- CHANGED:
- Update packages and installation method to full pyproject.
- FIXED:
- Correct scaling of Trunc-normal distribution.
- FIXED:
- Use weakref to reduce bloat.
- FIXED:
- conditional variables x in E[P | x] now enforced to be ordered correctly.
- CHANGED:
- Switch scipy backend of TruncNormal.
- CHANGED:
- Replaced poetry with native pip.
- Testing with CircleCI replaced with Github action.
- Remove assert check.
- REMOVED:
- Python2 testing.
- FIXED:
- Gamma pdf and moment breaks down for large values of alpha. Switching to new scipy formula.
- FIXED:
- Compatibility with numpoly: Bug workaround in polynomial division. Only affects numpoly v1.2.3 and generating polynomials with normed=True.
- FIXED:
- Compatibility with numpoly: Update string key access to ndpoly objects.
Refactoring orthogonal -> expansion module.
- ADDED:
- Dedicated classical orthogonal expansion schemes: chaospy.expansion.{chebyshev_1,chebyshev_2,gegenbauer,hermite,jacobi,laguerre,legendre}
- CHANGED:
- Function renames: chaospy.{orth_ttr,orth_chol,orth_gs,lagrange_polynomial} -> chaospy.expansion.{stieltjes,cholesky,gram_schmidt,lagrange}
- Docs update.
Refactoring quadrature module.
- ADDED:
- chaospy.quadrature.fejer_1 is added.
- Dedicated classical quadrature schemes: chaospy.quadrature.{chebyshev,gegenbauer,hermite,jacobi,laguerre,legendre}
- CHANGED:
- Bound checks for the triangle distribution. (Thanks to @yoelcortes.)
- Refactored hypercube quadrature to common backend. This gives lots of flags like seqments and
- Function renames: chaospy.quad_{clenshaw_curtis,discrete,fejer,gaussian,grid,gauss_lengendre,gauss_kronrod,gauss_lobatto,gauss_patterson,gauss_radau} -> chaospy.quadrature.{clenshaw_curtis,fejer_2,gaussian,grid,legendre_proxy,kronrod,lobatto,patterson,radau}
- Patterson growth rule changed from 0, 3, 7, ... to 0, 1, 2, ... but maps backwards. Defaults to not have growth parameter, as setting it false makes no sense.
- Renamed: chaospy.generate_sparse_grid -> chaospy.quadrature.sparse_grid
- REMOVED:
- Genz-Keister quadrature quad_genz_keister is deprecated as it does not fit the chaospy scheme very well.
- ADDED:
- Support for approximate multivariate raw moments.
- FIXED:
- TruncNormal is a ShiftScaleDistribution, not a J operator.
- FIXED:
- Bugfix: sparse=True growth=None should se growth=True, but didn't this means that rules that require growth rules to be nested, i.e. clenshaw_curtis, fejer, discrete, newton_cotes were not benefiting fully from sparse-grid.
- FIXED:
- Correct bounds for Mean-Covariance distributions.
- FIXED:
- Correct triangle repr.
- ADDED:
- Increase the number of dimensions supported in Sobol sequence to 1111.
- Lower level override get_*_parameters functions to differentiate between density/fwd/inv, mom, ttr, and lower/upper.
- New allow_approx flag in Distribution.pdf.
- More docs and tests.
- Support for bibliography in docs.
- CHANGED:
- Updated Clenshaw-Curtis and Fejér algorithm which scales much better.
- More aggressive sample use in approximate_moment as bottleneck was the quadrature (Clenshaw-Curtis and Fejér).
- Better support for density approximation. Allow for more contexts by weaving a full density history.
- Documentation update.
- FIXED:
- Wrappers distribution no longer ignores wrapped distribution during dependency declaration. Ignoring them have in some cases caused some variables not to be declared correctly.
- FIXED:
- Bugfix in rounding for discrete distributions.
- Bugfix in rule for when to round discrete variables.
- ADDED:
- include_axis_dim flag added to Distribution.sample to force the inclusion of extra dimension. (Currently first dimension is omitted is len(dist) == 1.)
- Code of conduct and contribution descriptions in repo root.
- Tutorial for doing sequential polynomial chaos kriging.
- CHANGED:
- chaospy.E_cond changed to accept simple polynomials as second argument, allowing for e.g. chaospy.E_cond(q0*q1, q0, dist) which can be interpreted as "expectation of q0*q1 given q0 with respect to dist".
- Full refactorization of the documentation.
