Skip to content

Latest commit

 

History

History
108 lines (82 loc) · 3.33 KB

CHANGELOG.md

File metadata and controls

108 lines (82 loc) · 3.33 KB

Releases

1.0.2

  • updated dependencies to use CUDA 3.3

1.0.1

  • fixed: the fromdict function cannot be used with subtype of DefaultParameter

1.0.0

  • added functions todict and fromdict to allow easy storing of the parameters
  • project is now in version 1.0.0 (starting backward compatibility updates)

0.10.0

  • package is compatible with CuArray from the CUDA package
  • added reallocate function to change distribution's parameter type (i.e. from Array to CuArray)
  • Gamma distribution is "vectorized" i.e. it represents the joint distribution of independent Gamma distributed variables
  • remove Zygote (gradient of lognormalizers are implemented explicitely)
  • replace StatsFuns with SpecialFunctions package
  • license is MIT again (main reason: CeCILL-B is not OSI approved)
  • updated author's email in Project.toml

0.9.3

  • fixed: no type specialization for the splitgrad function to allow using AD toolkit such as AutoGrad.jl

0.9.2

  • replaced ForwardDiff with Zygote for efficiency reason
  • added Zygote adjoint for vec_tril and its inverse
  • specialized the Wishart gradlognorm function for optimal performances
  • fixed latex equations in the documentation

0.9.1

  • fixed: typo in calling Diagonal
  • added: test for the DefaultParameter struct

0.9.0

  • refactored parameters API:
    • abstract parameter structure and interface
    • default parameter implemenation
  • support julia 1.6 onward
  • change LICENSE to CeCILL-B

0.8.0

  • added an abstract type for each distribution to allow easy extension of the pre-defined distributions.
  • Wishart natural parameters and sufficient statistics are more compact (Symmetric matrices are replaced by a their diagonal and the lower triangular parts)
  • each distribution can have arbitrary parameterization through the Parameter object.
  • computation are differentiable (with respect to Zygote) to allow use of this package with automatic differentiation packages.
  • added the function splitgrad to replace the vectorize parameter in the gradlognorm function.

0.7.0

  • changed the parameterization of the Wishart: the precision matrix is decomposed into its diagonal and the lower-triangular part of the matrix.
  • added the mu parameter to the kldiv function: this parameter allows to provide directly the expectation of the sufficient statistcs and allows to easily compute the natural gradient of the KL digervence.

0.6.0

  • changed parameterization of the Normal: the sufficient statistics are composed of x, the diagonal of xx^T and the lower-triangular part of xx^T. This change is to simplify gradient-based inference algorithms: with this parameterization, they don't need to ensure that the resulting matrix will be symmetric.
  • added vec_tril and inv_vec_tril function to easily extract the lower-triangular part of a square matrix.

0.5.0

  • added sample function

0.4.0

  • added the Wishart distribution
  • bugfix: gradlognorm(gamma, vectorize = false) now returns a tuple

0.3.1

  • bugfix: added MIME type to the Base.show redifinition, this avoid to have chaotic printing of arrays of distributions
  • added this CHANGELOG file to the project

0.3.0

  • added function stdparam to be able to convert the natural parameters to the standard ones.

0.2.0

  • Added "delta" distributions

0.1.0

  • initial release