategy and derived classes.
    Added the MeixnerFactory class, in charge of building the orthonormal basis associated to the negative binomial distribution.
    Improved the FunctionalChaosRandomVector: it is now based on the polynomial meta model in the measure space instead of the input distribution based random vector. It provides the same output distribution for much cheaper realizations.
    Improved the performance of the RandomMixture class. Now, all the Normal atoms are merged into a unique atom, which greatly improve the performance in case of random mixture of many such atoms.
    Improved the MeixnerFactory class.
    Removed the test for the MeixnerFactory class in the CMake infrastructure.
    Removed the input/output strategies from all the Simulation classes as it is now a service of the NumericalMathFunction class.
    Removed the systematic computation of importance factors and mean point in event domain from the Simulation and SimulationResult classes, as it is not compatible with the process data model.
    Added the computeQuantileBound() method to the Wilks class, in order to ease the computation of guaranteed bounds on a quantile from Wilk's sampling size.
    Improved the use of the ResourceMap class within all the simulation algorithms.
    Added the save/load mechanism to the LHS class.
    Started the work on the RandomizedLHS class, in links with ticket #376.
    Added the save/load mechanism to the LowDiscrepancySequence class.
    Fixed a bug in the recurrence relations of the MeixnerFactory class
    Updated the validation file of the MeixnerFactory class.
    Fixed a minor bug in the GaussProductExperiment class.
    Improved the implementation of the computeCDF() method of the Multinomial class.
    Improved the verbose mode of the TNC class.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../eads-devel (1986:2181)
    Improved the WhiteNoise class: now, only distributions with zero mean are allowed.
    Improved the ARMA class. Now, it is possible to create non-stationary ARMA but no automatic thermalization is done in this case. The computeNThermalize() method became the computeNThermalization() method and now it takes a user specified epsilon.
    Removed the contructor based on tMin, tMax and stepNumber parameters. Now, the only constructor is tMin, N, timeStep.
    Removed AM_MAINTAINER_MODE macros from configure.ac files. See ticket #384 for explanations.
    Added TimeGrid class.
    Added TimeSeries / TimeSeriesImplementation classes.
    Added missing properties to many files.
    Added missing template specializations in SWIG files.
    Fixed numerous tests.
    Adapted the WhiteNoise, ARMA and TemporalNormalProcess classes to the new TimeGrid class.
    Improved the WhiteNoise class: now, the underlying distribution must have a zero mean.
    Fixed a bug in the ARMA class: when built using a given state, no thermalization must occur.
    Changed the signature of the computeNThermalization() method in the ARMA class. Now, it takes the vlue of the threshold below which the thermalization has occured. Switched the default threshold from about 2e-308 to about 2e-16.
    Renamed the getPossibleFutur() method into getFutur()
    Added a pretty-print method to the ARMAState class.
    Removed the TimeGrid source files, as this class is now defined in the TimeSeriesImplementation source files.
    Added the SpatialFunction class, that allows to generate time series from given time series by considering them as samples, without using the time information.
    Added the DynamicalFunction class, that aloow to generate time series from given time series using the full information in the time series, with a global use of the time grid.
    Fixed a bug in the definition of the ExponentialCauchy class.
    Fixed several bugs in the computeMean() and computeTemporalMean() class due to the new implementation of the TimeSeries class.
    Fixed a bug an inconsistency between the TimeSeries and the TimeGrid classes. In the TimeSeries class, one stores the number of points, whereas it is the number of intervals in the TimeGrid class. This choice is correct, but the difference has to be taken into account in the code.
    Added a drawMarginal() method to the ProcessSample class.
    Added a asNumericalSample() method to the TimeSeriesImplementation class.
    Added missing test for the TemporalNormalProcess class.
    Removed duplicate test.
    Removed duplicate installed programs.
    Fixed some non-critical bugs in the ARMA class.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    Removed useless intermediate namespaces.
    Make distcheck ok.
    CMake ok.
    MERGE: trunk> svn merge --reintegrate https://.../dutka@2246 (2195:2246)
    Reshaped C++ and C files in order to remove namespace indentation.
    Added new C macros: BEGIN_C_DECLS, END_C_DECLS, BEGIN_NAMESPACE_OPENTURNS and END_NAMESPACE_OPENTURNS.
    Added NumericalMathFunction::operator() on TimeSeries.
    Added new wrapper function func_exec_timeseries.
    Added Python side for these enhancements.
    Fixed bug in windows build introduced by previous commit.
    Fixed bug in TimeSeries copy ctor for Python binding. Closes #398.
    Fixed bug in NumericalSampleImplementation with TBB.
    Fixed ticket #400.
    Fixed bug introduced by previous commits.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../schueller (2198:2291)
    CMake: Fixed #387; Added TBB detection; Added uninstall, distcheck targets; Automatically set LIB_SUFFIX.
    Removed FunctionCache in favor of in/out History. Fixed callsNumbers being incorrecly incremented  in ComputedNumericalMathEvaluationImplementation. Added getCacheHits method. Added test cases for PythonNumericalMathEvaluationImplementation, ComputedNumericalMathEvaluationImplementation regarding cache behavior.
    Fixed #377. distcheck ok.
    Fixed bug in optimization algorithms where some non-null initialized variable (on windows) could lead to a significant computation error
    Fixed save/load bug in BoundConstrainedAlgorithmImplementationResult
    Added cmake Numpy detection
    Fixed #369
    Added NumericalMathFunction::addCacheContent to add keys/output values to the cache from input/output samples.