- Updates numpoly to version 1.1.0. (some small breaking changes).
- FIXED:
- Bugfixes to chaospy.Spearman
- REMOVED:
- Deprecated report_on_exception. Caused recursion problems, and only a semi-useful diagnostic tool to begin with.
- No more support for Python 3.5. This allows the poetry install to use newer version of numpy and scipy. (This relates to poetry install, so working in py35 might still be possible in practice.)
- ADDED:
- include_axis_dim flag added to Distribution.sample to force the inclusion of extra dimension. (Currently first dimension is omitted is len(dist) == 1.)
- CHANGED:
- chaospy.E_cond changed to accept simple polynomials as second argument, allowing for e.g. chaospy.E_cond(q0*q1, q0, dist) which can be interpreted as "expectation of q0*q1 given q0 with respect to dist".
- Bugfixes to chaospy.Spearman
- Updates to the documentation.
- REMOVED:
- Deprecated report_on_exception. Caused recursion problems, and only a semi-useful diagnostic tool to begin with.
- No more support for Python 3.5. This allows the poetry install to use newer version of numpy and scipy.
Refactored chaospy.quadrature.recurrence -> chaospy.recurrence.
- CHANGED:
- chaospy.constructor removed in favor for chaospy.UserDistribution.
- Bugfix: chaospy.InverseGamma moments needed to be reciprocal.
- Increased range on distributions: StudentT.
- Moved submodule chaospy{.orthogonal->}.recurrence.
- Stieltjes method get common interface chaospy.stieltjes which uses analytical TTR if present, and approximation if not.
- Refactor discretized_stieltjes to be an iterative method with tolerance criteria instead of brute forced. Also added max iterations and scaling.
- Flag: Default recurrence_algorithm default changed to stieltjes (as it covers both analtical and discretized Stieltjes).
- Discretization default in Lanczos and Stieltjes changed from fejer to clenshaw_curtis as edge evaluation is better handled these days, and the latter is better for when edges are finite.
- REMOVED:
- chaospy.basis and chaospy.prange (which was superseded by chaospy.monomial in June).
- Removal of "analytical" TTR where it is approximated: Triangle.
- chaospy.chol modules and the Cholesky functions: bastos_ohagen, gill_murry_wright and schnabel_eskow. gill_king moved to chaospy.orthogonal.cholesky as it is used by orth_chol.
- Flag: accuracy deprecated in favor for tolerance.
- CHANGED:
- lower > upper illegal for all LowerUpperDistribution and Trunc.
- scale <= 0 illegal for all ShiftScaleDistribution.
- Add epsilon buffer to all quadrature rules that evaluate at the edges.
- numpoly update to version 1.0.8.
Release!
- ADDED:
- Gaussian Mixture Model: GaussianMixture.
- Tutorial for how to use scikit-learn mixture models to fit a model, and chaospy to generate quasi-random samples and orthogonal polynomials.
- CHANGED:
- chaospy.Trunc updated to take both lower and upper at the same time.
- REMOVED:
- chaospy.SampleDist removed in favor of chaospy.GaussianKDE.
Additive recursion sampler.
- ADDED:
- Support for additive recursive sampling scheme.
- Tutorial for Monte-Carlo now includes compare of difference sampling schemes.
- CHANGED:
- Bugfix to antithetic variate.
Mv-KDE support!
- ADDED:
- Added support for multivariate kernel density estimation distribution GaussianKDE.
- Tutorial for KDE.
- CHANGED:
- Default tolerance for the accuracy in approximate inverse lowered from 10^-5 to 10^-12.
- Lots of distribution have fixes such that dist.inv([0, 1]) is now allowed in general.
- Update to lots of docs to include example with dist.inv([0, 1]).
- Update nbval config to be more relaxed during tests.
Distribution operations are now all one-dimensional. One pass per dimension.
- ADDED:
- New report_on_error decorator to get more understandable error output.
- New helper functions: format_repr_kwargs, init_dependencies, declare_dependencies, check_dependencies.
- New intermediate distribution baseclasses: ItemDistribution, LowerUpperDistribution, MeanCovarianceDistribution, OperatorDistribution, ShiftScaleDistribution.
- New basic distribution: InverseGamma.
- New error type of error UnsupportedFeatureError to differentiate illegal operations (covered by StochasticallyDependentError) and unsupported features.
- Lots of new tests.
- CHANGED:
Lots and lots of positional idx arguments everywhere to indicate the dimensions worked on. Except for _mom which is kept as is.
