NEWS

New in version 1.6.1:

* fix fft imports in newer versions of numpy
* other bug fixes

New in version 1.6:

* Python 3 support (single code base used for Python 2 and 3)
* bug fixes
* disable parallel computiation by default on non-Posix systems
  (set pacal.params.parallel=True to enable)
* this is the last version of PaCal with support for Python 2

New in version 1.5.1:

* implement the tanh function
* fixes for newest versions of numpy and scipy
* fix for parallel computation with openblas

New in version 1.5:

* Initial implementation of arithmetic of dependent random variables
  (Bayesian network style)
* Functionality not related to dependent variables is as stable as in
  previous releases, so it is safe to use 1.5 as default
* Parallel computation (multicore).  Set params.general.parallel=False
  to disable.

New in version 1.1:

* operations on two dependent random variables
* joint distribution models: normal, copulas
* joint distribution of two order statistics
* PaCal now depends on sympy

New in version 1.0:

* order statistics
* noncentral distributions: Chi^2, F, Beta and T
* optimized routines for sum, average, max, min etc. of i.i.d. random variables
* Extreme Value Distributions: Gumbel, Weibull, Frechet
* better accuracy of Beta distribution
* methods interp_error and interp_error_by_segment return estimates of
  interpolation errors
* a warning is issued when dependent RVs are used in operations
  (use params.general.warn_on_dependent = False to suppress)
* bug fixes