    Added Indices typemaps, closing #381
    Added conversion list / tuple -> Description
    Missing test in previous commit
    Fixed error in latest tests
    Added test for BoundConstrainedAlgorithmImplementationResult save/load
    Added test for array/Description conversion
    Removed useless constructor from range of elements added in previous commits
    Exposed matrix constructors from 1-d scalar collections in python
    Made cmake find scripts compliant with cmake standards
    Some cleaning in main CMakeLists.txt
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../eads-devel (2230:2298)
    Fixed a bug in the getFuture() method of the RandomWalk class.
    Added pretty-print to the DrawableImplementation class.
    Moved the Uncertainty/Algorithm/IsoProbabilisticTransformation folder into Uncertainty/Algorithm/Transformation folder, in order to prepare the development of the process transformations.
    Added the computeQuantilePerComponent() method to the ProcessSample class.
    Fixed several minor bugs in autotest files.
    New classes:
    ===========
    Added the TrendTransform class
    Added the BoxCoxFactory class
    Added the BoxCoxTransform class
    Added the InverseBoxCoxTransform class
    Added the BoxCoxEvaluationImplementation class
    Added the InverseBoxCoxEvaluationImplementation class
    Added the CompositeProcess class
    Improvements / Enhancements:
    ===========================
    Changed method isStationnary to isStationary in Process class
    Changed the frequency grid support. Now we make benefits of the RegularGrid class
    Fixed a bug in the SpectralNormalProcess setTimeGrid() method. We must also update the frequencyGrid and the spectral density matrices
    Fixed minor bug with TimeSeries constructor
    Added the TrendFactory class.
    Added the InverseTrendTransform class.
    Modified the drawMarginal() method of the TimeSeries class in order to improve the readability of the graph.
    Added a flag to tell if the BoxCoxFactory must build a direct or inverse Box Cox transformation.
    Added a getInverse() method to the BoxCoxTransform class.
    Added a getInverse() method to the InverseBoxCoxTransform class.
    Improved the ExponentialCauchy class by using the ExponentialModel and CauchyModel classes.
    Improved the SpectralModelFactory class.
    Improved the CovarianceModel class.
    Implemented the interface/implementation pattern for the SecondOrderModel class.
    Added missing files in the previous commit.
    Added flags to the SecondOrderModelImplementation class to check if the underlying spectral and covariance models are present.
    Added missing files in previous commit. Applying patches is not the best way to interact with the versioning system...
    Made some minor cleaning to the FFT related classes.
    Moved the file COPYING.kissfft to the right place.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (2197:2323)
    Updated the MuParser version. OpenTURNS is now based on MuParser version 2.0.0.
    Removed useless and confusing exception declarations in header files.
    Removed useless and confusing exception declarations in source files.
    Improved the use of ResourceMap throughout the library.
    Added a specialized implementation of the solveLinearSystem() method for the CovarianceMatrix class.
    Improved the test coverage of the solveLinearSystem() method for the various matrix types.
    Improved the speed and accuracy of moments computation for the ZipfMandelbrot distribution.
    Factorized code between the LHS and LHSExperiment classes.
    Worked on the RandomizedLHS class. Ongoing work.
    Added the RandomizedLHS class. It is linked to ticket #376.
    Fixed a bug in the drawProbabilityConvergence() method of the Simulation class.
    Added the RandomizedQuasiMonteCarlo class. It closes ticket #376.
    Improved the getRealization() method of the UserDefined class. Now, it has a O(log(n)) complexity instead of O(n) complexity.
    Added the getMarginal() methods to the UserDefined class.
    Added the MinCopula class.
    Fixed a minor bug in the Pairs class.
    Reindented the source as the last merge was mostly manual and messed up a lot of files with respect to the indentation.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (2327:2375)
    Added a warning in the build() method of the LogNormal distribution in order to inform the user when the MLE estimator failed.
    Improved the drawPDF() and drawCDF() methods of the CopulaImplementation class.
    Simplified the importance factors computation in the AnalyticalResult class.
    Started the implementation of the SimulationSensitivityAnalysis class.
    Improved the buildDefaultLevels() method of the Contour class. Now, the levels are based on quantiles of the value to be sliced.
    Added the normalize() and normalize2() methods to the NumericalPoint class.
    Improved the computeQuantilePerComponent() method of the NumericalSampleImplementation class. Now, it uses the algorithm used by Octave and Matlab. It closes ticket #406.
    Worked on the SimulationSensitivityAnalysis class. Ongoing work.
    Fixed a bug in the Simulation class: the convergence history was not reseted  between runs.
    Fixed a bug in the initialization part of the store() method of the Full, Compact and Last class.
    Improved the SimulationSensitivityAnalysis class.
    Restored the ability to compute importance factors and mean point in event domain to the SimulationResult class, using the SimulationSensitivityAnalysis class.
    Improved the StandardDistributionPolynomialFactory class to take into account the NegativeBinomial special case using Meixner factory.
    Changed the default behaviour of the OrthonormalizationAlgorithm class. Now, it uses the GramSchmidtAlgorithm as a default implementation.
    Improved the DrawableImplementation derivative classes to avoid name conflict with R.
    Checked for null size in various places of the NumericalSampleImplementation class.
    Added methods to define color using the Hue, Saturation, Value color space to the Drawable class.
    Fixed a bug for large arguments in the computeCDF() method of the Multinomial class.
    Renamed an internal variable in DrawableImplementation class to avoid conflict with a Windows macro.
