v1.4.0

ExponentialFamily v1.4.0

v1.3.0

ExponentialFamily v1.3.0

Diff since v1.2.2

Merged pull requests:

• Use Julia 1.10 for tests (#177) (@bvdmitri)

v1.2.2

ExponentialFamily v1.2.2

Diff since v1.2.1

Merged pull requests:

• Faster logpdf implementation for container based inputs (#147) (@Nimrais)
• add getgradlogpartition function for Bernoulli (#150) (@HoangMHNguyen)
• add getgradlogpartition function for Poisson (#151) (@HoangMHNguyen)
• add getgradlogpartition Exponential (#152) (@HoangMHNguyen)
• Add gradient of binomial (#153) (@ismailsenoz)
• Add gradient of beta (#154) (@ismailsenoz)
• Add gradient calculation for log partition (#157) (@bartvanerp)
• Add gradient of Wishart (#158) (@ismailsenoz)
• Add gradient for inverse wishart (#159) (@ismailsenoz)
• Add gradient of Weibull (#160) (@ismailsenoz)
• Gradient of vmf (#161) (@ismailsenoz)
• Add gradient of Rayleigh (#162) (@ismailsenoz)
• Add gradient Pareto (#163) (@ismailsenoz)
• Add gradient normal gamma (#164) (@ismailsenoz)
• Add gradient of negative binomial (#165) (@ismailsenoz)
• Add gradient of matrix Dirichlet (#166) (@ismailsenoz)
• Add gradient of log Normal (#167) (@ismailsenoz)
• Add gradient for categorical (#170) (@ismailsenoz)
• Add gradient inverse gamma (#171) (@ismailsenoz)
• Add gradient for Laplace (#172) (@ismailsenoz)
• Update README.md (#174) (@bartvanerp)

Closed issues:

• The logpdf from ExponentialFamily.jl is slower then from Distributions.jl (#146)

v1.2.1

ExponentialFamily v1.2.1

Diff since v1.2.0

Merged pull requests:

• Update the definition of logpdf_optimized for Gaussians (#143) (@bvdmitri)
• Relax compat constraints (#144) (@bvdmitri)

v1.2.0

ExponentialFamily v1.2.0

Merged pull requests:

• Add README.md (#3) (@albertpod)
• Add Erlang distribution (#14) (@albertpod)
• Update CI & toml (#15) (@albertpod)
• Exponential distribution (#16) (@wouterwln)
• CompatHelper: add new compat entry for StaticArrays at version 1, (keep existing compat) (#26) (@github-actions[bot])
• Fix normal tests (#27) (@ismailsenoz)
• Add formatter (#28) (@albertpod)
• added ef extensions for beta and tests (#29) (@ismailsenoz)
• Categorical ef extensions (#30) (@ismailsenoz)
• incorporate fix for ForwardDiff hessian error. (#33) (@ismailsenoz)
• Contingency ef extensions (#34) (@ismailsenoz)
• Simplify the interface for natural parameters (#36) (@ismailsenoz)
• Dev Von Mises (#37) (@ismailsenoz)
• add von mises fisher distribution (#38) (@ismailsenoz)
• Multinomial distribution (#39) (@HoangMHNguyen)
• Geometric distribution (#40) (@HoangMHNguyen)
• CompatHelper: add new compat entry for JuliaFormatter at version 1, (keep existing compat) (#41) (@github-actions[bot])
• add check for number of trials (#43) (@HoangMHNguyen)
• Add conditioner field to natural parameters structure and tests (#44) (@ismailsenoz)
• fix: Gamma and \approx (#45) (@Nimrais)
• fix: convert into one dist type (#46) (@Nimrais)
• Fix wishart conversions (#47) (@Nimrais)
• Add Pareto distribution (#48) (@Chengfeng-Jia)
• Generic plus (#49) (@ismailsenoz)
• docs: add general lognormalizer doc string (#50) (@Nimrais)
• Continuous bernoulli (#51) (@ismailsenoz)
• Binomial distribution (#53) (@albertpod)
• add Rayleigh and tests (#56) (@ismailsenoz)
• Fix exponential, clean binomial (#58) (@albertpod)
• Weibull (#59) (@Nimrais)
• Dev generic operations (#60) (@ismailsenoz)
• LogNormal distribution added. The Natural parameters are calculated f… (#61) (@Sepideh-Adamiat)
• Dev negative binomial (#62) (@ismailsenoz)
• add Laplace distribution and tests (#64) (@ismailsenoz)
• Fix negative binomial (#65) (@albertpod)
• add chi squared distribution and tests (#66) (@ismailsenoz)
• Add Poisson distribution extension (#67) (@albertpod)
• Fix prod for Binomial Distributions (#68) (@albertpod)
• Remove sample friendly (#69) (@ismailsenoz)
• Test alignment (#71) (@albertpod)
• Dev normal wishart (#72) (@ismailsenoz)
• Add Normal-Gamma distribution (#73) (@albertpod)
• Gammafix (#74) (@ismailsenoz)
• Update tests, README (#75) (@albertpod)
• Add Fisher Information for Poisson distribution (#76) (@albertpod)
• fisherinformation Gamma (#77) (@Nimrais)
• fisherinformation Chisq (#78) (@Nimrais)
• Fisher information for normal gamma (#79) (@Nimrais)
• Add fisher information for exponential distribution (#80) (@albertpod)
• Fisher Rayleigh (#82) (@albertpod)
• Add Fisher information Weibull distribution (#83) (@albertpod)
• add fisher information and tests for pareto distribution (#84) (@ismailsenoz)
• Fisher negative binomial (#85) (@Nimrais)
• Fix normal gamma tests (#87) (@albertpod)
• Fix isproper (#88) (@Nimrais)
• add fisher information for laplace and test (#89) (@ismailsenoz)
• fix Multinomial (#91) (@HoangMHNguyen)
• fix lognormal parameterization and add fisher information (#92) (@ismailsenoz)
• Fix categorical and contingency parameterisations (#93) (@HoangMHNguyen)
• Fisher information for Von Mises (#95) (@ismailsenoz)
• Fisher binomial (#96) (@ismailsenoz)
• feat: implement Fisher information for Univariate Normal (#97) (@Nimrais)
• Add Fisher for Geometric distribution (#98) (@albertpod)
• Add Fisher information for Multinomial distribution (#99) (@albertpod)
• Fix format throughout test and src folders (#100) (@albertpod)
• Fisher information for Erlang distribution (#101) (@ismailsenoz)
• Add reconstruct function (#102) (@albertpod)
• Fisher vmf (#103) (@ismailsenoz)
• Add Fisher to Categorical (#104) (@HoangMHNguyen)
• add fisher information for continuous Bernoulli (#105) (@ismailsenoz)
• Fisher Wishart (#106) (@albertpod)
• Add Fisher information for Multivariate Normal (#107) (@Nimrais)
• Fix format and tests (#108) (@albertpod)
• feat: fisher information InverseWishart (#109) (@Nimrais)
• Revise prod (#110) (@ismailsenoz)
• Sufficient statistics, logpdf, pdf and duplications (#113) (@ismailsenoz)
• Generic prod tests (#115) (@ismailsenoz)
• feat: add mean for KnownExponentialFamilyDistribution Bernoulli (#116) (@Nimrais)
• feat: add mean, cov methods for EF, fix: conversion for normal-gamma and wishart (#117) (@ismailsenoz)
• fix: resolve ambiguity (nothing case) (#118) (@Nimrais)
• Fisher information for Dirichlet (#119) (@ismailsenoz)
• Matrix Dirichlet Fisher information (#120) (@ismailsenoz)
• Fix wishart conversion (#124) (@ismailsenoz)
• Update README.md (#125) (@albertpod)
• Change ef structure (#127) (@ismailsenoz)
• Add documentation to the package (#128) (@albertpod)
• Fix docs (#129) (@albertpod)
• Major refactoring (#133) (@bvdmitri)
• Parallel tests + VSCode integration (#135) (@bvdmitri)
• Add Aqua to tests (#136) (@bvdmitri)
• Update README.md with href to mit license (#139) (@bartvanerp)
• Diffrentiabilty of ExponentialFamily(ExponentialFamily.MvNormalMeanCovariance) logpdf (#140) (@Nimrais)
• InverseGamma Fisher (#141) (@Nimrais)
• Integrate BayesBase.jl (#142) (@bvdmitri)

Closed issues:

• List of exponential family distributions that should be added (#1)
• Some distributions do not have the required exponential family methods (#2)
• Bernoulli log normalizer appears to be incorrect (#31)
• The interface for natural parameters is over-complicated (#32)
• Fisher information matrix (#35)
• Missing check for the number of trial in prod(), Base.+ and Base.- (#42)
• Inconsistency in methods arrangement and input variable names (#54)
• Categorical and contingency parameterizations (#55)
• Normal-Gamma sampling is not stable (#81)
• Issue with isproper for Negative Binomial (#86)
• Incorrect product of Multinomial (#90)
• Improve prod function and add comprehensive tests (#94)
• Sufficients statistics and conversions in the pdfs (#111)
• Logpdf out-of-support Traits (#114)
• Tests needs to be revised (#121)
• Generic conversion causes stackoverflow (#122)
• Conversion 'WishartImproper' to 'Distribution' is incorrect (#123)
• Add comprehensive documentation (#126)
• Pareto (#131)
• TagBot trigger issue (#138)