== EasyABC changelog ==


=== Version 1.5.1 ===

  * Bugfix in SABC implementation
  * fix email adress of maintainer
  * bugfix in distance computation considering weights, thanks to Clémentine Straub

=== Version 1.5 ===
  
  * new algorithm introduced by Carlo Albert: Simulated Annealing
  * bugfix in Delmoral implementation distance computation (thanks to Milton Pifano)
  * bugfixes from Stephen Nawara

=== Version 1.4 ===

  * optional weighted euclidean distance (suggested by Wim Delva and Maxime Lenormand)
  * new method based on emulation
  * bugfixes in cluster mode
  * bugfix: NA values for summary statistics are excluded from the computed points

=== Version 1.3.1 ===

  * bugfix in cluster mode for MCMC functions
  * feature restored: constants description in prior

=== Version 1.3 ===

  * bugfix in compute_dist methods (thanks to Matteo Fasiolo): variance is now fixed in a better way
  * some code optimizations thanks to Matteo Fasiolo
  * possibility to use user-defined prior distributions
  * possibility to add constraints to prior distributions (e.g. parameter 1 < parameter 2)
  * examples of how to link Java models or shell scripts to EasyABC

=== Version 1.2.2 - 3 June 2013 ===

  * bugfix in data reading from system call in binary_model function (bug reported by Albert Ko)
  * fixes for line width compliance with R-devel

=== Version 1.2.1 - 17 May 2013 ===

  * compilation fix for Solaris

=== Version 1.2 - 27 February 2013 ===

  * intermediary results are stored in a list for sequential algorithms (with the option verbose=TRUE)
  * various minor corrections in the help files and in the vignette.

=== Version 1.1 - 29 January 2013 ===

  * simple toy model introduced for better examples
  * algorithm parameters have now default values
  * use_seed parameter is now set to False by default
  * four prior distributions are now availables: uniform, normal, lognormal and exponential
  * verbose mode enhanced
  * tigher integration of package abc for rejection method
  * bugfix: data handling when model take only one parameter