    Make distcheck ok
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../schueller (2294:2382)
    Fixed several memory leaks related to the repeated use of PythonNumericalMathEvaluationImplementation over NumericalSamples
    Fixed cmake FTBFS without blas
    Added size check for AnalyticalNumericalMathEvaluationImplementation
    Fixed bug in Pairs : the clean method of the parent class was not called
    Reverted changeset 2296 as branch eads-devel cleans up
    MERGE: svn merge https://.../trunk -r 2294:2306. Synchronizaton with the trunk.
    Fixed a memory leak in PythonNumericalMathEvaluationImplementation. Make
    Added type conversion for complex matrix
    Added matrix->array conversion
    workaround for a bug in UseSWIG script not supporting multiple modules and parallel compilation
    Added array->Matrix conversion.
    Fixed save/load mistake in Pairs, and Curve
    Added Rpy2 detection. Some process headers were not installed by cmake. Made cmake to properly install openturns-config. Fixed swig warning related to SecondOrderModelImplementation.
    Fixed bug in NS.exportToCSV method.
    Fixed #407.
    Added operators NumericalPoint::__div__, __rmul__; NumericalSample::operator==; Matrix::__rmul__.
    Applied Benoit's patch for OSX for #356.
    Added missing HermitianMatrix conversion
    MERGE: svn merge https://.../trunk -r2294:2379.
    Fixed wrong namespace in RandomizedQuasiMonteCarlo & RandomizedLHS python wrapper files
    Removed useless header inclusion
    Make distcheck ok
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../process (2307:2389)
    Added isComposite method to the Process class
    Added isStationary to the CovarianceModel class
    Changed NormalProcess constructors in order to be more coherent with probabilistic. Now the time grid is the last argument
    Added new constructors to the NormalProcess class. Now we may build a TemporalNormalProcess with argument of type StationaryCovarianceModel and a SpectralNormalProcess with an argument of type SpectralModel
    Updated corresponding tests
    Reworked the TrendFactory class. Now the basis of function is an argument of the constructor.
    Need to build the TrendEval class. Work in progress
    Added ARMAFactory sparse class
    Updated the BoxCoxFactory class. The flag direct was removed as the result should be a BoxCoxTransform
    Changed attribut position in some classes in order to be coherant with the constructors. This might be a potential source of bugs.
    Added the Whittle sparse class
    Started the implementation of the Whittle algorithm
    Added the TrendEvaluationImplementation class.
    Added the InverseTrendEvaluationImplementation class.
    Removed useless sample based operator() method for BoxCoxEvaluationImplementation and InverseBoxCoxEvaluationImplementation classes.
    Slightly rewrite TriangularComplexMatrix::operator() to avoid a GCC warning: control reaches end of non-void function.
    Let order in initialization lists match member declaration.
    Remove unused variables from SpectralNormalProcess::initialize and ARMACoefficients constructor.
    Remove unused variables from t_KissFFT_std.cxx and t_BoxCoxFactory_std.cxx.
    Added missing copy constructors for SpectralFactory class
    Added missing typemap in some swig files (CovarianceModel, SpectralFactory and SpectralModel)
    Made the WhittleFactory class more "user freindly". User may set the algorithm of periodogram estimate
    Removed internal function computeSpectrum in WhittleFactory. The function is implemented in the computeLogLikelihood function in order to optimize 
    some loop calls. Removed also some attributs
    Work is in progress. Need to check if result is a stationary ARMA and to implement tests
    Removed references to old namespaces in RandomWalk.i
    Added missing save / load methods for SpectralModel and SpectralModelFactory classes
    Implemented WhittleFactory estimate method using a ProcessSample argument
    Added missing WhittleFactory tests
    MERGE: svn merge -r2306:2328 https://.../eads-devel
    Fixed a bug in UserDefinedSpectralModel due to bad index
    Fixed bug in WhittleFactory
    Fixed a minor bug in SecondOrderModelImplementation save/load
    Replaced the interpolation of index in the computeSpectralDensity method of UserDefinedSpectralModel by using standard function 
    Renamed lambda value in BoxCoxTransform and InverseBoxCoxTransform because of swig warning
    Removed useless using::operator in TriangularComplexMatrix
    Added the missing swig template HermitianMatrixCollection. This helps users to define their own spectral model
    Fixed wrong comments in the openturns.conf.in file.
    Fixed a scaling bug in the WhittleFactory class.
    Added the missing getFuture() method in the WhiteNoise class.
    Completely reworked the SpectralNormalProcess class in order to take into account the symmetry of the spectral density function. It reduces the number of Cholesky factorization by a factor of two, and the cache mechanism allows to manipulate much larger processes.
    Removed the NormalProcess class: all the processes have an isNormal() method that is enough to trigger the use of specific algorithms for this kind of processes.
    Reworked the TemporalNormalProcess class in order to improve both the time and space complexity of this kind of processes when there is no spatial correlation. In this case, the memory consumption is an O(dimension) instead of O(dimension^2).
    Fixed a bug in the isStationary() method of the RandomWalk class.
    Improved the use of the time grid in the TrendFactory class.
    Fixed wrong comments in the ResourceMap class.
    Added all the hard-coded constants in the process-related classes into the ResourceMap class.
    Fixed a bug in the getMarginal() method of the AnalyticalNumericalMathEvaluationImplementation class.
    Changed the definition of the CauchyModel class. Now, it implements the bilateral spectral density function as it is the natural parameterization of the SpectralNormalProcess class.
    Fixed wrong formulas in the CauchyModel class.
    Improved the memory performance of the CauchyModel class for diagonal spatial correlation. In this case, both the memory and the CPU consumption are O(dimension) instead of O(dimensin^2).