Adding consistent baseclass naming convention: Copula{->Distribution}, Mul->Multiply, Neg->Negative, DistributionCore->SimpleDistribution.
Qoi_Dist will no longer returns a numpy array in the multivariate case. This is because Distribution no play will as a numpy object type.
Changes to cache system: * Cache content changed from Dict[Distribution, ndarray] to
Dict[Distribution, Tuple[ndarray, ndarray]] to store both inputs and outputs for each calculations.
- backend function _value replaced with _cache for consistency.
- Backend interface _get_value replaced with _get_cache_1 and _get_cache_2. For former is new, the latter is a renaming.
Iid is changed from being a function wrapper to a subclass wrapper, allowing once again isinstance(dist, Iid).
- REMOVED:
- Deprecating topological soring in J, as this is now handled by the evaluation order.
- Old function interfaces add, mul, neg, trunk, trunc, pow.
- Comparison operators <, <=, > and => for distributions. These were used as syntactic sugar referencing chaospy.Trunc. This to support == operator.
Adding rotation: changing dist backend.
ADDED: CHANGED:
- Baseclass distribution baseclass refactoring:
- Split old Dist into two: Abstract baseclass Distribution and convenience structure DistributionCore.
- Cleaned up __init__ structured to be more standardized.
- Much improved REPR handle.
- standardized __len__.
- Lots more pre-flight checks for distribution integrity.
- Simplification and standardization of distributions.operators.
- Better recursive caching of values during evaluations.
- Some hierarchy changes.
- Tiny changes in argument signature for some distribution. Same arguments, but some change in names or order to standardize. These changes affect: Angelit, Burr, Cauchy, ChiSquared, F, FoldedNormal, GeneralizedExtreme, HyperbolicSecant, Levy, LogWeibull, Logistic, MvStudentT, Pareto1, Pareto2, PowerLogNormal, PowerNormal, StudentT,
- REMOVED:
- chaospy.distributions.evaluation is removed in favor for method on the Distribution class.
- DependencyError deprecated in favor of StochasticallyDependentError.
- matmul operator is in practice an really odd duckling that is highly incompatible with the rotation idea. If linear map is needed, use MeanCovariance.
Adding rotation: the fundamentals.
- ADDED:
- Property for checking for dependencies: Dist.stochastic_dependent.
- Lots of illegal probability distribution configuration that would cause trouble during execution are now caught earlier with an appropriate error.
- Logging of samples out-of-bound for Dist methods: pdf, cdf, inv, fwd.
- Dist.pdf get the extra flag decompose to split density into parts (like inv and fwd does by default.) Should work with all distribution, with a few exception. (MvLogNormal comes to mind.)
- New LocScale baseclass for all generic distributions with location and and covariance structure.
- Lots of new tests.
- CHANGED:
- New and improved dependency system based on underlying variable declaration.
- Some probability distribution boundaries moved from hardcoded to automatically detected.
- Update Iid to not be J subclass.
- Test cases for the new LocScale baseclass: MvNormal and Alpha.
- REMOVED:
- Precedence order system. Was not ready yet, and a new one is being made with the rotation system.
- FIXED:
- Correct length for multivariate binary operators.
- CHANGED:
- numpoly update to version 1.0.6.
- ADDED:
- Support for quadrature dispatching for Mul, Add and (independent) J.
- CHANGED:
Refactor approximate_moment: * Remove unused antithetic variate. (Really never used.) * Remove redundant support for multiple exponents at once. In practice only
one is called at the time anyway.
- Adding buffering for both quadrature and results, so to reduce needed computation for recursive methods.
- New order default: 1000 -> 1000/log2(len(dist)+1) About the same for lower dimensions, but scales better with higher dimensions.
Update requirements to include numpy.
- CHANGED:
Refactor discrete distribution: * Allowing "offset" (up to 0.5 on each side), making all discrete
distributions piece-wise constants.
- Use linear interpolation in dist.fwd and dist.inv between the edges, making them piece linear function.
- dist.cdf adjusted 0.5 to the right to replicate old behavior.
- Update the two implemented discrete distributions DiscreteUniform and
Binomial.
- ADDED:
- Added changelog (the file you currently are reading).
- Support for polynomial saving to and loading from disk.
- CHANGED:
- Refactor descrete distributions to work better with quadrature.
- numpoly update to version 1.0.5.
- FIXED:
- Bugfix: Poly-division with large relative error caused infinity-loops.
- CHANGED:
- Move chaospy/tutorial -> chaospy/docs/tutorials.
