Skip to content

Latest commit

 

History

History
236 lines (190 loc) · 10.5 KB

CHANGELOG.md

File metadata and controls

236 lines (190 loc) · 10.5 KB
Raw

Changelog

All notable changes to the UQTestFuns project is documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

Added

  • The three-dimensional sensitivity test function from Moon (2010).
  • Two new abstract base classes are added, namely UQTestFunFixDimABC and UQTestFunVarDimABC to deal with the construction of UQ test functions of fixed and variable dimensions, respectively. Both abstract classes are derived from UQTestFunABC such that the interfaces remain consistent.
  • function_id and input_id are now property of ProbInput.
  • output_dimension is now property of UQTestFunBareABC and inherited to all concrete classes of UQ test functions.
  • Printing a test function instance now shows whether the function is parameterized or not.
  • The information related to the parameterization of a function is now shown in the output of list_functions()
  • New class FunParams to organize function parameters.
  • The six-dimensional and ten-dimensional Friedman functions from Friedman et al. (1983) and Friedman (1991), respectively.
  • The three-dimensional simple portfolio model from Saltelli et al. (2004).
  • The 20-dimensional polynomial test function from Alemazkoor and Meidani (2018).
  • The exponential distribution as a distribution of UnivDist instances.

Changed

  • Application tags are now displayed when an instance of test function is printed on the terminal.
  • list_functions() is now printed in grid format and include information regarding the output dimension and the parameterization. Furthermore, filtering can be done based on the input dimension, output dimension, tag, and parameterization.
  • The class UnivDist has been renamed to Marginal. The name more clearly refers to one-dimensional marginal distributions (of a univariate random variable), which form a ProbInput.
  • The property spatial_dimension of ProbInput and UQTestFunBareABC is renamed to input_dimension for clarity (as opposed to output_dimension).
  • The property name of UQ test function instances has been renamed to function_id that implies uniqueness although it is not strictly enforced.
  • The parameter in the Gramacy 1D sine function is now removed. Noise can be added on the fly if needed.
  • evaluate() abstract method is now must be implemented directly in the concrete UQ test function; eval_() in the UQTestFunABC has been removed.

0.4.1 - 2023-10-27

Added

  • The two-dimensional polynomial function of high-degree for metamodeling exercises from Alemazkoor and Meidani (2008).
  • New tutorials (how the package may be used in a sensitivity analysis or reliability analysis exercises) have been added to the documentation following the review process in the submission of the package to the Journal of Open Source Software (JOSS).

Changed

  • The documentation landing page now includes explicit statement regarding the purpose of the package.

Fixed

  • Several typos in the documentation have been fixed with an additional minor improvements overall.

0.4.0 - 2023-07-07

Added

  • The two-dimensional convex failure domain problem for reliability analysis exercises from Borri and Speranzini (1997).
  • The two-dimensional Quadratic RS problem for reliability analysis exercises from Waarts (2000). This is a variant of the classic RS problem with one quadratic term.
  • The one-dimensional damped cosine function for metamodeling exercises from an example in Santner et al. (2018).
  • The two-dimensional circular bar RS problem for reliability analysis exercises taken from an example in Verma et al. (2015).
  • The two-dimensional polynomial function with random inputs from Webster et al. (1996) for metamodeling exercises.
  • New instance method for UnivDist and ProbInput classes called reset_rng(). When called (optionally with a seed number), a new instance of NumPy default RNG will be created and attached to the instance.
  • GitHub actions now include testing on Python v3.11 via Tox.

Changed

  • rng_seed_prob_input keyword parameter has been removed from the list of parameters to the constructor of all UQ test functions. The accepted way to reset an RNG with a seed is to use the instance method reset_rng() (optionally with a seed number) of the ProbInput instance attached.
  • Some background information in the documentation has been changed to match the description in the JOSS paper draft.

Fixed

  • A mistake in one the parameter values of the Sobol'-G function has been fixed.