    Changed the definition of the overlap parameter of the WelchFactory class.
    Fixed the build() method based on a TimeSeries in the WelchFactory class.
    Improved the performance of the build() method based on a ProcessSample in the WelchFactory class.
    Reworked the UserDefinedSpectralModel class. Now, it implements a continuous-frequency extension of a dicrete grid-based model snapped to the nearest point in the grid.
    Improved the use of the time grid in the TimeSeriesImplementation class.
    Improved the use of the CovarianceModel and SpectralModel classes in the ExponentialCauchy class.
    Changed the default behaviour of the SpectralModel class. Now, it is based on the CauchyModel class.
    Reworked the SecondOrderModelImplementation class. Now, it uses the CovarianceModel and SpectralModel classes instead of pointers. It avoids most of the pifalls associated with the direct use of pointers, in particular for the save/load mechanisms and the Python integration.
    Improved the parameters accessors of the ExponentialModel class.
    Changed the default behaviour of the CovarianceModel class. Now, it is based on the ExonentialModel class.
    Improved the RegularGird class. Now, it gives a direct access to any value in the grid.
    Added the isDiagonal() method to the SymmetricMatrix class.
    Removed duplicate pretty-print method between the Matrix and MatrixImplementation classes. Fixed a bug in this method for empty matrices.
    More tests are needed for the SpectralNormalProcess, WelchFactory and WhittleFactory classes.
    MERGE: svn merge -r2328:2387 https://.../trunk
    Fixed minor bugs with precision output on windows platform.
    Still a problem with the automatic cast between python structures and OpenTURNS objects.
    Fixed type conversion issues. Fixed rpy2 detection with cmake.
    Make distcheck ok
    Module load ok.
    Windows port ok.
    CMake build ok.
    Updated headers
    Set version number 1.0rc1
    Allowed to create an HermitianMatrixCollection in python. distcheck ok.
    Added UserDefinedSpectralModel test. distcheck ok.

2011-08-02  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

  * [0.15] Added missing HyperbolicAnisotropicEnumerateFunction tests for cmake.
    Fixed #340. Distcheck ok.
    MERGE: hand merge branch barate (1995-2001). Added explicit destructor to class Description. Closes #314. distcheck ok
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (1997:2043)
    Excluded the first coefficient from the cleaning step in the CleaningStrategy. It closes ticket #337.
    Changed the policy in the use of the range information in the computeQuantile() method of the DistributionImplementation class. It closes ticket #338.
    Added missing reference factory accessors in the GramSchmidtAlgorithm class.
    Added missing implementation of the inverse() method of the LinearEnumerateFunction class.
    Fixed a bug in the run() method of the FunctionalChaosAlgorithm class that prevented the vectorial coefficients to be aggregated correctly. It fixes ticket #343.
    Enhanced the MetaModelResult class in order to have access to marginal residuals and relative errors instead of global residual and relative error. It fixes part of ticket #344.
    Fixed several bugs in the tests related to the FunctionalChaosAlgorithm class.
    Improved the tests related to distribution factories.
    Added the missing getParametersCollection() and setParametersCollection() methods to the ZipfMandelbrot class.
    Enhanced the ChiFactory class.
    Added missing implementation of the inverse() method of the LinearEnumerateFunction class.
    Improved the robustness of the Multinomial class.
    Added the build() method from parameters to the NormalCopula class.
    Fixed a minor bug in the getParametersCollection() method in the Mixture class for the nD case (confusing names for the dependence parameters).
    Added the dependence parameters to the getParametersCollection() method of the EllipticalDistribution class.
    Added the missing GaussProductExperiment class to the library (the code was almost complete but not included).
    Improved the robustness of the LowDiscrepancyExperiment class.
    Changed the behaviour of the setDistribution() method of the FixedExperiment class. Now it has no side effect, as the distribution is fixed by the initial sample.
    Adapted the FunctionalChaosAlgorithm, ProjectionStrategy, LeastSquaresStrategy and IntegrationStrategy classes in order to perform functional chaos expansions based on databases. Nevertheless, the user has to fit a reasonable distribution to the input sample in order to get useful results.
    Improved the StandardDistributionPolynomialFactory class in order to recognize and use classical polynomial families when possible.
    Added the ability to draw the points for a given curve to the Curve class.
    Added the ability to use logarithmic scales on one or both axes to the Graph class.
    Fixed numerical precision of several tests.
    Added missing tests in the TUI.
    Switched from exception to warning in case of non-critical non-convergence of optimization algorithms. It closes ticket #350.
    Fixed a minor bug in the CleaningStrategy class, when maximumSize > maximumDimension.
    Make distcheck ok.
    Module load ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../dutka (1998:2047)
    Changed default SWIG detection. Now detection occurs by default.
    Fixed bug in product package detection when autotools reparses the configure files.
    Fix bug in product package detection.
    Forced SWIG module build for distributed tarball so SWIG modules are not preprocessed any more on the packager's system. This should close #287. Be aware that the module template has changed also so update your own module.
    Fixed bug in SWIG macro preventing from using an alternative Python install. Closes #284.
    Added vanished Python function getCorrelationMatrixFromMap()
    Added Repeat static method to Log class in order to see messages as soon as they are received.
    Changed SetFile static method in Log class to automatically switch to non colorized output when writing messages to files.
    Added ColoredOutput static method to TTY class to query the current state.
    Added assignment operator to AtomicInt class to easy value setting.
    Some code reordering to ease debugging in Python module.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> svn merge --reintegrate https://.../schueller (2023:2047)
    Enhanced PYNMF exception handling. Fixed an exception shadowing in OpenTURNSPythonFunction.
