v0.3.0

JuliaBUGS v0.3.0

Diff since v0.2.4

Merged pull requests:

• Remove unnecessary code (#123) (@sunxd3)
• Add _inv function for ReverseDiff compat (#124) (@sunxd3)
• Update Project.toml (#125) (@yebai)
• Transform parser.jl content to a module for ease of understanding (#128) (@sunxd3)
• Change BUGSGraph generation function name (#129) (@sunxd3)
• Add WithinGibbs sampler implementations (#131) (@sunxd3)
• Improve parsers and @bugs macro (#136) (@sunxd3)
• Improve CollectVariables (#138) (@sunxd3)
• Bump some compat versions (#139) (@sunxd3)
• Add description for Leuk (#140) (@sunxd3)
• Rename analysis functions (#141) (@sunxd3)

Closed issues:

• Simple type inference to avoid using try_cast_to_int (#112)
• inv function error in gradient computation (#115)
• Inference tests with HMC are not enabled in runtests.jl (#117)
• Support condition/decondition syntax similiar to DynamicPPL. (#132)
• A "hello world" entry example is missing (#134)

v0.2.4

JuliaBUGS v0.2.4

Diff since v0.2.3

Merged pull requests:

• Add support for AdvancedMH (#120) (@sunxd3)
• Better error message for b[1][1] syntax (#121) (@sunxd3)
• Add is_transformed to compile (#122) (@sunxd3)

Closed issues:

• Current chain creation code in AHMCExt can generalize to a MCMCChainsExt and a lighter-weight AHMCExt (#92)
• evaluate!!(model, SamlingContext()) return a single SimpleVarInfo instead of tuple of varinfo and logp (#100)
• beta[1][1] pass the parser but should be illegal (#101)
• AbstractMCMC.sample with AdvancedMH.jl currently doesn't do bundle_samples (#104)
• By default all model parameters are in unconstrained space (#105)

v0.2.3

JuliaBUGS v0.2.3

Diff since v0.2.2

Merged pull requests:

• Update README, add section on parallel and distributed sampling (#109) (@sunxd3)
• Add getparams and get_params_varinfo functions (#113) (@sunxd3)
• More robust constant propagation in the compilation process (#116) (@sunxd3)
• Update Project.toml (#119) (@yebai)

Closed issues:

• Assess the possibility of doing parallel sampling and add to documentations. (#70)
• Luek and Luekfr are skipped because they rely on more thorough constant propagation (#95)
• dmnorm implementation should invserve the precision matrix (#103)
• Some code issues (#110)
• Skip evaluating RHS when it might be expensive (#111)
• Support for in-place construct of arrays (#114)

v0.2.2

JuliaBUGS v0.2.2

Diff since v0.2.1

Merged pull requests:

• Add kwargs pass through to plotting library extensions (#98) (@sunxd3)
• Fix wrong implementation of distributions (#106) (@sunxd3)
• Disallow empty indexing (#107) (@sunxd3)
• Update Project.toml (#108) (@sunxd3)

Closed issues:

• Add kwargs to the functions in plotting extensions (#91)
• graphplot(model; Spring(; dim = 3) errors (#99)
• Empty indexing like x[] should raise an error just like x[ , ] (#102)

v0.2.1

JuliaBUGS v0.2.1

Diff since v0.2.0

Merged pull requests:

• Explicitly convert the type of values before create a Chain (#97) (@sunxd3)

v0.2.0

JuliaBUGS v0.2.0

Diff since v0.1.1

Merged pull requests:

• Move code around for better understandability (#85) (@sunxd3)
• Remove link functions and refactor tests (#86) (@sunxd3)
• Fix math typo in README (#87) (@yebai)
• Support trans-dimensional bijectors, flatten MarkovBlanketCoveredModel (#88) (@sunxd3)
• Fix broken case when variables are multi-dimensional and add tests for AdvancedHMC extension and inference (#89) (@sunxd3)
• Add birats to test (#90) (@sunxd3)
• Fix some compilation related bugs; fix AdvancedHMCExt so that the values stored to the Chain are in original space (#93) (@sunxd3)
• Update Project.toml (#96) (@sunxd3)

Closed issues:

• Models with dwish and ddirich are not yet supported with AdvancedHMC (#66)
• Add logdensity_and_gradient and/or few cycles of HMC sampling to the tests (#71)

v0.1.1

JuliaBUGS v0.1.1

Diff since v0.1.0

v0.1.0

JuliaBUGS v0.1.0

Closed issues:

