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)