    Fixed lintian spelling warnings.
    Fixed #343.
    Bison/Flex detection occurs by default for cmake.
    Removed wrong dependency pyinstallcheck->cppinstallcheck.
    Workaround for swig 2.0.4 regression by removing useless typedef.
    Took FiniteDifferenceSteps into account in FiniteDifference classes.
    BlendedStep replaces ProportionalStep.
    Corrected cmake package version.
    Split NearestPointCheckerResult class.
    Split class SVMRegressionSpecificParameters.
    Fixed calculus of relative error.
    Fixed typo in EnumerateFunctionImplementation.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    mingw: allow several retry if rmdir command failed.
    Updated package sources.

2011-07-07  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

  * [0.15rc1] Changed version number to 0.14.0-based-devel
    MERGE: trunk> svn merge --reintegrate https://.../chaos (1693:1969) 
    Moved Optim/PenalizedLeastSquaresAlgorithm to Algo/PenalizedLeastSquaresAlgorithm. there was a duplicate. had to change ivan's interface a bit
    Added EnumerateFunctionImplementation as OrthogonalProductPolynomialFactory embbeds an instance (cannot instanciate former abstract EnumerateFunction)
    Completely modified BasisSequence coding as NMF doesnt provide op== properly. work in progress on LAR
    Added HyperbolicAnisotropicEnumerateFunction::getStrate*; fixed indent
    MERGE: svn merge -r1693:1746 https://.../trunk
    Created class ApproximationAlgorithmImplementationFactoryImplementation. Modified LeastSquaresStrategy
    Work on CorrectedLeaveOneOut. Work in progress.
    Modified ApproximationAlgorithm interface to take design experiment weights into account. Work on python interface.
    Modified FittingAlgorithm interface to take the BasisSequenceFactory into account
    Replaced LeavePOut with KFold cross-validation.
    Added Sobol' function test
    Added verbosity output. LAR ok. MCV ok.
    LeastSquaresMetaModelSelection to take weights into account
    Removed useless casts to base classes in Python wrapping for EnumerateFunction
    Added default q for HyperbolicAnisotropicEnumerateFunction
    Proposal for adaptative LAR algorithm.
    Fixed HyperbolicAnisotropicEnumerateFunction test.
    Fixed LeastSquaresMetaModelSelection residual.
    Added test for the evaluation of the metamodel produced by PCE
    Sparse representation of metamodel through FunctionalChaosResult.
    Made relative error accessible from FunctionalChaosResult
    Changed definition of stratas for HyperbolicAnisotropicEnumerateFunction.
    Made LAR faster using cholesky 1 rank update.
    Removed BasisSequenceFactory coefficients path validation.
    Removed useless attribute in LAR.
    Allowed use of save/load methods.
    Minor optimization to Corrected LOO.
    MERGE:chaos> svn merge -r1746:1885 https://.../trunk.
    Added few comments. distcheck ok.
    Applied lebrun's patch.
    MERGE:>chaos svn merge https://.../trunk -r1885:1948. Synchronization with trunk. Make distcheck ok.
    Modified sparse chaos tests to use FunctionalChaosAlgorithm directly.
    Applied lebrun's patch to enhance CorrectedLeaveOneOut. distcheck ok.
    Fixed sparse FunctionalChaos test at files.
    Distcheck ok.
    CMake build ok.
    Module build ok.
    MERGE: trunk> svn merge --reintegrate https://.../schueller (1946:1972) 
    Memory efficient computation of Sobol' indices. Dropped dependency to R-sensitivity. Deprecated CorrelationAnalysis's Sobol' method.
    CMakeLists also updated by setVersionNumber.sh. Distcheck ok.
    CMake build ok.
    Module build ok.
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (1945:1980)
    Added pretty print to all the distributions.
    Added access to the Catalog class in the TUI.
    Added pretty print to the marginal transformation related classes.
    Renamed the Instance() method into GetInstance() and Initialization() method into Initialize() in the ResourceMap class.
    Added methods to create colors from red, green, blue or red, green, blue, alpha components in the Drawable class.
    Fixed a minor bug in the Clour class that created a mess when using the 'dot' point style.
    Added the LogUniform class.
    Fixed a bug in the computeCDF() method of the ComposedDistribution class in the 1D case.
    Fixed a bug in the computeCDF() method of the Gumbel class in the case of complementary CDF computation for small arguments.
    Added a missing file in the python TUI.
    Added a missing file in the CMake infrastructure for the TUI.
    Improved the stability of the evaluation operator of the OrthogonalUniVariatePolynomial class (Clenshaw's algorithm is now done in the correct order!).
    Added a missing test for the MultinomialFactory class in the TUI.
    Added the LogUniformFactory class, based on maximum likelihood estimation.
    Added a parameter to change the threshold between CDF and complementary CDF usage in the MarginalTransformationEvaluation class.
    Added pretty-print to the Interval class.
    Fixed a bug in the new __getitem__ method of the NumericalSample class.
    Changed the result type of the getCoefficients() method in the FunctionalChaosResult class.
    Fixed typos in the comments of classes in the OrthogonalBasis directory.
    Added the CharlierFactory class, which provides orthonormal polynomials for the Poisson distribution.
    Added the KrawtchoukFactory class, which provides orthonormal polynomials for the Binomial distribution.
    Added tests on the dimension of the input for the evaluation methods of all the NumericalMathFunction derivates.
    Added a constructor based on input and output samples to the ProjectionStrategyImplementation and LeastSquaresStrategy classes.