0.3.0 - 2023-07-03

Added

  • The two-dimensional Gayton Hat function from Echard et al. (2013) used in the context of reliability analysis.
  • The eight-dimensional damped oscillator reliability problem from Der Kiureghian and De Stefano (1990); the problem is based on the existing Damped Oscillator model in the code base.
  • The two-dimensional hyper-sphere bound reliability problem from Li et al. (2018).
  • The two-dimensional cantilever beam reliability problem from Rajashekhar and Ellington (1993).
  • The two-dimensional four-branch function for reliability analysis from Katsuki and Frangopol (1994).
  • The five-dimensional speed reducer shaft reliability problem from Du and Sudjianto (2004).
  • The two-dimensional reliability problem of a circular pipe crack under a bending moment under Verma et al. (2015).
  • New docs section on list of functions for reliability analysis including a brief description on the reliability analysis problem.

0.2.0 - 2023-06-26

Added

  • The two-dimensional Franke functions (1st, 2nd, 3rd, 4th, 5th, and 6th), relevant for metamodeling exercises, are added as UQ test functions.
  • The two-dimensional McLain functions (S1, S2, S3, S4, and S5), relevant for metamodeling exercises, are added as UQ test functions.
  • An implementation of the Welch et al. (1992) test function, a 20-dimensional function used in the context of metamodeling and sensitivity analysis.
  • Four M-dimensional test functions from Bratley et al. (1992) useful for testing multi-dimensional numerical integrations as well as global sensitivity analysis methods.
  • Add a new parameterization to the Sobol'-G function taken from Bratley et al. (1992) and Saltelli and Sobol' (1995).
  • An implementation of the one-dimensional function from Forrester et al. (2008). The function was used as a test function for optimization approaches using metamodels.
  • An implementation of the Gramacy (2007) one-dimensional sine function, a function with two regimes.
  • Two base classes are now available UQTestFunBareABC and UQTestFunABC. The former is used to implement a bare UQ test function (with only evaluate() and ProbInput), while the latter is used to implement published UQ test functions in the code base (i.e., with additional metadata such as tags and description).
  • An instance of NumPy random number generator is now attached to instances of UnivDist and ProbInput. The random seed number may be passed to the corresponding constructor for reproducibility.
  • CITATION.cff file to the code base.

Changed

  • The date format in CHANGELOG.md has been changed from YYYY-DD-MM to the ISO format YYYY-MM-DD.
  • The bibliography style in the docs has been changed to 'unsrtalpha' (alphanumeric labels, sorted by order of appearance).
  • When list_functions() is called with a tag argument, then the application tags are no longer displayed to save terminal spaces.
  • The one-dimensional OakleyOHagan1D function has been renamed to Oakley1D.

Fixed

  • The original citation for the Sobol'-G function has been fixed; it is now referred to Saltelli and Sobol' (1995).
  • If a function is used as parameters in a test function (e.g., if variable dimension), then it must have the keyword parameter "spatial_dimension" for the function to be called when an instance of a UQ test function is created. This is to allow an arbitrary function (without a parameter named "spatial_dimension") to be a parameter of UQ test function.
  • One-dimensional test function now returns a one-dimensional array.

0.1.1 - 2023-03-07

Fixed

  • v0.1.0 was erroneously already registered at PyPI; latest version of UQTestFuns that was planned for v0.1.0 is now v0.1.1
  • Missing link in CHANGELOG.md

0.1.0 - 2023-03-07

Added

  • Publishing to PyPI is now automated once a tagged (with semantic version) release is carried out via GitHub
  • Add a few additional classifiers in setup.cfg for the PyPI record
  • DOI from Zenodo in README.md

Fixed

  • Wrong classifier specification in setup.cfg causing upload to PyPI to fail
  • Issue with RTD document built crashing from time to time; probably due to a problematic matplotlib version

Changed

  • The HTML representation of ProbInput instances now takes less space
  • Relax the numerical tolerance of a test (i.e., univariate beta distribution)
  • Minor edit in the docs

0.0.1 - 2023-03-06

First public release of UQTestFuns.

Added

  • An abstract class (UQTestFunABC) to unify the interface of a UQ test function
  • Probabilistic input modeling (via ProbInput class) with a joint independent distribution function
  • UnivDist class to represent one-dimensional marginal distributions (a univariate continuous random variable)
  • Nine univariate distributions
  • A total of 11 UQ test functions (concrete implementations of UQTestFunABC) typically used for the metamodeling, sensitivity analysis, and optimization applications
  • A concrete class implementation (UQTestFun) to create a UQ test function on runtime
  • A minimal documentation built using Jupyter Book
  • CI/CD to build and serve the documentation on ReadTheDocs
  • Mirror GitHub action to the CASUS organization