- Use nbsphinx to integrate notebooks straight into the RTD docs.
- Renamed chaospy/{doc -> docs}.
- Include numpoly documentation content directly in toctree.
- numpoly update version 1.0.3
- Chaospy logger now capture Numpoly as well.
- Aligning Numpoly properly, making a wrapper redundant.
- REMOVED:
- Announcing deprecation of chaospy.basis and chaospy.prange in favor of chaospy.monomial.
- Deprecating chaospy.setdim in favor for numpoly.set_dimensions.
- ADDED:
- Add Joe-copula back into the fold.
- Add chaospy.example to simplify the Jupyter notebook tutorials creation.
- CHANGED:
Remove CircleCI build-cache system in favor of simpler linear builds. * Reduce checks to 2.7 and 3.8. Anything between is assumed to be covered
by the two.
Clean up sensitivity analysis tools.
Clean up copula docs.
Move lots of doc examples from .rst to .ipynb.
- REMOVED:
- Remove Sens_*_nataf as they were a one-shot project for a paper and no longer work.
- Deprecate old Archemedean base copula
- CHANGED:
- Switch numpoly.bindex with new numpoly.glexindex.
- REMOVED:
- Removing unused Bertran functions.
- Deprecating old Distribution names (which have been announce for over a year through warning messages)
- ADDED:
- Added chaospy.orthogonal.frontend:generate_expansion as an one stop expansion generation function.
- Add tag-check when deploying using tags.
- Add logging which activates on env CHAOSPY_DEBUG=1. Log to file with env CHAOSPY_LOGFILE=/path/to/file
- Added Program Evaluation and Review Technique (PERT) distribution.
- Adding support for Dist.__matmul__ (which obviously does nothing in python 2).
- Adding tests to the hard-to-get-right sub-module: chaospy.distributions.operators.
- Added LRU cache to some quadrature schemes.
- Added segments to Newton-Cotes, Fejer and Clenshaw-Curtis (as this is recommended to have to discretized Stieltjes).
- Added experimental Jupyter notebooks with user tutorials/recipes GITROOT/tutorial
- Gumbel and Clayton copulas get analytical recursive Rosenblatt transformations.
- CHANGED:
Update numpoly to version 0.3.0. * Replace explicit numpoly import, with an implicit one with a
smart-wrapper.
- Docs updated with new polynomial string representation order.
Update to documentation.
Replace sample and quadrature scheme name from one letter ["G", "E", "C", "H", ...], to new full name strings: ["gaussian", "legendre", "clenshaw-curtis", "halton", ...]. (Old style still works, but is undocumented.)
Increase quadrature sample rate 100->200 when doing discretized Stieltjes to increase accuracy (at the computational cost).
Increased sample rate for approximate inverse (used when inverse is missing), increasing accuracy at extra computational cost.
New style Archemedean copula.
Refactor chaospy.distributions.operators to become less messy.
Some adjustment to the expansion functions to align with the new frontend.
Update lagrange to use numpoly.bindex in the backend.
Use graded: bool and reverse: bool as a replacement for sort: str = "GRI": * The "I" in "GRI" is deprecated: It can always be achieved with
values = values[::-1], so it serves little purpose.
- The "R" was implemented backwards. R present is equivalent with reverse=False.
- sort still works, but raises an warning about future deprecation.
- Using one letter strings is less readable, and needs to be removed. Splitting them up, simplifies documentation.
- REMOVED:
- Deprecating copulas Frank, Joe and Ali-Mikhail-Haw, as their accuracy is not good enough.
- Remove really old tutorial stuff not longer in use.
- FIXED:
- Bugfix for evaluate_lower and evaluate_upper for operators like addition, multiply, power, etc.
- Fix to interpret_as_integer of joint distribution (now covering mixed content).
- ADDED:
- Added chaospy.__version__
- CHANGED:
Upper and lower methods: * Replace Dist.bnd with Dist.lower and Dist.upper to have better
control.
- Issue future deprecation warning if Dist._bnd is used.
- Deprecate chaospy.distributions.approximation:find_interior_point as its use falls away with the new methods.
- Add new chaospy.distributions.evauation.bound:evaluate_lower and evaluate_upper
Fix to interpret_as_integer of joint distribution with discrete components.
- REMOVED:
- Deprecated trigonometric distribution transformations, as the were hard to transfer over, undocumented and likely not used.
- CHANGED:
- numpoly version 0.1.6.
- CHANGED:
numpoly introduced, version 0.1.4: * Replacing backend for polynomial handle with numpoly, leaving just a
compatibility wrapper.
