v0.8.0

AdvancedHMC v0.8.0

Diff since v0.7.1

Breaking changes

• To make an MCMC transition from phasepoint z using trajectory τ(or HMCKernel κ) under Hamiltonian h, use transition(h, τ, z) or transition(rng, h, τ, z)(if using HMCKernel, use transition(h, κ, z) or transition(rng, h, κ, z)).
• The initial_step_size in find_good_stepsize for heuristic search of a good initial leap-frog step-size can be manually specified, defaulting to 1//10.
• The printing interface has been upgraded to a more user-friendly design.

Merged pull requests:

• Revive GPU tests (#430) (@ErikQQY)
• Add quality tests (#431) (@ErikQQY)
• Fix CUDA tests (#433) (@ErikQQY)
• Allow specifying initial step-size (#434) (@ErikQQY)
• Unify argument order in phasepoint and transition (#435) (@ErikQQY)
• Skip CUDA tests when no CUDA devices are found. (#436) (@yebai)
• Move Riemannian HMC files (#438) (@yebai)
• DocsGHA: Grant write permission to pull-requests event (#441) (@shravanngoswamii)
• Fix experimental CI (#442) (@ErikQQY)
• Better printing (#446) (@ErikQQY)
• CompatHelper: bump compat for MCMCChains in [weakdeps] to 7, (keep existing compat) (#448) (@github-actions[bot])
• Update CHANGELOG (#451) (@ErikQQY)

Closed issues:

• Canonical momentum and kinetic momentum (#129)
• Vectorize NUTS (#140)
• Discontinuous Hamiltonian Monte Carlo for discrete parameters (#209)
• A support list of all combinations (#212)
• Initial Epsilon in find_good_step_size function (#216)
• Reduce number of allocations (#224)
• Mixed Hamiltonian Monte Carlo for Mixed Discrete and Continuous Variables (#234)
• Unify argument orders (#248)
• Disable warning about rejection because of numerical errors (#268)
• Integration test for Turing (#269)
• Forecasting (#308)
• Objects of type LogTargetDensity are not callable (#324)
• Retest is broken (#327)
• Define a model (#363)
• GPU tests via buildkite (#429)

v0.7.1

AdvancedHMC v0.7.1

Diff since v0.7.0

Merged pull requests:

• Reduce allocations in stepsize.jl (#390) (@devmotion)
• Simplify find_good_stepsize (#394) (@devmotion)
• Remove SimpleUnpack (#401) (@ErikQQY)
• Set up formatting (#404) (@ErikQQY)
• Allow ADTypes AD backend selection in Hamiltonian (#405) (@ErikQQY)
• Format README (#406) (@ErikQQY)
• Add ComponentArrays extension (#407) (@ErikQQY)
• Ignore formatting commit in git-blame (#408) (@devmotion)
• Remove Requires dependency (#409) (@devmotion)
• Better documentation (#417) (@ErikQQY)
• Simpler worked example in README.md (#421) (@yebai)
• Add examples for AD backend switching (#423) (@ErikQQY)
• Add HISTORY.md (#425) (@ErikQQY)
• Fix gradient length mismatch in PhasePoint (#426) (@ErikQQY)
• Reduce allocation in mass matrix adaptor (#427) (@ErikQQY)

Closed issues:

• Moving ReadMe to Docs (#330)
• Component Arrays with DenseEuclideanMetric fails (#344)
• AdvancedHMC.sample might not be respecting the passed rng object (#379)
• Improve AHMC documentation with tutorials, API docs etc. (#402)
• update readme and docs to use ADTypes (#413)
• Introduce a HISTORY.md for breaking/major changes (#419)
• Minor doc inconsistency (#422)

v0.7.0

AdvancedHMC v0.7.0

Diff since v0.6.4

Breaking changes

• Type piracies of Base.rand and Base.randn for vectors of RNGs are removed: Replace rand(rngs::AbstractVector{<:Random.AbstractRNG}) with map(rand, rngs), randn(rngs::AbstractVector{<:Random.AbstractRNG}) with map(randn, rngs), rand(rngs::AbstractVector{<:Random.AbstractRNG}, T, n::Int) (for n == length(rngs)) with map(Base.Fix2(rand, T), rngs), and randn(rngs::AbstractVector{<:Random.AbstractRNG}, T, m::Int, n::Int) (for n == length(rngs)) with eg reduce(hcat, map(rng -> randn(rng, T, m), rngs)).
• Type piracy Base.isfinite(x::AbstractVecOrMat) is removed: Switch to all(isfinite, x) if you (possibly implicitly) relied on this definition
• Abstract fields of NesterovDualAveraging, HMCDA, SliceTS, and MultinomialTS are made concrete by adding type parameters: Update occurrences of these types (eg. in function signatures) if necessary
• Definitions of Base.rand for metrics are removed: Use the (internal) AdvancedHMC.rand_momentum function if you depend on this functionality and open an issue to further discuss the API