• Add a lot of test models (#1)
• Actual parser -- external language? (#2)
• Document BUGS syntax that is difficult to support via Julia macro (#3)
• String manipulation to convert BUGS code into Julia AST? (#4)
• Support interpolation in @bugsast (#5)
• Use pattern matching (#6)
• Support incremental construction of GraphInfo (#16)
• Interesting package that might help implement our BUGS parser (#7)
• Transform syntax errors (#8)
• Type checking at IR level instead of AST level (#9)
• Roadmap to going public (#11)
• Review Readme and other old files lying around; clean up (#12)
• @bugsmodel_str doesn't support multiple-line expressions (#13)
• Documentation Checklist (#14)
• Fix the example in README (#17)
• Fix function identifier conflicts and precompilation (#18)
• Separate model construction and conditioning of toturing. (#19)
• Disallowing using stochastic variables as loop bounds/indices(even when they have observed values). (#20)
• Warn user about BUGS-style link function syntax. (#21)
• Only stochastic variable is reported in MCMCChain summury (#25)
• Setting up CI workflows for docs, tests and formating. (#26)
• Compile BUGS model to multiple Turing models / Distributions to enable Gibbs sampling (#28)
• Additional compilation target: LogDensityProblems (#30)
• More Julian syntax (#31)
• Verify the BUGS distribution functions in primitives.jl are correct. (#32)
• Add logdensity correctness tests for all BUGS example models (#39)
• Add support for DataFrames.jl (#42)
• Add LogdensityProblemsAD extension (#49)
• Update README for the new parse and compile function (#54)
• Implement AbstractMCMC.bundle_samples for MCMC samples conversion (#63)

Merged pull requests:

• Steps towards a functional compiler from Bugs to GraphPPL. (#10) (@phipsgabler)
• Towards the first public release. (#15) (@sunxd3)
• Fixed a small typo. (#22) (@yebai)
• Remove Turing dependency. (#23) (@sunxd3)
• Improve documentations. (#24) (@sunxd3)
• Stochastic indexing (#29) (@yebai)
• Formatting and improve docs. (#33) (@sunxd3)
• Fix bugs related to stochastic indexing. (#34) (@sunxd3)
• Fix link function syntax in tests. (#35) (@yebai)
• Fix testCI fails. (#36) (@sunxd3)
• More efficient dependency graph based compiler for BUGS (#37) (@yebai)
• LogDensityProblems target (#38) (@sunxd3)
• Fix errors to successfully compile and compute logdensity for all examples from Volume 1. (#41) (@sunxd3)
• Global variable store (#43) (@yebai)
• Some tiny fix (#44) (@sunxd3)
• Update docstrings for distributions (#45) (@sunxd3)
• Raise Julia version to 1.8 (#46) (@sunxd3)
• Improve docs (#47) (@sunxd3)
• Some refactoring (#48) (@sunxd3)
• Testing and bugs fixes with DynamicPPL (#50) (@sunxd3)
• Fix name clash for test models. (#51) (@yebai)
• Add a parser for BUGS (#52) (@sunxd3)
• Some optimization to the compilation time (#53) (@sunxd3)
• Update parser.md (#55) (@yebai)
• Markov Blanket Interface and some other minor fixes. (#56) (@sunxd3)
• Add link to BUGS syntax BNF definition (#57) (@yebai)
• Fix Typos in ReadMe (#58) (@ParadaCarleton)
• Update README.md (#60) (@yebai)
• Some minor improvements of README (#61) (@yebai)
• Switch to ADgradient (#62) (@yebai)
• Remove AuxiliaryNodes from the final graph (#64) (@sunxd3)
• Delete JuliaBUGSMCMCChainsExt.jl (#65) (@yebai)
• Restrict model type for bundle_samples (#67) (@yebai)
• Remove auxiliary nodes plotting from graph extensions. (#68) (@yebai)
• Restrict sampler to subtypes AbstractHMCSampler (#72) (@yebai)
• Improve documentations. (#73) (@sunxd3)
• Structure the folder of SymbolicPPL code and add README (#74) (@sunxd3)
• Update README.md (#75) (@yebai)
• Update graph_plotting.md (#76) (@sunxd3)
• Add compat and license (#77) (@sunxd3)
• Fix upper bounds for dependencies. (#78) (@yebai)
• Switch to inner constructor for MarkovBlanketCoveredBUGSModel (#79) (@sunxd3)