- Refactor descriptive to utilize new backend
- Update all docstring containing a polynomial as the string representation has changed.
- Declare chaospy.Poly as soon-to-be deprecated
Replace setuptools+pipenv for installation and development management to poetry for both
Introduce CircleCI build-cache step.
Distribution update: * Added Dist.interpret_as_integer to better support discrete
distributions.
- Update lots of method docs in chaospy.distributions.collection to look better.
- CHANGED:
- Making a logger.warning into logger.info (as requested by user).
- ADDED:
- Added support for openturns Distributions (thanks Régis Lebrun).
- Added "Related Projects" section to root README with thanks and shout-outs.
- Added discrete distributions: Binomial, DiscreteUniform.
- Added recipe for stochastic dependent distributions: doc/recipes/dependent.rst
- CHANGED:
- Moved external interfaces to new submodule: chaospy.external: SampleDist (KDE), OTDistribution (OpenTURNS), scipy_stats.
- Update Chaospy logo.
- CHANGED:
Replace chaospy.bertran.operators.bertran_indices with chaospy.bertran.bindex:
- Faster execution by using more numpy for heavy lifting
Moved chaospy.{quad -> quadrature} to finalize the refactor from v3.0.6.
Documentation polish to chaospy.quadrature.
- FIXED:
- Bugfixes in handling of three-terms-recursion
- REMOVED:
- Remove chaospy.quad.collection.probabilitic as it is much easier to implement from the user side.
- ADDED:
- Added license to setup.py
- New quadrature rules (thanks to Nico Schlömer): Gauss-Lobatto, Gauss-Kronrod, Gauss-Radau, Newton-Cotes.
- CHANGED:
- Update CircleCI to test for Python versions 2.7.16, 3.6.8 and 3.7.3
- Update dependencies.
- Refactored chaospy.quadrature: * Standardize quadrature interface. * Lots of new docs.
- Move version number chaospy.{version -> __init__}.
- REMOVED:
- Deprecating chaospy.distributions.collection.raised_cosine as hyp1f2 is no longer supported by scipy.
- Removing local set_state for Sobol indices and instead rely on numpy.random's random seed.
- ADDED:
- Added new method Dist._range to override the lower and upper bound calculations on some distributions.
- Added readme to setup.py
- CHANGED:
- Adding caching to some of the functionality in chaspy.bertran
- Use new cached functions to improve on raw statistical moments of multivariate Gaussian and multivariate Student-T distributions.
- Update polynomial output, as update to Bertran changes a few things in str handle.
- ADDED:
- Adding chaospy.distributions.evauation submodule to deal with graph resolution.
- Added CircleCI tests for Python 2.7.15
- CHANGED:
- Update CircleCI Python {3.6.2 -> 3.7.1}
- Some adjustments added to support Python 2.
- REMOVED:
- Remove dependency to networkx (as evaluation now does this task).
- Deprecating chaospy.distributions.cores (as each distribution are now locally defined in chaospy.distributions.collection)
- FIXED:
- Fixes to CircleCI testing.
- ADDED:
- Sparse segmentation function chaospy.bertran.sparse:sparse_segment
- CHANGED:
- Move install source {ROOT/src/chaospy -> ROOT/chaospy}
- Documentation update (mostly chaospy.orthogonal).
- REMOVED:
- Deprecated cubature module; Does not work with the chaospy v3, and is hard to maintain.
- CHANGED:
Update install dependencies to newest version
Refactor documentation * Update Sphinx configuration to newest version * Restructured the documentation a bit to make more sense with the new
code.
- Added some extra docs here and there.
- ADDED:
- Added Fejer quadrature
- CHANGED:
Full refactor of the chaospy.dist submodule: * Move: chaospy.dist -> chaospy.distributions * Deprecate chaospy.distributions.graph in favor of new
chaospy.distributions.evaluation which will not depend on networkx and should be easier to maintain.
- Move distributions from the two files distributions.{cores,collection} to the folder distributions.collection, where each file now is one core and one (or more) wrapper(s).
- Rename some old distributions; Kept the old ones for now, but they issue deprecation warnings.
- Split distributions.copulas.collection into individual components.
- Tests distribution using black-list instead of current white-list system.
- Rewritten a lot of documentation.
Replace absolute import paths with relative ones.
Refactor chaospy.descriptives to look better docs and code wise.
Adapt to Python 2+3 support.
Turn on automatic logging for warnings and upwards