Merged pull requests:

• Initial AbstractMCMC.step should not sample (#366) (@torfjelde)
• Revert "Initial AbstractMCMC.step should not sample" (#385) (@torfjelde)
• Documentation and Turing Navigation CI improvement (#386) (@shravanngoswamii)
• Fix rand and randn type piracy (#387) (@devmotion)
• Reduce allocations in all + any and remove isfinite type piracy (#388) (@devmotion)
• Remove DEBUG constant (#389) (@devmotion)
• FIx incorrect ratios of divergent transitions (#391) (@devmotion)
• Use Random.default_rng() instead of Random.GLOBAL_RNG (#392) (@devmotion)
• Use Random.randexp (#393) (@devmotion)
• Add Aqua tests (#395) (@devmotion)
• Rename testset to avoid that it's overwritten by ReTest (#396) (@devmotion)
• Use Julia binaries matching the runner's architecture in CI (#397) (@devmotion)
• Reduce allocations of pm_next! (#398) (@devmotion)
• Fix fields with abstract types (#399) (@devmotion)
• Remove rand overloads for sampling of momentum (#400) (@devmotion)

Closed issues:

• Step size initialization uses GLOBAL RNG instead of reproducible RNG from caller (#306)
• Weighted Log Likelihood #featurerequest (#317)
• In sample progress=true prints more than a progress-bar (#364)
• Progress message ratio_divergent_transitions_during_adaption bug (#373)

v0.6.4

AdvancedHMC v0.6.4

Diff since v0.6.3

Merged pull requests:

• Specialize on Types and use if-else instead of Val-dispatches (#383) (@devmotion)

v0.6.3

AdvancedHMC v0.6.3

Diff since v0.6.2

Merged pull requests:

• Update GH actions and add dependabot (#377) (@devmotion)
• Add getparams and setparams!! following AbstractMCMC v5.5 and v5.6 (#378) (@sunxd3)

v0.6.2

AdvancedHMC v0.6.2

Diff since v0.6.1

Merged pull requests:

• Fix Step (#359) (@JaimeRZP)
• Fix broken links and tags in docs (#367) (@mhauru)
• added navabr workflow (#368) (@jeetsuthar)
• feat: support position-dependent kinetic (#369) (@xukai92)
• Generalized leapfrog integrator (#370) (@xukai92)
• Update DocsNav.yml (#371) (@shravanngoswamii)
• Use Base.RefValue (#374) (@devmotion)
• Use n_adapts instead of nadapts (#375) (@devmotion)
• Require AbstractMCMC 5 (#376) (@devmotion)

v0.6.1

AdvancedHMC v0.6.1

Diff since v0.6.0

Merged pull requests:

• Allow CUDA v5 (#360) (@ChrisRackauckas)
• fix: conflict of @main from Base (nightly) (#362) (@xukai92)

v0.6.0

AdvancedHMC v0.6.0

Diff since v0.5.5

Merged pull requests:

• fix: arg order (#349) (@xukai92)
• CompatHelper: bump compat for AbstractMCMC to 5, (keep existing compat) (#352) (@github-actions[bot])
• Deprecate init_params which is no longer in AbstractMCMC (#353) (@torfjelde)
• CompatHelper: add new compat entry for Statistics at version 1, (keep existing compat) (#354) (@github-actions[bot])
• Removed deprecation of init_params + bump minor version (#355) (@torfjelde)
• Fix some tests. (#356) (@yebai)
• Fix docs CI (#357) (@yebai)

Closed issues:

• Doc string error for NUTS (#346)

v0.5.5

AdvancedHMC v0.5.5

Diff since v0.5.4

Merged pull requests:

• Find step_size if zero (#348) (@yebai)

v0.5.4

AdvancedHMC v0.5.4

Diff since v0.5.2

Closed issues:

• Riemannian Manifold HMC (#6)
• Sensible Defaults (#184)
• Struct naming improvments (#246)
• Broken link in README (#256)

Merged pull requests:

• Fix duplicate of include and some bugs in tests (#343) (@yebai)
• Fixes for constructor docstrings (#347) (@torfjelde)