Releases | Open Source Science

StochasticAD v0.1.26

Merged pull requests: - Add basic reverse mode (#129) (@gaurav-arya) - Fix formatting issues in note (#130) (@gaurav-arya)

- Julia
Published by github-actions[bot] over 1 year ago

StochasticAD v0.1.25

Diff since v0.1.24

Merged pull requests: - Small tweaks for type stability / inferrability issues (#122) (@gaurav-arya) - Update patch version (#123) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 2 years ago

StochasticAD v0.1.24

Diff since v0.1.23

Merged pull requests: - Miscellaneous interface improvements (#119) (@gaurav-arya) - Bump patch version (#120) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 2 years ago

StochasticAD v0.1.23

Diff since v0.1.22

Merged pull requests: - Add randst (#115) (@gaurav-arya) - Add 1-arg randst method (#116) (@gaurav-arya)

Closed issues: - Typo in the paper for the categorical distribution w+ and w-? (#109) - Add more general rand-based interface to allow users to specify custom couplings, etc. (#111) - Most simple function to differentiate (#112) - Different results inputing 0.4 and 0.40 (#113) - Add documentation + example for StochasticAD.propagate (#114)

- Julia
Published by github-actions[bot] about 2 years ago

stochasticad.jl - v0.1.22

StochasticAD v0.1.22

Diff since v0.1.21

Merged pull requests: - Fixes to devdocs (#106) (@gaurav-arya) - Small tweaks (#107) (@gaurav-arya) - Format (#108) (@gaurav-arya)

Closed issues: - Document StochasticAD.propagate (#95)

- Julia
Published by github-actions[bot] over 2 years ago

stochasticad.jl - v0.1.21

StochasticAD v0.1.21

Diff since v0.1.20

Merged pull requests: - Fix docs build (#104) (@gaurav-arya) - Add some basic developer docs for custom rules (#105) (@gaurav-arya)

Closed issues: - Fix docs build (#98)

- Julia
Published by github-actions[bot] over 2 years ago

stochasticad.jl - v0.1.20

StochasticAD v0.1.20

Diff since v0.1.19

Merged pull requests: - Refactor wrapping logic into an abstract type (#99) (@gaurav-arya) - Minor tweaks to random walk tutorial (#100) (@gaurav-arya) - Provide enumeration for discrete random distributions when possible (#101) (@gaurav-arya) - Make weightedmapΔs the underlying primitive for mapping and scaling perturbations (#102) (@gaurav-arya) - Handled signed weights correctly in pruning backend (#103) (@gaurav-arya)

- Julia
Published by github-actions[bot] over 2 years ago

stochasticad.jl - v0.1.19

StochasticAD v0.1.19

Diff since v0.1.18

Merged pull requests: - Add straight through strategy (#85) (@gaurav-arya) - Support duals in derivative contribution (#88) (@gaurav-arya) - Run formatter after bracket alignment removal (#89) (@gaurav-arya) - fix tutorial deps (#90) (@Moelf) - Add GeometryBasics as tutorials dep (#91) (@gaurav-arya) - Make strategy customizable via StrategyWrapperFIs (#94) (@gaurav-arya) - Tweaks to propagate (#96) (@gaurav-arya) - Some miscellaneous tweaks (#97) (@gaurav-arya)

Closed issues: - Comparing stochastic derivatives to derivatives from traditional AD (#86)

- Julia
Published by github-actions[bot] over 2 years ago

stochasticad.jl - v0.1.18

StochasticAD v0.1.18

Diff since v0.1.17

Merged pull requests: - Remove unused tag in lazy pruning backend (#87) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.17

StochasticAD v0.1.17

Diff since v0.1.16

Merged pull requests: - Add smooth_triple (#84) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.16

StochasticAD v0.1.16

Diff since v0.1.15

Closed issues: - Clean up interface for creating an empty backend (#51)

Merged pull requests: - Add stochastic triple smoothing backend (#81) (@gaurav-arya) - Make array indexing test try out all backends (#82) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.15

StochasticAD v0.1.15

Diff since v0.1.14

Closed issues: - Infinitesimal derivative part of StochasticTriples end up as NaN for some instances of MVNormal (#79)

Merged pull requests: - Fix zero'ing behaviour with NaN derivative component (#80) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.14

StochasticAD v0.1.14

Diff since v0.1.13

Merged pull requests: - Allow custom kwargs in map functions (#74) (@gaurav-arya) - Add smoothing frule (#75) (@gaurav-arya) - Propagate kwargs in alternate f eval (#76) (@gaurav-arya) - Don't preserve structs when structural mapping perturbations in propagate (#78) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.13

StochasticAD v0.1.13

Diff since v0.1.12

Merged pull requests: - Add interface for directional perturbations and derivatives (#73) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.12

StochasticAD v0.1.12

Diff since v0.1.11

Merged pull requests: - Switch to using singleton types to specify backend algorithm (#72) (@gaurav-arya)

- Julia
Published by github-actions[bot] almost 3 years ago

stochasticad.jl - v0.1.11

StochasticAD v0.1.11

Diff since v0.1.10

Merged pull requests: - Add experimental forward propagate function (internal) (#55) (@gaurav-arya) - Add scalarize to backend interface (#67) (@gaurav-arya) - Add accidentally-not-included simple ops benchmark to benchmarks (#68) (@gaurav-arya) - Minor improvements to smoothing rules (#70) (@gaurav-arya)