v0.12.5

Fix the optimizer to use all of the optimizers available through the Optim.jl package.

v0.12.4

<compat update>

v0.12.3

GaussianProcesses v0.12.3

Diff since v0.12.2

Closed issues:

• Remove or parameterize the positive definite retry mechanism (#131)

Merged pull requests:

• remove positive definite retry mechanism (#171) (@maximerischard)
• CompatHelper: bump compat for "Distributions" to "0.24" (#172) (@github-actions[bot])
• CompatHelper: bump compat for "Distances" to "0.10" (#173) (@github-actions[bot])
• CompatHelper: bump compat for "SpecialFunctions" to "1.0" (#175) (@github-actions[bot])
• CompatHelper: bump compat for "StaticArrays" to "1.0" (#176) (@github-actions[bot])

v0.12.2

GaussianProcesses v0.12.2

Diff since v0.10.0

Closed issues:

• Regression with known error bars (#70)
• Reduce size of notebooks (#121)
• Failed to compile GaussianProcesses.jl (#123)
• identifying parameters from output (#130)
• Vector-valued GPs (#140)
• Weighted Regression Options (#143)
• GPE ScikitLearn fit! gives error (#149)
• plotting error (#150)
• Missing KernelData for ElasticCovStrat? (#155)
• Fix arbitrary hyperparameters (#156)
• Question: Are multiple inputs possible? (#161)
• Documenter compat issue (#162)
• Implementation of LOO Cross-Validation (#164)
• Posterior does not go through zero noise points (#168)

Merged pull requests:

• Added a unit test for elliptical slice sampling (#122) (@Red-Portal)
• Allow more recent version of StaticArrays and prepare for bugfix release (#124) (@andreasnoack)
• Integrate elliptical slice sampler (ESS) (#125) (@thomaspinder)
• CompatHelper: bump compat for "SpecialFunctions" to "0.9" (#127) (@github-actions[bot])
• CompatHelper: add new compat entry for "Documenter" at version "0.24" (#128) (@github-actions[bot])
• CompatHelper: bump compat for "Distributions" to "0.22" (#133) (@github-actions[bot])
• CompatHelper: bump compat for "SpecialFunctions" to "0.10" (#135) (@github-actions[bot])
• CompatHelper: bump compat for "Optim" to "0.20" (#136) (@github-actions[bot])
• fix tests that have been failing - WIP (#137) (@maximerischard)
• CompatHelper: bump compat for "RecipesBase" to "0.8" (#138) (@github-actions[bot])
• fixed bug in the elliptical slice implementation (#139) (@Red-Portal)
• Allow positional arguments to Optim.optimize (#142) (@baggepinnen)
• CompatHelper: bump compat for "Distributions" to "0.23" (#144) (@github-actions[bot])
• CompatHelper: bump compat for "RecipesBase" to "1.0" (#146) (@github-actions[bot])
• support for heteroscedastic noise in GPE (#148) (@maximerischard)
• CompatHelper: bump compat for "Optim" to "0.21" (#151) (@github-actions[bot])
• CompatHelper: bump compat for "Distances" to "0.9" (#152) (@github-actions[bot])
• CompatHelper: bump compat for "PDMats" to "0.10" (#153) (@github-actions[bot])
• Update ScikitLearn.jl (#154) (@OkonSamuel)
• CompatHelper: bump compat for "Zygote" to "0.5" (#157) (@github-actions[bot])
• CompatHelper: bump compat for "Optim" to "0.22" (#158) (@github-actions[bot])
• Fixed description for MeanPeriodic (#159) (@mgcth)
• CompatHelper: bump compat for "Documenter" to "0.25" (#160) (@github-actions[bot])
• Drop Documenter as a dependency (#163) (@Nosferican)
• CompatHelper: bump compat for "Optim" to "1.0" (#166) (@github-actions[bot])
• CompatHelper: bump compat for "Optim" to "1.1" (#167) (@github-actions[bot])
• Add BibTeX syntax color highlighting (#169) (@blegat)
• Typo fix: Sqaured -> Squared (#170) (@blegat)

v0.10.0

• Introduced sparse approximation methods
  • Subset of regressors, deterministic training conditional, fully independent training conditional and full-scale approximation are all available
• Extended functionality to include leave-one-out cross-validation
• Introduced functionality to enable variational inference in GPs with non-Gaussian data.
  • The approach used is a variant of that presented in Khan et. al.
  • Currently limited to only Poisson data. Additional likelihood functionality will be added soon.
• Deprecated GPMC in favor of GPA. This is to be inline with that fact that approximate inference in a GP is not limited to MCMC, but variational methods can now be used.

v0.9.0

Update NEWS.md

v0.8.0

See NEWS.md for details.

v0.7.0

See NEWS.md for summary of changes.