    Improved the log messages of the CleaningStrategy class.
    Fixed a minor bug in the computePDF() method of the TruncatedDistribution class.
    Improved the generic implementation of the computeGeneratingFunction method of the DistributionImplementation class.
    Improved the management of the derivated attributes of the ZipfMandelbrot class.
    Improved the run() method of the FunctionalChaos class. Now, only the non-zero coefficients are stored in the result.
    Added missing python tests.
    distcheck ok.
    MERGE: trunk> svn merge --reintegrate https://.../dutka (1945:1990)
    Removed test for bc command. Useless.
    Changed default TBB configuration preventing OpenTURNS from scaling. Closes #322.
    Fixed bug in regular expression shortcut substitution. Closes #324.
    Added a method to toggle colorization in log outputs. Closes #319.
    Fixed bug of not found TBB library. Closes #310.
    Improved OpenTURNS vs Python/Numpy type conversions.
    Fixed bug in NumericalSample output when size==1.
    Updated expout test files in relation with NumericalSample previous bug fix.
    Fixed bug in readFile and WriteFile functions.
    Removed Python static library build.
    Renamed duplicate filenames (for case insensitive filesystems). Closes #307.
    Fixed bug in openturns-config. Closes #302.
    Improved NumericalSample tests for speed.
    Fixed locale issue in CSV file read and write. Closes #227.
    Make distcheck ok.
    Module load ok.
    Windows build ok.
    CMake build ok.
    Fixed KrawtchoukFactory test.
    Added missing wrapper poutre_fullspeed in CMakeLists
    Embedded strndup implementation for windows instead of #def...
    Fix missing MonteCarlo_threads in cmake test.
    Put MonteCarlo_threads cmake test entry in correct position.
    mingw: simple_python_wrapper example checks the computed values.
    mingw: retry deleting temporary dir several time.

2011-06-14  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

  * [0.14] MERGE: trunk> svn merge --reintegrate https://.../schueller (1870:1902)
    Fixed #280.
    CMake : removed python libraries detection when python module is disabled. distcheck ok.
    New distribution module to reduce memory usage 1.3G->800M. distcheck ok.
    CMake : made pyinstallcheck occur after cppinstallcheck
    Fixed i386 build. (Added sync routines in AtomicFunctions thanks to Ivan's ASM code when building for i386 arch).
    Fixed warning in PythonWrappingFunctions. distcheck ok.
    Fixed missing cmake files/tests.
    Moved some swig files to their right module.
    Updated cmake version. distcheck ok.
    Work on AtomicFunctions.
    Fixed warning in Multinomial. distcheck ok.
    Fixed #312.
    Fixed #311.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    Renamed model_dist to model_copula. Make distcheck ok.
    MERGE: trunk> svn merge --reintegrate https://.../lebrun (1869:1907)
    Applied the patch from Nolwenn Balin concerning gcc-4.5 failed build (see ticket #308). It closes the ticket #308.
    Improved the management of exceptional cases in the NonCentralChiSquare distribution.
    Improved the automatic drawing range computation for discrete classes in the DistributionImplementation class.
    Worked on a bug in Ev3 in the simplification process. Still in progress.
    Added the InverseNormalFactory class based on maximum likelihood.
    Added the BurrFactory class, based on maximum likelihood estimation.
    Added an add() method to the Graph class, which is an alias for the addDrawable() method.
    Added the division by a scalar operator to the NumericalPoint and NumericalPointImplementation classes.
    Made some cleaning in the tests, removing duplicate using clauses.
    Added the BurrFactory class, based on maximum likelihood estimator.
    Renamed the addDrawable() and addDrawables() methods of the Graph class into add(). The old methods are still available but are now obsolete.
    Enhanced the computePDF() and computeCDF() methods of the UserDefined class in the 1D case, using a logarithmic algorithm instead of a linear one.
    Fixed a bug in the save/load mechanism of the FunctionalChaosResult class.
    Added the save/load mechanism to all the classes in the OrthogonalBasis folder.
    Added an operator for the division by a scalar to the NumericalPoint class.
    Removed the useless NumericalPointImplementation and NumericalPointImplementationWithDescription classes.
    Added the DualLinearCombinationEvaluationImplementation, DualLinearCombinationGradientImplementation and DualLinearCombinationHessianImplementation classes for linear combinations of NumericalPoint with 1D output dimension NumericalMathFunction coefficients.
    Added pretty-print and save/load mechanism to several classes in Base/Func.
    Added missing tests for the library and the TUI.
    Added a constructor from implementation to the Graph class.
    Imposed a 17 digit precision for writing in the XMLStorageManager class.
    Added missing keys in openturns.conf and the ResourceMap class.
    Added a missing file in python/src/Makefile.am
    Fixed a bug in the computeConditionalQuantile() method of the DistributionImplementation class.
    Added a generic implementation of the getRealization() class in the DistributionImplementation class.
    Deprecated the buildImplementation() methods of all the DistributionImplementationFactory classes. One should use the build() method instead.
    Deprecated the buildLM() method of the LinearModelFactory class. One should use the build() method instead.
    Deprecated the LMAdjustedRSquared(), LMAdjustedRSquared(), LMFisher(), LMResidualMean() and LMRSquared() methods in the LinearModelTest class. One should use the LinearModelAdjustedRSquared(), LinearModelAdjustedRSquared(), LinearModelFisher(), LinearModelResidualMean() and LinearModelRSquared() methods instead.
    Fixed a bug in the DrawLinearModel() method of the VisualTest class.
