Skip to content

HISTORY.md

Latest commit

 

History

History
67 lines (36 loc) · 2.65 KB

HISTORY.md

File metadata and controls

67 lines (36 loc) · 2.65 KB

0.15.24

Export a few more functions from Bijectors.VectorBijectors, namely scalar_to_scalar_bijector, TypedIdentity, Log, and Untruncate.

0.15.23

Implement vector bijectors for Distributions.MvTDist.

0.15.22

Implement vector bijectors for Bijectors.TransformedDistribution.

0.15.21

Add compatibility with Roots.jl v3.

0.15.20

Export Bijectors.VectorBijectors.has_constant_vec_bijector to allow downstream packages to overload it.

0.15.19

Optimise performance of vector bijectors for products of univariate distributions.

0.15.18

Implemented vector bijectors for Bijectors.OrderedDistribution.

0.15.17

Adds a new module, Bijectors.VectorBijectors. This module implements bijectors that always return vectors as the output of a linking operation. This provides additional guarantees over the existing Bijectors interface, which in general will return some collection of values that are in Euclidean space, but whose shape is not generally a vector (it could be a scalar or a multidimensional array). The intention is to make it easier to implement bijectors that are compatible with libraries that use vector inputs, such as LogDensityProblems.jl (which is in turn heavily used in the Turing ecosystem).

Please see the documentation for further information.

0.15.16

Added compatibility with Mooncake.jl v0.5.

0.15.15

Fixed a case where calling various combinations of VecCholeskyBijector, PDBijector, PDVecBijector, and their inverses on a wrong input would cause a stack overflow instead of a more normal error.

0.15.14

Added a new documentation page describing how to implement custom bijectors (this release contains no code changes; it only exists to make sure that the 'stable' docs are updated).

0.15.13

Exports extra functionality that should probably have been exported, namely ordered, isinvertible, and columnwise, from Bijectors.jl.

The docs have been thoroughly restructured.

0.15.12

Improved implementation of the Enzyme rule for Bijectors.find_alpha.

0.15.11

Bijectors for ProductNamedTupleDistribution are now implemented.

Bijectors.output_size is now exported. This function provides information about the size of transformed variables. There are two main invocations:

  • output_size(b, input_size::Tuple) returns the size of the output of b, given an input that has size input_size.
  • output_size(b, dist::Distribution) returns the size of the output of b, given an input sampled from distribution dist. For most distributions this is implemented by calling output_size(b, size(dist)); however, ProductNamedTupleDistribution does not implement size, so this method is necessary.