v0.5.0

November 18, 2024

v0.5.0 is a minor release that further expands the library of available UQ test functions. This update introduces 14 new test functions, bringing the total to 56.

See the complete CHANGELOG.

New features

Test functions

This release includes the following newly added test functions:

• Five-dimensional single-diode solar cell model: from Constantine et al. (2015) for sensivitiy analysis and metamodeling.
• Six-dimensional undamped non-linear oscillator function: for reliability analysis.
• M-dimension Sobol'-Levitan function: from Sobol' and Levitan (1999), for sensitivity analysis.
• M-dimensional test function: from Morris et al. (2006), for sensitivity analysis.
• Modified Sobol'-G test function (i.e., Sobol'-G)*: from Saltelli et al. (2010), sensitivity analysis.
• M-dimensional linear function: from Saltelli et al. (2008), for sensitivity analysis.
• Two-dimensional test function: from Cheng and Sandu (2010), for metamodeling exercises.
• Two-dimensional non-polynomial test function: from Lim et al. (2002), for metamodeling.
• Two-dimensional polynomial test function: from Lim et al. (2002), for metamodeling.
• Three-dimensional test function: from Moon (2010), for sensitivity analysis.
• Six-dimensional and ten-dimensional Friedman functions: from Friedman et al. (1983) and Friedman (1991), respectively.
• Three-dimensional simple portfolio model: from Saltelli et al. (2004), for sensitivity analysis.
• 20-dimensional polynomial test function: Alemazkoor and Meidani (2018), for metamodeling.
• New Sobol'-G parameter set: from Sun et al. (2022).

Documentation

• Added concise overviews of uncertainty quantifiction frameworks, metamodeling, and sensitivity analysis.

Probabilistic input

• Support for exponential distribution is now available.

Core

• Introduced FunParams, an abstraction for managing collection of fixed function parameters.
• function_id and input_id added as properties of ProbInput.
• The printout of test function instances now indicates the number of output dimension and whether the function is parameterized.
• Parameterization details are displayed in the output of list_functions()
• output_dimension added as a property of UQ test functions.

Changes

• Renamed spatial_dimension to input_dimension.
• UnivDist class renamed to Marginal, closely reflecting its role as a marginal distribution of a univariate random variable.
• name property of UQ test functions is now function_id.
• Removed noise parameterization from the Gramacy 1D test function; noise, if necessary, can be added on the fly after the function has been evaluated.

Fixes

• Corrected minor typos and grammatical mistakes in documentation and in-code comments.