Release Notes

Release Notes#

Version 0.2.0#

  • \(N^{th}\)-order polynomial expansion

  • user-defined function input

  • Latin hypercube sample generation

  • command line inputs and options

  • BetaVariable

  • ExponentialVariable

  • GammaVariable

  • user-input Variable

Version 0.2.1#

  • bug fixes to sample generation for the general Variable

Version 0.2.2#

  • bug fixes to sample generation for BetaVariable, GammaVariable, and ExponentialVariable classes when variables have non-standard bounds

Version 0.2.3#

  • bug fixes to sample generation

Version 0.2.4#

  • updated sample generation that is accurate no matter how many points are generated

Version 0.2.5#

  • recursive variable basis equation for BetaVariable to reduce time

  • norm squared values for each variable calculated with scipy integral to reduce time and improve accuracy

  • PRESS residual

  • speed improvements

Version 0.3.0#

  • statistics

    • mean statistics

    • variance statistics

    • mean error statistics

    • R-squared

    • R-squared adjusted

  • uncertainty bounds

    • mean confidence interval

    • confidence interval uncertainty

    • Sobol uncertainty

    • coefficient uncertainty

    • mean uncertainty

  • discrete variables

    • user-input

    • Poisson

    • discrete uniform

    • negative binomial

    • hypergeometric

  • stepwise regression

  • A-optimal design

  • D-optimal design

  • report generation capability

  • adaptive sampling

  • parallelized using mpi4py

  • robust unittest

  • UQPCE wrapper

  • Robust Design Capability

  • orthogonal polynomial updates

    • replaced SymPy integrals in Gram-Schmidt method with scipy integrals to reduce time

    • added a symbolic option to calculate symbolic integrals for debugging purposes

  • improved variable initialization to allow user to easily use variables for other purposes

  • optional seeding for random values

  • less rounding error in variable basis and norm squared values

  • errors raised when users input parameters that are not in the correct ranges for the variable type

Version 1.0.0#

  • OpenMDAO compatability

  • analytical derivatives

  • stepwise regression

  • backward elimination

  • more efficient ProbabilityBox class

  • programmer-focused package and PCE object

  • restructured model construction to build N models simultaneously

  • improved logo

Contents