    Deprecated the DrawLMVisualTest() method of the VisualTest class. Use DrawLinearModel() instead.
    Deprecated the DrawLMResidual() method of the VisualTest class. Use DrawLinearModelResidual() instead.
    Added the DrawKendallPlot() method to the VisualTest class.
    Added missing tests for the TUI.
    Added new elements accessors to the NumericalSample class in the TUI.
    Added the HyperbolicAnisotropicEnumerateFunction class (imported from the chaos branche).
    Changed the default color of the diagonal from grey to red in the VisualTest class.
    Added missing copyOnWrite() calls in the Graph class.
    Improved const correctness of the NumericalSample class.
    Added missing copyOnWrite() calls in the NumericalSample class.
    Added double indices accessors at() and operator() () to the NumericalSample class.
    Added a linear index accessor to the NumericalSample class.
    Removed the automatic drawing of a grid for graphs that includes a Pie instance.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    MERGE: trunk> hand merge https://.../dutka (1854:1919) 
    Improved wrapper file management. Added multiple DTD management.
    Improved substitution in wrapper external files.
    Fixed reopened bug described in ticket #14.
    Fixed build failure with GCC 4.6. Closes ticket #305
    Improved wrapper library:
    * Added variable name checks in <subst> tag.
    * Subtitution now only occurs for listed variables.
    Added colorization for Log messages.
    Improved wrapper library:
    * Fixed bug concerning back reference in replace strings (\nnn) when they appeared just before some digit (the digit was interpreted as being part of the back reference).
    Changed generic wrapper.
    Changed log output for wrapper messages.
    Added internal variable substitution (OPENTURNS_EXECUTION_DIR).
    Changed default wrapper implementation.
    Added missing files in CMakeLists.txt. Make distcheck ok.
    Python NMF can now be saved in studies. Fixed #316.
    Prepared refundation of wrapper library.
    Fixed bug in ResourceMap preventing it from being printed. Closes #317.
    Fixed warning in SWIG file for NumericalSample.
    Make distcheck ok.
    Module load ok.
    Windows port ok.
    CMake build ok.
    Added operator== for Interval & Indices
    Enhanced ot python function cache and calls number count.
    Added some docstring.
    MERGE:>trunk hand merge branch-souchaud (1874:1875) Mingw: Improve README file. Disable distcc by default.
    Raise a warning rather than an error if unable to remove directory.
    Mingw: remove shell colors.
    Fix bug in regex shortcuts.
    Fixed use of deprecated methods in wrapper templates.
    Added docstring for env variables.
    Added missing wrappers for cmake.
    Updated copyright date.
    Reverted changeset 1927.
    Mingw: simplify compilation configuration environment.
    Mingw: add linux parameter example to python wrapper.
    Fix bug with tabulation in regex shortcuts.
    Fix bug : remove multiple copy of exchangedata in defaultWrapper (ticket #326).
    Added missing environment var docstring.

2010-04-15  Julien Schueller <schueller(:AT:)phimeca(:DOT:)com>

  * [0.14rc1] Added CMake support. Work in progress.
    Make distcheck ok.
    CMake support. Work in progress.
    Make distcheck ok.
    CMake support. Work in progress.
    Make distcheck ok.
    CMake support. Work in progress.
    CMake support. Work in progress.
    CMake support. Work in progress. Now make, make check, make install and make installcheck work.
    Make distcheck ok.
    CMake support. Work in progress. Now compiles python modules.
    Note: in order to support simultaneously Autotools and CMake build systems, the python modules and files have been renamed.
    CMake support. Work in progress. Now install python modules. Python importation of module works.
    Make distcheck ok.
    CMake support. Work in progress. Python tests are ok.
    Make distcheck ok.
    CMake support. Work in progress. Added missing files to installation.
    Make distcheck ok.
    CMake support. Work in progress. Packaging for source tree ok.
    Make distcheck ok.
    CMake support. Work in progress. Add R detection.
    Make distcheck ok.
    MERGE:trunk> svn merge --reintegrate https://.../lebrun (1455-1580)
    Created the AggregatedNumericalMathEvaluationImplementation class, that allows to stack several functions with the same input space and output spaces of different dimensions.
    Created the ProductPolynomialHessianImplementation class.
    Created the ProductPolynomialGradientImplementation class.
    Added a getMarginal() method to the NumericalMathEvaluationImplementation, NumericalMathGradientImplementation and NumericalMathHessianImplementation classes.
    Added the derivate() and derivative() methods to the UniVariatePolynomial class, that allow to compute the derivate UniVariatePolynomial and the value of this derivate at a given point without having to form the derivate polynomial.
    Added the LnBeta() method to the SpecFunc class.
    Added an accessor to set simultaneously the log-parameters of the LogNormal class and improved the robustness of the parameters accessor, in order to prevent parameters that would exceed the double precision capacities. It closed ticket #255.
    Added the computeCharacteristicFunction() method to the Beta class.
    Added a generic implementation of the computeCharacteristicFunction() method to the DistributionImplementation class. Warning: the current algorithm has a linear cost with respect to the magnitude of the argument.
    Added a line width modifier for the Staircase and the BarPlot classes.
    Added the save/load mechanism to the FunctionalChaosResult class and all the IsoProbabilisticTransformation related classes.
    Added a new stopping criterion based on the standard deviation for the Simulation class. It closes ticket #260.
    Added a constructor to the NumericalMathFunction class in order to build aggregated functions.
    Added a default constructor for the save/load mechanism for the Event and EventRandomVectorImplementation classes.
    Added missing accessors to the QuadraticCumul class.
    Added the NumericalMathGradientImplementation and NumericalMathHessianImplementation classes. The symbolic differenciation is done using a reworked version of the Ev3 library. See http://www.lix.polytechnique.fr/~liberti/.
    Removed redundant methods in the KernelMixture class with respect to its base classes.
    Removed the useless destructor in the NumericalPoint class.
    Changed the signature of the getMarginal() methods for the NumericalMathEvaluationImplementation, NumericalMathGradientImplementation and NumericalMathHessianImplementation classes.
    Changed the setPairCollection() method in the UserDefined class. The probabilities are now automatically normalized.
    Changed the implementation of the getMarginal() methods of the NumericalMathFunctionImplementation class.
    Improved the robustness of the setPairCollection() method of the Histogram class.
    Improved the performance of the computePDF() method of the KernelMixture class.
    Improved the drawPDF() and drawCDF() methods of the DiscreteDistribution class.
    Improved the drawPDF() method of the Histogram class.
    Improved the build() method of the OrthogonalProductPolynomialFactory class. Now, the polynomial has an exact evaluation of its gradient and hessian instead of a finite difference approximation.
    Improved the getMarginal() methods of the AnalyticalNumericalMathEvaluationImplementation class.
    Improved debug messages of the DistributionImplementation class.
    Improved the accuracy of the computeQuantile() method of the RandomMixture class.
    Improved the boud checking in the computeQuantile method of the DistributionImplementation class.
    Improved the windows portability of the Ev3 library.
    Improved the numerical robustness of the computePDF() method in the Beta class. It closes ticket #265.
    Fixed a bug and added sample accessors to the ProjectionStrategyImplementation and ProjectionStrategy classes. It closes ticket #263.
    Fixed a bug in the computeCharacteristicFunction() method in the Exponential class. It was a problem of sign that showed up only when gamma <> 0.
    Fixed a bug in the generic implementation of the getRealization() method in the DistributionImplementation class: the same algorithm was used for 1D and multivariate distributions, which is a nonsense.
    Fixed a bug that prevented full accuracy computations in the RandomMixture class when the underlying distribution was smoother than a Normal distribution.
    Fixed a spurious bug in the computeCDF() method of the RandomMixture class.
    Fixed several typos that prevented OpenTURNS to be built using gcc 4.5.0.
    Make distcheck ok.
    MERGE: trunk> svn merge --reintegrate https://.../branches/dutka (1457:1616)
    Library:
    ========
    * Improvements:
      - Read static class members from openturns.conf and ResourceMap object.
      - Changed NumericalSampleImplementation to make it much more compact.
      - Removed some useless casts in many parts of the C++ code.
      - Changed __str__() signature.
      - Added pretty print to WrapperFile, WrapperData, Collection, NumericalMathFunction, NumericalMathFunctionImplementation, NumericalMathEvaluationImplementation, NumericalMathGradientImplementation, NumaricalMathHessianImplementation, CenteredFiniteDifferenceGradient, CenteredFinitedifferenceHessian, NonCenteredFiniteDifferenceHessian, ComputedNumericalMathEvaluationImplementation, ComputedNumericalMathGradientImplementation, ComputedNumericalMathHessianImplementation, WrapperObject, Library classes.
    * Parallelization:
      - Added Threading Building Blocks (aka TBB) support for multithreading.
      - Changed implementation of some counters with atomic variables
      - Parallelization of NumericalSample::computeMean(), NumericalSample::scale(), NumericalSample::computeVariancePerComponent(), NumericalSample::getMin(), NumericalSample::getMax(), NumericalSample::translate(), NumericalSample::computeEmpiricalCDF(), NumericalSample::computeKendallTau().
      - Added thread cancellation mechanism when performing multithreaded computations.
      - Added lock to WrapperExchangedData structure.
      - Improved Log thread safety.
      - Created one temporary directory by thread when executing code.
      - Multithreaded sample execution function embedded in library. No more need to define it in wrapper code.
      - Speeded computations by removing useless output messages when not wanted.
    * Misc:
      - Add deprecated tag to some obsolete methods.
      - Changed svn:mergeinfo property on some files in order to ease incoming merge.
      - Changed default log level in order to see wrapper error messages
      - Removed useless rm command detection. Replaced it with internal recursive remove function.
      - Fixed some typos.
    * Fixes:
      - Fixed bug in output of tests.
      - Changed linktest compilation flags.
    Python:
    =======
    * Improvements:
      - Better management for NumericalPoint and NumericalSample objects when passed as arguments. Automatic cast for python sequence objects (tuple and list) in NumericalPoint and NumericalSample objects.
      - Improvement for NumericalSample class. Now constructible from a collection of NumericalPoints. Acts as a python sequence object.
      - Python module 'openturns' handles Ctrl-C.
      - Updated the test for the Python based NumericalMathFunction.
      - Improved Distribution argument management. No more need to cast DistributionImplementation objects into Distribution in method arguments.
    * Fixes:
      - Added selective sample evaluation for wrapper based NumericalMathFunction. Closes bug #236.
      - Added sample execution method for Python based NumericalMathFunction. Closes #237.
      - Fixed bug #240 concerning DistributionCollection copy constructor.
    Wrappers:
    =========
    * Improvements:
      - Add user prefix in temporary directory to wrapper description file.
      - Added <user-prefix> tag to wrapper description file in order to create a generic wrapper for command line callable codes.
      - Created a new generic wr
Results 1 - 1 of 1
Help - FTP Sites List - Software Dir.
Search over 15 billion files
© 1997-2017 FileWatcher.com